import math
import pandas as pd
from fastai.vision.all import *
from singleline_dataset.embeddings import *
from singleline_dataset.fileorg import *no env var SINGLELINE_DATA_HOME, defaulting to: Noneclustering
Fill in a module description here
EXTERNAL_PARENT_DIR = Path('/Users/al/Dropbox/2-Areas/2-Sketchbooks')
EXTERNAL_SKETCHBOOKS_DIR = EXTERNAL_PARENT_DIR / 'sketchbooks'
EXTERNAL_DATA_HOME = EXTERNAL_PARENT_DIR / 'singleline_data'EPOCH = SketchbookEpoch(epoch='20231214', data_home=EXTERNAL_DATA_HOME)
CURR_01_FLAT = EPOCH.dir_01_FLAT()
CURR_02_CATEGORIZED = EPOCH.dir_02_CATEGORIZED()
CURR_03_HAND = EPOCH.dir_03_HANDLABELED()no env var SINGLELINE_DATA_HOME, defaulting to: NoneEXTERNAL_SKETCHBOOKS_DIR.ls()[:2](#2) [Path('/Users/al/Dropbox/2-Areas/2-Sketchbooks/sketchbooks/sb67'),Path('/Users/al/Dropbox/2-Areas/2-Sketchbooks/sketchbooks/sb69')]# flatten_sketchbooks(EXTERNAL_SKETCHBOOKS_DIR, CURR_01_FLAT)CURR_01_FLATPath('/Users/al/Dropbox/2-Areas/2-Sketchbooks/singleline_data/raster/epoch-20231214/01_FLAT')separate DF’s for drawings, watercolors
drawings_df = cdf[cdf.handlabeled_metacluster_name == "0_drawings"]drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | indiv_fname | visual_rel_path | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| idx | |||||||||||||||||||||
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | sb77p043.jpg | 1_watercolors/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb77p043.jpg | 0_drawings/sb77p043.jpg | 0_drawings | 0 | False | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | sb67p021.jpg | 0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb67p021.jpg | 0_drawings/sb67p021.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | sb25p227.jpg | 0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb25p227.jpg | 0_drawings/sb25p227.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... |
| 6 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | sb48p057.jpg | 0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb48p057.jpg | 0_drawings/sb48p057.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... |
| 8 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | sb77p044.jpg | 0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb77p044.jpg | 0_drawings/sb77p044.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... |
5 rows × 21 columns
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | indiv_fname | visual_rel_path | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| idx | |||||||||||||||||||||
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | sb77p043.jpg | 1_watercolors/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb77p043.jpg | 0_drawings/sb77p043.jpg | 0_drawings | 0 | False | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | sb67p021.jpg | 0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb67p021.jpg | 0_drawings/sb67p021.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | sb25p227.jpg | 0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb25p227.jpg | 0_drawings/sb25p227.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... |
| 6 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | sb48p057.jpg | 0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb48p057.jpg | 0_drawings/sb48p057.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... |
| 8 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | sb77p044.jpg | 0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb77p044.jpg | 0_drawings/sb77p044.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... |
5 rows × 21 columns
# reindex so I have you
drawings_df.set_index(pd.Index(range(len(drawings_df))), inplace=True)drawings_df.label.hist()
drawings_df.cluster.hist()
watercolors_df = cdf[cdf.handlabeled_metacluster_name == "1_watercolors"]
watercolors_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | indiv_fname | visual_rel_path | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| idx | |||||||||||||||||||||
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb07p104.jpg | art/sb07p104.jpg | art | xtra | 4 | 0.177406,0.008799,0.008245,0.017177,0.788373 | -0.39105213,0.96309596,-0.73929733,0.4489988,0.8297614,-0.3937308,-0.7267234,-0.32461396,-0.7537334,-0.42715576,-0.62038505,-0.72562873,0.3037846,-0.6266761,-0.5213815,-0.31669965,-0.5467274,-0.6520286,-0.22544897,-0.6497359,-0.64108604,0.6374447,-0.5446661,-0.6779002,-0.14143991,-0.6690378,-0.44753593,0.66945183,-0.73986673,-0.45211998,-0.05996394,-0.21525535,0.82928157,-0.04412651,-0.600525,-0.04148197,-0.73825717,-0.57338566,-0.25227422,2.3566113,-0.61049384,1.0027854,-0.40134108,1.3733256,-0.75794554,-0.59257406,-0.4395525,-0.6001429,1.0358506,1.1911299,0.8162154,0.032713354,-0.6482407... | 5 | 238.33170 | 1 | ... | sb07p104.jpg | 1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb07p104.jpg | 1_watercolors/sb07p104.jpg | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb07p104.jpg | -0.6198182,0.8895926,-0.73929733,0.6418346,0.46164584,1.0085131,-0.7267234,-0.4896673,-0.7537334,-0.42715576,-0.62038505,-0.72562873,1.4399182,-0.6266761,0.474914,-0.57515705,-0.5467274,-0.6520286,1.2274435,-0.6497359,-0.27622125,-0.45870692,-0.5446661,-0.6779002,1.3343718,-0.6690378,-0.4518578,0.42277813,-0.73986673,-0.45211998,1.6291834,-0.67007643,-0.09817025,-0.00090405345,-0.600525,-0.6186755,-0.73933,-0.57338566,0.006386578,1.034553,-0.61049384,-0.58021957,-0.09483403,0.78779536,-0.73080033,0.10947651,0.3367887,-0.5382227,-0.26548737,0.5630663,2.15088,0.2266624,-0.6482407,0.049348265... |
| 5 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb31p017.jpg | art/sb31p017.jpg | art | xtra | 4 | 0.172119,0.008924,0.008226,0.016950,0.793781 | -0.05104202,-0.3935787,-0.73929733,0.86878717,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14799604,-0.62038505,-0.72562873,-0.2654881,-0.6266761,-0.68991846,-0.57515705,-0.41532665,-0.6520286,-0.6898426,-0.34099144,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.2674706,-0.53598946,-0.13334358,-0.5982413,-0.73986673,-0.45211998,-0.68509877,0.033780396,-0.52171326,0.054827005,-0.20721096,0.3301326,-0.73933,-0.57338566,0.34032136,-0.6179389,-0.20170414,0.16225064,-0.44437563,0.44581896,-0.069196045,0.20976627,-0.07846111,-0.6001429,0.38122594,-0.09738222,-0.59497476,-0.665527... | 11 | 191.66959 | 1 | ... | sb31p017.jpg | 1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb31p017.jpg | 1_watercolors/sb31p017.jpg | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb31p017.jpg | 1.491735,-0.5714155,-0.73929733,0.05867392,0.5154146,0.0936929,-0.7267234,-0.4896673,0.4246894,-0.42715576,-0.33477074,-0.72562873,1.2274611,-0.6266761,-0.68991846,-0.57515705,-0.5467274,1.1312869,-0.06054777,-0.028933167,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.46393585,0.4817328,0.22738433,-0.73986673,-0.45211998,0.0057709217,-0.67007643,-0.52171326,0.68283725,0.060967267,0.7253895,-0.73933,-0.57338566,-0.65925777,-0.6179389,-0.32987997,0.86336035,-0.6189488,1.3613329,1.2250904,-0.59257406,-0.5857397,-0.6001429,-0.5687719,-0.14814761,0.5325973,-0.66552746,-0.6482407,0.... |
| 21 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb05p079.jpg | art/sb05p079.jpg | art | xtra | 4 | 0.172158,0.008873,0.008536,0.017017,0.793416 | 1.8089037,-0.5714155,-0.73929733,-0.10773122,-0.22753507,0.34884918,0.124596715,-0.4896673,-0.27447522,-0.42715576,0.61659455,-0.03226477,-0.662768,-0.6266761,-0.68991846,-0.57515705,-0.39897114,-0.6520286,-0.6898426,-0.6497359,0.16184372,-0.55193967,-0.5446661,-0.6779002,0.04647112,-0.6690378,-0.4518578,0.9576235,0.29423904,-0.45211998,-0.68509877,0.09260088,-0.3948297,-0.4592305,-0.600525,-0.6186755,-0.64046603,-0.57338566,0.08902073,0.09568572,-0.61049384,0.7515181,-0.15892029,-0.4098586,1.8483992,-0.59257406,-0.6157326,-0.56642026,-0.651804,-0.44185895,-0.19970179,0.19587982,1.3608505,... | 11 | 140.32997 | 1 | ... | sb05p079.jpg | 1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb05p079.jpg | 1_watercolors/sb05p079.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb05p079.jpg | -0.6013173,-0.5714155,-0.7368871,-0.6553513,-0.6254566,-0.65341413,0.071327925,-0.4896673,-0.7537334,-0.29406685,1.2854458,-0.72562873,1.4279991,-0.2084853,0.4439276,-0.57515705,-0.5467274,-0.2426523,0.6567551,-0.6497359,0.284827,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.39760548,-0.5982413,-0.16403872,0.82081807,-0.68509877,-0.67007643,-0.16007048,-0.23898152,0.4340614,-0.6186755,-0.73933,-0.57338566,-0.65925777,-0.60700464,-0.22504222,1.2887287,0.17025876,-0.5827259,3.245521,-0.59257406,1.607188,-0.6001429,-0.1800242,-0.5315948,0.2059763,0.33526373,0.109642506,-0.0013180... |
| 24 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb26p082.jpg | art/sb26p082.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | 1.8392665,0.3796271,-0.73929733,0.82189727,1.6830277,0.9302285,-0.5532687,-0.4896673,-0.30764756,-0.42715576,-0.44450876,-0.72562873,-0.662768,0.54702616,-0.68991846,-0.57515705,-0.31611928,-0.59368926,1.2819774,-0.6497359,-0.64108604,-0.55193967,0.84984857,-0.6779002,-0.06536412,-0.6690378,0.34403068,-0.16627067,0.3990196,-0.45211998,0.65170527,-0.67007643,0.6049255,-0.4367706,-0.600525,-0.0046679974,-0.73933,-0.109048486,-0.65925777,0.65358853,-0.61049384,0.70374626,-0.6189488,-0.6682474,-0.0177418,-0.59257406,1.1068444,-0.29651338,-0.651804,-0.5315948,1.8200746,-0.66552746,0.11623299,-0... | 7 | 176.11029 | 1 | ... | sb26p082.jpg | 1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb26p082.jpg | 1_watercolors/sb26p082.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb26p082.jpg | 0.8175719,0.48385304,-0.73929733,1.0291914,1.4682066,1.225123,-0.43518582,-0.4896673,-0.7537334,-0.42715576,0.72162604,-0.72562873,-0.662768,-0.6266761,-0.55470747,-0.57515705,-0.5467274,-0.6520286,2.099903,-0.6497359,-0.64108604,-0.55193967,1.3572993,-0.6779002,1.3149762,0.8512635,0.81857395,-0.5982413,0.056738913,-0.45211998,2.0946088,-0.67007643,-0.52171326,-0.4592305,-0.600525,1.1100725,-0.73933,-0.57338566,-0.24140868,0.9004108,-0.61049384,0.4742828,-0.6189488,-0.6682474,0.335356,0.40166396,0.7790335,-0.6001429,-0.651804,-0.4778452,0.7360641,-0.66552746,-0.6482407,-0.13845,-0.7311095,... |
| 28 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb69p101-color.jpg | art/sb69p101-color.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | -0.23259398,-0.5714155,-0.73929733,-0.6553513,0.34417403,1.046661,-0.42288247,-0.4896673,0.39017296,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.93892807,-0.57515705,-0.5222921,-0.6520286,-0.6898426,-0.6497359,-0.6366938,-0.55193967,-0.5446661,-0.6779002,-0.46687827,-0.6690378,-0.4518578,-0.5982413,0.7252437,-0.45211998,1.8137504,-0.67007643,-0.52171326,-0.36457688,-0.600525,0.8178272,-0.73933,0.23574609,-0.65925777,-0.6179389,-0.61049384,-0.15158218,-0.41958112,-0.6682474,-0.75794554,0.4943592,0.8210626,-0.6001429,-0.651804,-0.080083996,0.48374963,-0.42743486,-0.6482407,-0.4... | 7 | 187.45346 | 1 | ... | sb69p101-color.jpg | 1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb69p101-color.jpg | 1_watercolors/sb69p101-color.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb69p101-color.jpg | 0.14908737,-0.5714155,-0.73929733,0.44245207,-0.6254566,1.4611686,-0.7267234,-0.4896673,0.87923527,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.4918049,-0.57515705,-0.5467274,-0.6520286,-0.6898426,-0.2860955,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.69009775,-0.6690378,-0.4518578,0.18292433,-0.004632473,-0.45211998,1.7543458,-0.67007643,-0.52171326,-0.41669255,-0.600525,1.4654884,0.5742926,-0.57338566,-0.65925777,-0.6179389,-0.52983093,0.26484245,-0.25630984,-0.6682474,0.19898295,0.95297796,0.38033813,-0.6001429,-0.651804,-0.5315948,1.3264011,0.31547034,-0.6482407,-0.47... |
5 rows × 21 columns
watercolors_df.set_index(pd.Index(range(len(watercolors_df))), inplace=True)
watercolors_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | indiv_fname | visual_rel_path | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb07p104.jpg | art/sb07p104.jpg | art | xtra | 4 | 0.177406,0.008799,0.008245,0.017177,0.788373 | -0.39105213,0.96309596,-0.73929733,0.4489988,0.8297614,-0.3937308,-0.7267234,-0.32461396,-0.7537334,-0.42715576,-0.62038505,-0.72562873,0.3037846,-0.6266761,-0.5213815,-0.31669965,-0.5467274,-0.6520286,-0.22544897,-0.6497359,-0.64108604,0.6374447,-0.5446661,-0.6779002,-0.14143991,-0.6690378,-0.44753593,0.66945183,-0.73986673,-0.45211998,-0.05996394,-0.21525535,0.82928157,-0.04412651,-0.600525,-0.04148197,-0.73825717,-0.57338566,-0.25227422,2.3566113,-0.61049384,1.0027854,-0.40134108,1.3733256,-0.75794554,-0.59257406,-0.4395525,-0.6001429,1.0358506,1.1911299,0.8162154,0.032713354,-0.6482407... | 5 | 238.33170 | 1 | ... | sb07p104.jpg | 1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb07p104.jpg | 1_watercolors/sb07p104.jpg | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb07p104.jpg | -0.6198182,0.8895926,-0.73929733,0.6418346,0.46164584,1.0085131,-0.7267234,-0.4896673,-0.7537334,-0.42715576,-0.62038505,-0.72562873,1.4399182,-0.6266761,0.474914,-0.57515705,-0.5467274,-0.6520286,1.2274435,-0.6497359,-0.27622125,-0.45870692,-0.5446661,-0.6779002,1.3343718,-0.6690378,-0.4518578,0.42277813,-0.73986673,-0.45211998,1.6291834,-0.67007643,-0.09817025,-0.00090405345,-0.600525,-0.6186755,-0.73933,-0.57338566,0.006386578,1.034553,-0.61049384,-0.58021957,-0.09483403,0.78779536,-0.73080033,0.10947651,0.3367887,-0.5382227,-0.26548737,0.5630663,2.15088,0.2266624,-0.6482407,0.049348265... |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb31p017.jpg | art/sb31p017.jpg | art | xtra | 4 | 0.172119,0.008924,0.008226,0.016950,0.793781 | -0.05104202,-0.3935787,-0.73929733,0.86878717,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14799604,-0.62038505,-0.72562873,-0.2654881,-0.6266761,-0.68991846,-0.57515705,-0.41532665,-0.6520286,-0.6898426,-0.34099144,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.2674706,-0.53598946,-0.13334358,-0.5982413,-0.73986673,-0.45211998,-0.68509877,0.033780396,-0.52171326,0.054827005,-0.20721096,0.3301326,-0.73933,-0.57338566,0.34032136,-0.6179389,-0.20170414,0.16225064,-0.44437563,0.44581896,-0.069196045,0.20976627,-0.07846111,-0.6001429,0.38122594,-0.09738222,-0.59497476,-0.665527... | 11 | 191.66959 | 1 | ... | sb31p017.jpg | 1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb31p017.jpg | 1_watercolors/sb31p017.jpg | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb31p017.jpg | 1.491735,-0.5714155,-0.73929733,0.05867392,0.5154146,0.0936929,-0.7267234,-0.4896673,0.4246894,-0.42715576,-0.33477074,-0.72562873,1.2274611,-0.6266761,-0.68991846,-0.57515705,-0.5467274,1.1312869,-0.06054777,-0.028933167,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.46393585,0.4817328,0.22738433,-0.73986673,-0.45211998,0.0057709217,-0.67007643,-0.52171326,0.68283725,0.060967267,0.7253895,-0.73933,-0.57338566,-0.65925777,-0.6179389,-0.32987997,0.86336035,-0.6189488,1.3613329,1.2250904,-0.59257406,-0.5857397,-0.6001429,-0.5687719,-0.14814761,0.5325973,-0.66552746,-0.6482407,0.... |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb05p079.jpg | art/sb05p079.jpg | art | xtra | 4 | 0.172158,0.008873,0.008536,0.017017,0.793416 | 1.8089037,-0.5714155,-0.73929733,-0.10773122,-0.22753507,0.34884918,0.124596715,-0.4896673,-0.27447522,-0.42715576,0.61659455,-0.03226477,-0.662768,-0.6266761,-0.68991846,-0.57515705,-0.39897114,-0.6520286,-0.6898426,-0.6497359,0.16184372,-0.55193967,-0.5446661,-0.6779002,0.04647112,-0.6690378,-0.4518578,0.9576235,0.29423904,-0.45211998,-0.68509877,0.09260088,-0.3948297,-0.4592305,-0.600525,-0.6186755,-0.64046603,-0.57338566,0.08902073,0.09568572,-0.61049384,0.7515181,-0.15892029,-0.4098586,1.8483992,-0.59257406,-0.6157326,-0.56642026,-0.651804,-0.44185895,-0.19970179,0.19587982,1.3608505,... | 11 | 140.32997 | 1 | ... | sb05p079.jpg | 1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb05p079.jpg | 1_watercolors/sb05p079.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb05p079.jpg | -0.6013173,-0.5714155,-0.7368871,-0.6553513,-0.6254566,-0.65341413,0.071327925,-0.4896673,-0.7537334,-0.29406685,1.2854458,-0.72562873,1.4279991,-0.2084853,0.4439276,-0.57515705,-0.5467274,-0.2426523,0.6567551,-0.6497359,0.284827,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.39760548,-0.5982413,-0.16403872,0.82081807,-0.68509877,-0.67007643,-0.16007048,-0.23898152,0.4340614,-0.6186755,-0.73933,-0.57338566,-0.65925777,-0.60700464,-0.22504222,1.2887287,0.17025876,-0.5827259,3.245521,-0.59257406,1.607188,-0.6001429,-0.1800242,-0.5315948,0.2059763,0.33526373,0.109642506,-0.0013180... |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb26p082.jpg | art/sb26p082.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | 1.8392665,0.3796271,-0.73929733,0.82189727,1.6830277,0.9302285,-0.5532687,-0.4896673,-0.30764756,-0.42715576,-0.44450876,-0.72562873,-0.662768,0.54702616,-0.68991846,-0.57515705,-0.31611928,-0.59368926,1.2819774,-0.6497359,-0.64108604,-0.55193967,0.84984857,-0.6779002,-0.06536412,-0.6690378,0.34403068,-0.16627067,0.3990196,-0.45211998,0.65170527,-0.67007643,0.6049255,-0.4367706,-0.600525,-0.0046679974,-0.73933,-0.109048486,-0.65925777,0.65358853,-0.61049384,0.70374626,-0.6189488,-0.6682474,-0.0177418,-0.59257406,1.1068444,-0.29651338,-0.651804,-0.5315948,1.8200746,-0.66552746,0.11623299,-0... | 7 | 176.11029 | 1 | ... | sb26p082.jpg | 1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb26p082.jpg | 1_watercolors/sb26p082.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb26p082.jpg | 0.8175719,0.48385304,-0.73929733,1.0291914,1.4682066,1.225123,-0.43518582,-0.4896673,-0.7537334,-0.42715576,0.72162604,-0.72562873,-0.662768,-0.6266761,-0.55470747,-0.57515705,-0.5467274,-0.6520286,2.099903,-0.6497359,-0.64108604,-0.55193967,1.3572993,-0.6779002,1.3149762,0.8512635,0.81857395,-0.5982413,0.056738913,-0.45211998,2.0946088,-0.67007643,-0.52171326,-0.4592305,-0.600525,1.1100725,-0.73933,-0.57338566,-0.24140868,0.9004108,-0.61049384,0.4742828,-0.6189488,-0.6682474,0.335356,0.40166396,0.7790335,-0.6001429,-0.651804,-0.4778452,0.7360641,-0.66552746,-0.6482407,-0.13845,-0.7311095,... |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb69p101-color.jpg | art/sb69p101-color.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | -0.23259398,-0.5714155,-0.73929733,-0.6553513,0.34417403,1.046661,-0.42288247,-0.4896673,0.39017296,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.93892807,-0.57515705,-0.5222921,-0.6520286,-0.6898426,-0.6497359,-0.6366938,-0.55193967,-0.5446661,-0.6779002,-0.46687827,-0.6690378,-0.4518578,-0.5982413,0.7252437,-0.45211998,1.8137504,-0.67007643,-0.52171326,-0.36457688,-0.600525,0.8178272,-0.73933,0.23574609,-0.65925777,-0.6179389,-0.61049384,-0.15158218,-0.41958112,-0.6682474,-0.75794554,0.4943592,0.8210626,-0.6001429,-0.651804,-0.080083996,0.48374963,-0.42743486,-0.6482407,-0.4... | 7 | 187.45346 | 1 | ... | sb69p101-color.jpg | 1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb69p101-color.jpg | 1_watercolors/sb69p101-color.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb69p101-color.jpg | 0.14908737,-0.5714155,-0.73929733,0.44245207,-0.6254566,1.4611686,-0.7267234,-0.4896673,0.87923527,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.4918049,-0.57515705,-0.5467274,-0.6520286,-0.6898426,-0.2860955,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.69009775,-0.6690378,-0.4518578,0.18292433,-0.004632473,-0.45211998,1.7543458,-0.67007643,-0.52171326,-0.41669255,-0.600525,1.4654884,0.5742926,-0.57338566,-0.65925777,-0.6179389,-0.52983093,0.26484245,-0.25630984,-0.6682474,0.19898295,0.95297796,0.38033813,-0.6001429,-0.651804,-0.5315948,1.3264011,0.31547034,-0.6482407,-0.47... |
5 rows × 21 columns
watercolors_df.label.hist()
watercolors_df.cluster.hist()
drawings_tsv_fname = f"{preprocessing_dir}/drawings.tsv"
drawings_df.to_csv(
drawings_tsv_fname, index=True, index_label="idx", sep="\t", header=True
)
drawing_df = pd.read_csv(drawings_tsv_fname, delimiter="\t", index_col="idx")
drawing_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | indiv_fname | visual_rel_path | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| idx | |||||||||||||||||||||
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | sb77p043.jpg | 1_watercolors/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb77p043.jpg | 0_drawings/sb77p043.jpg | 0_drawings | 0 | False | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | sb67p021.jpg | 0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb67p021.jpg | 0_drawings/sb67p021.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | sb25p227.jpg | 0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb25p227.jpg | 0_drawings/sb25p227.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | sb48p057.jpg | 0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb48p057.jpg | 0_drawings/sb48p057.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | sb77p044.jpg | 0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb77p044.jpg | 0_drawings/sb77p044.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... |
5 rows × 21 columns
watercolors_tsv_fname = f"{preprocessing_dir}/watercolors.tsv"
watercolors_df.to_csv(
watercolors_tsv_fname, index=True, index_label="idx", sep="\t", header=True
)
watercolor_df = pd.read_csv(watercolors_tsv_fname, delimiter="\t", index_col="idx")
watercolor_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | indiv_fname | visual_rel_path | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| idx | |||||||||||||||||||||
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb07p104.jpg | art/sb07p104.jpg | art | xtra | 4 | 0.177406,0.008799,0.008245,0.017177,0.788373 | -0.39105213,0.96309596,-0.73929733,0.4489988,0.8297614,-0.3937308,-0.7267234,-0.32461396,-0.7537334,-0.42715576,-0.62038505,-0.72562873,0.3037846,-0.6266761,-0.5213815,-0.31669965,-0.5467274,-0.6520286,-0.22544897,-0.6497359,-0.64108604,0.6374447,-0.5446661,-0.6779002,-0.14143991,-0.6690378,-0.44753593,0.66945183,-0.73986673,-0.45211998,-0.05996394,-0.21525535,0.82928157,-0.04412651,-0.600525,-0.04148197,-0.73825717,-0.57338566,-0.25227422,2.3566113,-0.61049384,1.0027854,-0.40134108,1.3733256,-0.75794554,-0.59257406,-0.4395525,-0.6001429,1.0358506,1.1911299,0.8162154,0.032713354,-0.6482407... | 5 | 238.33170 | 1 | ... | sb07p104.jpg | 1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb07p104.jpg | 1_watercolors/sb07p104.jpg | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb07p104.jpg | -0.6198182,0.8895926,-0.73929733,0.6418346,0.46164584,1.0085131,-0.7267234,-0.4896673,-0.7537334,-0.42715576,-0.62038505,-0.72562873,1.4399182,-0.6266761,0.474914,-0.57515705,-0.5467274,-0.6520286,1.2274435,-0.6497359,-0.27622125,-0.45870692,-0.5446661,-0.6779002,1.3343718,-0.6690378,-0.4518578,0.42277813,-0.73986673,-0.45211998,1.6291834,-0.67007643,-0.09817025,-0.00090405345,-0.600525,-0.6186755,-0.73933,-0.57338566,0.006386578,1.034553,-0.61049384,-0.58021957,-0.09483403,0.78779536,-0.73080033,0.10947651,0.3367887,-0.5382227,-0.26548737,0.5630663,2.15088,0.2266624,-0.6482407,0.049348265... |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb31p017.jpg | art/sb31p017.jpg | art | xtra | 4 | 0.172119,0.008924,0.008226,0.016950,0.793781 | -0.05104202,-0.3935787,-0.73929733,0.86878717,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14799604,-0.62038505,-0.72562873,-0.2654881,-0.6266761,-0.68991846,-0.57515705,-0.41532665,-0.6520286,-0.6898426,-0.34099144,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.2674706,-0.53598946,-0.13334358,-0.5982413,-0.73986673,-0.45211998,-0.68509877,0.033780396,-0.52171326,0.054827005,-0.20721096,0.3301326,-0.73933,-0.57338566,0.34032136,-0.6179389,-0.20170414,0.16225064,-0.44437563,0.44581896,-0.069196045,0.20976627,-0.07846111,-0.6001429,0.38122594,-0.09738222,-0.59497476,-0.665527... | 11 | 191.66959 | 1 | ... | sb31p017.jpg | 1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb31p017.jpg | 1_watercolors/sb31p017.jpg | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb31p017.jpg | 1.491735,-0.5714155,-0.73929733,0.05867392,0.5154146,0.0936929,-0.7267234,-0.4896673,0.4246894,-0.42715576,-0.33477074,-0.72562873,1.2274611,-0.6266761,-0.68991846,-0.57515705,-0.5467274,1.1312869,-0.06054777,-0.028933167,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.46393585,0.4817328,0.22738433,-0.73986673,-0.45211998,0.0057709217,-0.67007643,-0.52171326,0.68283725,0.060967267,0.7253895,-0.73933,-0.57338566,-0.65925777,-0.6179389,-0.32987997,0.86336035,-0.6189488,1.3613329,1.2250904,-0.59257406,-0.5857397,-0.6001429,-0.5687719,-0.14814761,0.5325973,-0.66552746,-0.6482407,0.... |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb05p079.jpg | art/sb05p079.jpg | art | xtra | 4 | 0.172158,0.008873,0.008536,0.017017,0.793416 | 1.8089037,-0.5714155,-0.73929733,-0.10773122,-0.22753507,0.34884918,0.124596715,-0.4896673,-0.27447522,-0.42715576,0.61659455,-0.03226477,-0.662768,-0.6266761,-0.68991846,-0.57515705,-0.39897114,-0.6520286,-0.6898426,-0.6497359,0.16184372,-0.55193967,-0.5446661,-0.6779002,0.04647112,-0.6690378,-0.4518578,0.9576235,0.29423904,-0.45211998,-0.68509877,0.09260088,-0.3948297,-0.4592305,-0.600525,-0.6186755,-0.64046603,-0.57338566,0.08902073,0.09568572,-0.61049384,0.7515181,-0.15892029,-0.4098586,1.8483992,-0.59257406,-0.6157326,-0.56642026,-0.651804,-0.44185895,-0.19970179,0.19587982,1.3608505,... | 11 | 140.32997 | 1 | ... | sb05p079.jpg | 1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb05p079.jpg | 1_watercolors/sb05p079.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb05p079.jpg | -0.6013173,-0.5714155,-0.7368871,-0.6553513,-0.6254566,-0.65341413,0.071327925,-0.4896673,-0.7537334,-0.29406685,1.2854458,-0.72562873,1.4279991,-0.2084853,0.4439276,-0.57515705,-0.5467274,-0.2426523,0.6567551,-0.6497359,0.284827,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.39760548,-0.5982413,-0.16403872,0.82081807,-0.68509877,-0.67007643,-0.16007048,-0.23898152,0.4340614,-0.6186755,-0.73933,-0.57338566,-0.65925777,-0.60700464,-0.22504222,1.2887287,0.17025876,-0.5827259,3.245521,-0.59257406,1.607188,-0.6001429,-0.1800242,-0.5315948,0.2059763,0.33526373,0.109642506,-0.0013180... |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb26p082.jpg | art/sb26p082.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | 1.8392665,0.3796271,-0.73929733,0.82189727,1.6830277,0.9302285,-0.5532687,-0.4896673,-0.30764756,-0.42715576,-0.44450876,-0.72562873,-0.662768,0.54702616,-0.68991846,-0.57515705,-0.31611928,-0.59368926,1.2819774,-0.6497359,-0.64108604,-0.55193967,0.84984857,-0.6779002,-0.06536412,-0.6690378,0.34403068,-0.16627067,0.3990196,-0.45211998,0.65170527,-0.67007643,0.6049255,-0.4367706,-0.600525,-0.0046679974,-0.73933,-0.109048486,-0.65925777,0.65358853,-0.61049384,0.70374626,-0.6189488,-0.6682474,-0.0177418,-0.59257406,1.1068444,-0.29651338,-0.651804,-0.5315948,1.8200746,-0.66552746,0.11623299,-0... | 7 | 176.11029 | 1 | ... | sb26p082.jpg | 1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb26p082.jpg | 1_watercolors/sb26p082.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb26p082.jpg | 0.8175719,0.48385304,-0.73929733,1.0291914,1.4682066,1.225123,-0.43518582,-0.4896673,-0.7537334,-0.42715576,0.72162604,-0.72562873,-0.662768,-0.6266761,-0.55470747,-0.57515705,-0.5467274,-0.6520286,2.099903,-0.6497359,-0.64108604,-0.55193967,1.3572993,-0.6779002,1.3149762,0.8512635,0.81857395,-0.5982413,0.056738913,-0.45211998,2.0946088,-0.67007643,-0.52171326,-0.4592305,-0.600525,1.1100725,-0.73933,-0.57338566,-0.24140868,0.9004108,-0.61049384,0.4742828,-0.6189488,-0.6682474,0.335356,0.40166396,0.7790335,-0.6001429,-0.651804,-0.4778452,0.7360641,-0.66552746,-0.6482407,-0.13845,-0.7311095,... |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb69p101-color.jpg | art/sb69p101-color.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | -0.23259398,-0.5714155,-0.73929733,-0.6553513,0.34417403,1.046661,-0.42288247,-0.4896673,0.39017296,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.93892807,-0.57515705,-0.5222921,-0.6520286,-0.6898426,-0.6497359,-0.6366938,-0.55193967,-0.5446661,-0.6779002,-0.46687827,-0.6690378,-0.4518578,-0.5982413,0.7252437,-0.45211998,1.8137504,-0.67007643,-0.52171326,-0.36457688,-0.600525,0.8178272,-0.73933,0.23574609,-0.65925777,-0.6179389,-0.61049384,-0.15158218,-0.41958112,-0.6682474,-0.75794554,0.4943592,0.8210626,-0.6001429,-0.651804,-0.080083996,0.48374963,-0.42743486,-0.6482407,-0.4... | 7 | 187.45346 | 1 | ... | sb69p101-color.jpg | 1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb69p101-color.jpg | 1_watercolors/sb69p101-color.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb69p101-color.jpg | 0.14908737,-0.5714155,-0.73929733,0.44245207,-0.6254566,1.4611686,-0.7267234,-0.4896673,0.87923527,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.4918049,-0.57515705,-0.5467274,-0.6520286,-0.6898426,-0.2860955,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.69009775,-0.6690378,-0.4518578,0.18292433,-0.004632473,-0.45211998,1.7543458,-0.67007643,-0.52171326,-0.41669255,-0.600525,1.4654884,0.5742926,-0.57338566,-0.65925777,-0.6179389,-0.52983093,0.26484245,-0.25630984,-0.6682474,0.19898295,0.95297796,0.38033813,-0.6001429,-0.651804,-0.5315948,1.3264011,0.31547034,-0.6482407,-0.47... |
5 rows × 21 columns
drawings_emb = np.stack(
[np.array([float(f) for f in s.split(",")]) for s in list(drawings_df.emb_csv)]
).astype(np.float32)
drawings_cropped_emb = np.stack(
[
np.array([float(f) for f in s.split(",")])
for s in list(drawings_df.emb_csv_cropped)
]
).astype(np.float32)
drawings_emb.shape, drawings_cropped_emb.shape((1923, 512), (1923, 512))watercolors_emb = np.stack(
[np.array([float(f) for f in s.split(",")]) for s in list(watercolors_df.emb_csv)]
).astype(np.float32)
watercolors_cropped_emb = np.stack(
[
np.array([float(f) for f in s.split(",")])
for s in list(watercolors_df.emb_csv_cropped)
]
).astype(np.float32)
watercolors_emb.shape, watercolors_cropped_emb.shape((404, 512), (404, 512))drawings_X_tsne = TSNE(n_components=2, random_state=42).fit_transform(drawings_emb)
fig = plt.scatter(
x=drawings_X_tsne[:, 0], y=drawings_X_tsne[:, 1], c=drawings_df.cluster, s=2.0
)
drawings_cropped_X_tsne = TSNE(n_components=2, random_state=42).fit_transform(
drawings_cropped_emb
)
fig = plt.scatter(
x=drawings_cropped_X_tsne[:, 0],
y=drawings_cropped_X_tsne[:, 1],
c=drawings_df.cluster,
s=2.0,
)
drawings_index = faiss.IndexFlatL2(emb_dim)
drawings_index.add(drawings_emb)
drawings_cropped_index = faiss.IndexFlatL2(emb_dim)
drawings_cropped_index.add(drawings_cropped_emb)watercolors_X_tsne = TSNE(n_components=2, random_state=42).fit_transform(
watercolors_emb
)
fig = plt.scatter(
x=watercolors_X_tsne[:, 0],
y=watercolors_X_tsne[:, 1],
c=watercolors_df.cluster,
s=5.0,
)
watercolors_cropped_X_tsne = TSNE(n_components=2, random_state=42).fit_transform(
watercolors_cropped_emb
)
fig = plt.scatter(
x=watercolors_cropped_X_tsne[:, 0],
y=watercolors_cropped_X_tsne[:, 1],
c=watercolors_df.cluster,
s=5.0,
)
watercolors_index = faiss.IndexFlatL2(emb_dim)
watercolors_index.add(watercolors_emb)
watercolors_cropped_index = faiss.IndexFlatL2(emb_dim)
watercolors_cropped_index.add(watercolors_cropped_emb)drawings_emb
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | visual_rel_path | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | drawings_cluster_id | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 1_watercolors/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb77p043.jpg | 0_drawings/sb77p043.jpg | 0_drawings | 0 | False | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... | 3 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb67p021.jpg | 0_drawings/sb67p021.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... | 2 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb25p227.jpg | 0_drawings/sb25p227.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... | 3 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb48p057.jpg | 0_drawings/sb48p057.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... | 4 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb77p044.jpg | 0_drawings/sb77p044.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... | 0 |
5 rows × 22 columns
drawings_df = drawings_df.reindex()drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | visual_rel_path | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | drawings_cluster_id | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 1_watercolors/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb77p043.jpg | 0_drawings/sb77p043.jpg | 0_drawings | 0 | False | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... | 3 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb67p021.jpg | 0_drawings/sb67p021.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... | 2 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb25p227.jpg | 0_drawings/sb25p227.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... | 3 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb48p057.jpg | 0_drawings/sb48p057.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... | 4 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb77p044.jpg | 0_drawings/sb77p044.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... | 0 |
5 rows × 22 columns
ncentroids = 8
niter = 20
verbose = False
drawings_kmeans = faiss.Kmeans(
drawings_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_kmeans.train(drawings_emb)
drawings_D, drawings_I = drawings_index.search(drawings_kmeans.centroids, 16)
drawings_kD, drawings_kI = drawings_kmeans.index.search(drawings_emb, 1)
drawings_df.loc[:, "drawings_cluster_id"] = drawings_kI
drawings_df.loc[:, "drawings_cluster_dist"] = drawings_kD
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | drawings_cluster_id | drawings_cluster_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb77p043.jpg | 0_drawings/sb77p043.jpg | 0_drawings | 0 | False | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... | 3 | 217.249588 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb67p021.jpg | 0_drawings/sb67p021.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... | 2 | 169.947693 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb25p227.jpg | 0_drawings/sb25p227.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... | 3 | 162.882431 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb48p057.jpg | 0_drawings/sb48p057.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... | 4 | 94.257965 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/0_drawings/sb77p044.jpg | 0_drawings/sb77p044.jpg | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... | 0 | 173.559937 |
5 rows × 23 columns
show_all_clusters(clusters_df=drawings_df, clusters=drawings_I, title="drawings_emb")
fig = plt.scatter(
x=drawings_X_tsne[:, 0],
y=drawings_X_tsne[:, 1],
c=drawings_df.drawings_cluster_id,
s=2.0,
)
drawings_emb_cropped
ncentroids = 8
niter = 20
verbose = False
drawings_cropped_kmeans = faiss.Kmeans(
drawings_cropped_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cropped_kmeans.train(drawings_cropped_emb)
drawings_cropped_D, drawings_cropped_I = drawings_cropped_index.search(
drawings_cropped_kmeans.centroids, 16
)
drawings_cropped_kD, drawings_cropped_kI = drawings_cropped_kmeans.index.search(
drawings_cropped_emb, 1
)
drawings_df.loc[:, "drawings_cropped_cluster_id"] = drawings_cropped_kI
drawings_df.loc[:, "drawings_cropped_cluster_dist"] = drawings_cropped_kD
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | drawings_cluster_id | drawings_cluster_dist | drawings_cropped_cluster_id | drawings_cropped_cluster_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 0_drawings | 0 | False | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... | 3 | 217.249588 | 5 | 332.983734 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... | 2 | 169.947693 | 5 | 126.699768 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... | 3 | 162.882431 | 5 | 257.762512 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... | 4 | 94.257965 | 2 | 145.937225 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 0_drawings | 0 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... | 0 | 173.559937 | 5 | 197.142792 |
5 rows × 25 columns
show_all_clusters(
clusters_df=drawings_df, clusters=drawings_cropped_I, title="drawings_cropped_emb"
)
fig = plt.scatter(
x=drawings_cropped_X_tsne[:, 0],
y=drawings_cropped_X_tsne[:, 1],
c=drawings_df.drawings_cropped_cluster_id,
s=2.0,
)
watercolors_emb
watercolors_df = watercolors_df.reindex()ncentroids = 8
niter = 20
verbose = False
watercolors_kmeans = faiss.Kmeans(
watercolors_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
watercolors_kmeans.train(watercolors_emb)
watercolors_D, watercolors_I = watercolors_index.search(
watercolors_kmeans.centroids, 16
)
watercolors_kD, watercolors_kI = watercolors_kmeans.index.search(watercolors_emb, 1)
watercolors_df.loc[:, "watercolors_cluster_id"] = watercolors_kI
watercolors_df.loc[:, "watercolors_cluster_dist"] = watercolors_kD
watercolors_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | visual_abs_path | handlabeled_rel_path | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | watercolors_cluster_id | watercolors_cluster_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb07p104.jpg | art/sb07p104.jpg | art | xtra | 4 | 0.177406,0.008799,0.008245,0.017177,0.788373 | -0.39105213,0.96309596,-0.73929733,0.4489988,0.8297614,-0.3937308,-0.7267234,-0.32461396,-0.7537334,-0.42715576,-0.62038505,-0.72562873,0.3037846,-0.6266761,-0.5213815,-0.31669965,-0.5467274,-0.6520286,-0.22544897,-0.6497359,-0.64108604,0.6374447,-0.5446661,-0.6779002,-0.14143991,-0.6690378,-0.44753593,0.66945183,-0.73986673,-0.45211998,-0.05996394,-0.21525535,0.82928157,-0.04412651,-0.600525,-0.04148197,-0.73825717,-0.57338566,-0.25227422,2.3566113,-0.61049384,1.0027854,-0.40134108,1.3733256,-0.75794554,-0.59257406,-0.4395525,-0.6001429,1.0358506,1.1911299,0.8162154,0.032713354,-0.6482407... | 5 | 238.33170 | 1 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb07p104.jpg | 1_watercolors/sb07p104.jpg | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb07p104.jpg | -0.6198182,0.8895926,-0.73929733,0.6418346,0.46164584,1.0085131,-0.7267234,-0.4896673,-0.7537334,-0.42715576,-0.62038505,-0.72562873,1.4399182,-0.6266761,0.474914,-0.57515705,-0.5467274,-0.6520286,1.2274435,-0.6497359,-0.27622125,-0.45870692,-0.5446661,-0.6779002,1.3343718,-0.6690378,-0.4518578,0.42277813,-0.73986673,-0.45211998,1.6291834,-0.67007643,-0.09817025,-0.00090405345,-0.600525,-0.6186755,-0.73933,-0.57338566,0.006386578,1.034553,-0.61049384,-0.58021957,-0.09483403,0.78779536,-0.73080033,0.10947651,0.3367887,-0.5382227,-0.26548737,0.5630663,2.15088,0.2266624,-0.6482407,0.049348265... | 7 | 215.029266 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb31p017.jpg | art/sb31p017.jpg | art | xtra | 4 | 0.172119,0.008924,0.008226,0.016950,0.793781 | -0.05104202,-0.3935787,-0.73929733,0.86878717,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14799604,-0.62038505,-0.72562873,-0.2654881,-0.6266761,-0.68991846,-0.57515705,-0.41532665,-0.6520286,-0.6898426,-0.34099144,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.2674706,-0.53598946,-0.13334358,-0.5982413,-0.73986673,-0.45211998,-0.68509877,0.033780396,-0.52171326,0.054827005,-0.20721096,0.3301326,-0.73933,-0.57338566,0.34032136,-0.6179389,-0.20170414,0.16225064,-0.44437563,0.44581896,-0.069196045,0.20976627,-0.07846111,-0.6001429,0.38122594,-0.09738222,-0.59497476,-0.665527... | 11 | 191.66959 | 1 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb31p017.jpg | 1_watercolors/sb31p017.jpg | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb31p017.jpg | 1.491735,-0.5714155,-0.73929733,0.05867392,0.5154146,0.0936929,-0.7267234,-0.4896673,0.4246894,-0.42715576,-0.33477074,-0.72562873,1.2274611,-0.6266761,-0.68991846,-0.57515705,-0.5467274,1.1312869,-0.06054777,-0.028933167,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.46393585,0.4817328,0.22738433,-0.73986673,-0.45211998,0.0057709217,-0.67007643,-0.52171326,0.68283725,0.060967267,0.7253895,-0.73933,-0.57338566,-0.65925777,-0.6179389,-0.32987997,0.86336035,-0.6189488,1.3613329,1.2250904,-0.59257406,-0.5857397,-0.6001429,-0.5687719,-0.14814761,0.5325973,-0.66552746,-0.6482407,0.... | 4 | 199.030975 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb05p079.jpg | art/sb05p079.jpg | art | xtra | 4 | 0.172158,0.008873,0.008536,0.017017,0.793416 | 1.8089037,-0.5714155,-0.73929733,-0.10773122,-0.22753507,0.34884918,0.124596715,-0.4896673,-0.27447522,-0.42715576,0.61659455,-0.03226477,-0.662768,-0.6266761,-0.68991846,-0.57515705,-0.39897114,-0.6520286,-0.6898426,-0.6497359,0.16184372,-0.55193967,-0.5446661,-0.6779002,0.04647112,-0.6690378,-0.4518578,0.9576235,0.29423904,-0.45211998,-0.68509877,0.09260088,-0.3948297,-0.4592305,-0.600525,-0.6186755,-0.64046603,-0.57338566,0.08902073,0.09568572,-0.61049384,0.7515181,-0.15892029,-0.4098586,1.8483992,-0.59257406,-0.6157326,-0.56642026,-0.651804,-0.44185895,-0.19970179,0.19587982,1.3608505,... | 11 | 140.32997 | 1 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb05p079.jpg | 1_watercolors/sb05p079.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb05p079.jpg | -0.6013173,-0.5714155,-0.7368871,-0.6553513,-0.6254566,-0.65341413,0.071327925,-0.4896673,-0.7537334,-0.29406685,1.2854458,-0.72562873,1.4279991,-0.2084853,0.4439276,-0.57515705,-0.5467274,-0.2426523,0.6567551,-0.6497359,0.284827,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.39760548,-0.5982413,-0.16403872,0.82081807,-0.68509877,-0.67007643,-0.16007048,-0.23898152,0.4340614,-0.6186755,-0.73933,-0.57338566,-0.65925777,-0.60700464,-0.22504222,1.2887287,0.17025876,-0.5827259,3.245521,-0.59257406,1.607188,-0.6001429,-0.1800242,-0.5315948,0.2059763,0.33526373,0.109642506,-0.0013180... | 4 | 142.766022 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb26p082.jpg | art/sb26p082.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | 1.8392665,0.3796271,-0.73929733,0.82189727,1.6830277,0.9302285,-0.5532687,-0.4896673,-0.30764756,-0.42715576,-0.44450876,-0.72562873,-0.662768,0.54702616,-0.68991846,-0.57515705,-0.31611928,-0.59368926,1.2819774,-0.6497359,-0.64108604,-0.55193967,0.84984857,-0.6779002,-0.06536412,-0.6690378,0.34403068,-0.16627067,0.3990196,-0.45211998,0.65170527,-0.67007643,0.6049255,-0.4367706,-0.600525,-0.0046679974,-0.73933,-0.109048486,-0.65925777,0.65358853,-0.61049384,0.70374626,-0.6189488,-0.6682474,-0.0177418,-0.59257406,1.1068444,-0.29651338,-0.651804,-0.5315948,1.8200746,-0.66552746,0.11623299,-0... | 7 | 176.11029 | 1 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb26p082.jpg | 1_watercolors/sb26p082.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb26p082.jpg | 0.8175719,0.48385304,-0.73929733,1.0291914,1.4682066,1.225123,-0.43518582,-0.4896673,-0.7537334,-0.42715576,0.72162604,-0.72562873,-0.662768,-0.6266761,-0.55470747,-0.57515705,-0.5467274,-0.6520286,2.099903,-0.6497359,-0.64108604,-0.55193967,1.3572993,-0.6779002,1.3149762,0.8512635,0.81857395,-0.5982413,0.056738913,-0.45211998,2.0946088,-0.67007643,-0.52171326,-0.4592305,-0.600525,1.1100725,-0.73933,-0.57338566,-0.24140868,0.9004108,-0.61049384,0.4742828,-0.6189488,-0.6682474,0.335356,0.40166396,0.7790335,-0.6001429,-0.651804,-0.4778452,0.7360641,-0.66552746,-0.6482407,-0.13845,-0.7311095,... | 1 | 157.938324 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb69p101-color.jpg | art/sb69p101-color.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | -0.23259398,-0.5714155,-0.73929733,-0.6553513,0.34417403,1.046661,-0.42288247,-0.4896673,0.39017296,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.93892807,-0.57515705,-0.5222921,-0.6520286,-0.6898426,-0.6497359,-0.6366938,-0.55193967,-0.5446661,-0.6779002,-0.46687827,-0.6690378,-0.4518578,-0.5982413,0.7252437,-0.45211998,1.8137504,-0.67007643,-0.52171326,-0.36457688,-0.600525,0.8178272,-0.73933,0.23574609,-0.65925777,-0.6179389,-0.61049384,-0.15158218,-0.41958112,-0.6682474,-0.75794554,0.4943592,0.8210626,-0.6001429,-0.651804,-0.080083996,0.48374963,-0.42743486,-0.6482407,-0.4... | 7 | 187.45346 | 1 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/02_VISUAL_CATEGORIES/1_watercolors/sb69p101-color.jpg | 1_watercolors/sb69p101-color.jpg | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb69p101-color.jpg | 0.14908737,-0.5714155,-0.73929733,0.44245207,-0.6254566,1.4611686,-0.7267234,-0.4896673,0.87923527,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.4918049,-0.57515705,-0.5467274,-0.6520286,-0.6898426,-0.2860955,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.69009775,-0.6690378,-0.4518578,0.18292433,-0.004632473,-0.45211998,1.7543458,-0.67007643,-0.52171326,-0.41669255,-0.600525,1.4654884,0.5742926,-0.57338566,-0.65925777,-0.6179389,-0.52983093,0.26484245,-0.25630984,-0.6682474,0.19898295,0.95297796,0.38033813,-0.6001429,-0.651804,-0.5315948,1.3264011,0.31547034,-0.6482407,-0.47... | 1 | 167.278229 |
5 rows × 23 columns
show_all_clusters(
clusters_df=watercolors_df, clusters=watercolors_I, title="watercolors_emb"
)
fig = plt.scatter(
x=watercolors_X_tsne[:, 0],
y=watercolors_X_tsne[:, 1],
c=watercolors_df.watercolors_cluster_id,
s=8.0,
)
watercolors_emb_cropped
ncentroids = 8
niter = 20
verbose = False
watercolors_cropped_kmeans = faiss.Kmeans(
watercolors_cropped_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
watercolors_cropped_kmeans.train(watercolors_cropped_emb)
watercolors_cropped_D, watercolors_cropped_I = watercolors_cropped_index.search(
watercolors_cropped_kmeans.centroids, 16
)
(
watercolors_cropped_kD,
watercolors_cropped_kI,
) = watercolors_cropped_kmeans.index.search(watercolors_cropped_emb, 1)
watercolors_df.loc[:, "watercolors_cropped_cluster_id"] = watercolors_cropped_kI
watercolors_df.loc[:, "watercolors_cropped_cluster_dist"] = watercolors_cropped_kD
watercolors_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | watercolors_cluster_id | watercolors_cluster_dist | watercolors_cropped_cluster_id | watercolors_cropped_cluster_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb07p104.jpg | art/sb07p104.jpg | art | xtra | 4 | 0.177406,0.008799,0.008245,0.017177,0.788373 | -0.39105213,0.96309596,-0.73929733,0.4489988,0.8297614,-0.3937308,-0.7267234,-0.32461396,-0.7537334,-0.42715576,-0.62038505,-0.72562873,0.3037846,-0.6266761,-0.5213815,-0.31669965,-0.5467274,-0.6520286,-0.22544897,-0.6497359,-0.64108604,0.6374447,-0.5446661,-0.6779002,-0.14143991,-0.6690378,-0.44753593,0.66945183,-0.73986673,-0.45211998,-0.05996394,-0.21525535,0.82928157,-0.04412651,-0.600525,-0.04148197,-0.73825717,-0.57338566,-0.25227422,2.3566113,-0.61049384,1.0027854,-0.40134108,1.3733256,-0.75794554,-0.59257406,-0.4395525,-0.6001429,1.0358506,1.1911299,0.8162154,0.032713354,-0.6482407... | 5 | 238.33170 | 1 | ... | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb07p104.jpg | -0.6198182,0.8895926,-0.73929733,0.6418346,0.46164584,1.0085131,-0.7267234,-0.4896673,-0.7537334,-0.42715576,-0.62038505,-0.72562873,1.4399182,-0.6266761,0.474914,-0.57515705,-0.5467274,-0.6520286,1.2274435,-0.6497359,-0.27622125,-0.45870692,-0.5446661,-0.6779002,1.3343718,-0.6690378,-0.4518578,0.42277813,-0.73986673,-0.45211998,1.6291834,-0.67007643,-0.09817025,-0.00090405345,-0.600525,-0.6186755,-0.73933,-0.57338566,0.006386578,1.034553,-0.61049384,-0.58021957,-0.09483403,0.78779536,-0.73080033,0.10947651,0.3367887,-0.5382227,-0.26548737,0.5630663,2.15088,0.2266624,-0.6482407,0.049348265... | 7 | 215.029266 | 7 | 206.641174 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb31p017.jpg | art/sb31p017.jpg | art | xtra | 4 | 0.172119,0.008924,0.008226,0.016950,0.793781 | -0.05104202,-0.3935787,-0.73929733,0.86878717,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14799604,-0.62038505,-0.72562873,-0.2654881,-0.6266761,-0.68991846,-0.57515705,-0.41532665,-0.6520286,-0.6898426,-0.34099144,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.2674706,-0.53598946,-0.13334358,-0.5982413,-0.73986673,-0.45211998,-0.68509877,0.033780396,-0.52171326,0.054827005,-0.20721096,0.3301326,-0.73933,-0.57338566,0.34032136,-0.6179389,-0.20170414,0.16225064,-0.44437563,0.44581896,-0.069196045,0.20976627,-0.07846111,-0.6001429,0.38122594,-0.09738222,-0.59497476,-0.665527... | 11 | 191.66959 | 1 | ... | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb31p017.jpg | 1.491735,-0.5714155,-0.73929733,0.05867392,0.5154146,0.0936929,-0.7267234,-0.4896673,0.4246894,-0.42715576,-0.33477074,-0.72562873,1.2274611,-0.6266761,-0.68991846,-0.57515705,-0.5467274,1.1312869,-0.06054777,-0.028933167,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.46393585,0.4817328,0.22738433,-0.73986673,-0.45211998,0.0057709217,-0.67007643,-0.52171326,0.68283725,0.060967267,0.7253895,-0.73933,-0.57338566,-0.65925777,-0.6179389,-0.32987997,0.86336035,-0.6189488,1.3613329,1.2250904,-0.59257406,-0.5857397,-0.6001429,-0.5687719,-0.14814761,0.5325973,-0.66552746,-0.6482407,0.... | 4 | 199.030975 | 4 | 195.975464 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb05p079.jpg | art/sb05p079.jpg | art | xtra | 4 | 0.172158,0.008873,0.008536,0.017017,0.793416 | 1.8089037,-0.5714155,-0.73929733,-0.10773122,-0.22753507,0.34884918,0.124596715,-0.4896673,-0.27447522,-0.42715576,0.61659455,-0.03226477,-0.662768,-0.6266761,-0.68991846,-0.57515705,-0.39897114,-0.6520286,-0.6898426,-0.6497359,0.16184372,-0.55193967,-0.5446661,-0.6779002,0.04647112,-0.6690378,-0.4518578,0.9576235,0.29423904,-0.45211998,-0.68509877,0.09260088,-0.3948297,-0.4592305,-0.600525,-0.6186755,-0.64046603,-0.57338566,0.08902073,0.09568572,-0.61049384,0.7515181,-0.15892029,-0.4098586,1.8483992,-0.59257406,-0.6157326,-0.56642026,-0.651804,-0.44185895,-0.19970179,0.19587982,1.3608505,... | 11 | 140.32997 | 1 | ... | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb05p079.jpg | -0.6013173,-0.5714155,-0.7368871,-0.6553513,-0.6254566,-0.65341413,0.071327925,-0.4896673,-0.7537334,-0.29406685,1.2854458,-0.72562873,1.4279991,-0.2084853,0.4439276,-0.57515705,-0.5467274,-0.2426523,0.6567551,-0.6497359,0.284827,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.39760548,-0.5982413,-0.16403872,0.82081807,-0.68509877,-0.67007643,-0.16007048,-0.23898152,0.4340614,-0.6186755,-0.73933,-0.57338566,-0.65925777,-0.60700464,-0.22504222,1.2887287,0.17025876,-0.5827259,3.245521,-0.59257406,1.607188,-0.6001429,-0.1800242,-0.5315948,0.2059763,0.33526373,0.109642506,-0.0013180... | 4 | 142.766022 | 4 | 248.372314 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb26p082.jpg | art/sb26p082.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | 1.8392665,0.3796271,-0.73929733,0.82189727,1.6830277,0.9302285,-0.5532687,-0.4896673,-0.30764756,-0.42715576,-0.44450876,-0.72562873,-0.662768,0.54702616,-0.68991846,-0.57515705,-0.31611928,-0.59368926,1.2819774,-0.6497359,-0.64108604,-0.55193967,0.84984857,-0.6779002,-0.06536412,-0.6690378,0.34403068,-0.16627067,0.3990196,-0.45211998,0.65170527,-0.67007643,0.6049255,-0.4367706,-0.600525,-0.0046679974,-0.73933,-0.109048486,-0.65925777,0.65358853,-0.61049384,0.70374626,-0.6189488,-0.6682474,-0.0177418,-0.59257406,1.1068444,-0.29651338,-0.651804,-0.5315948,1.8200746,-0.66552746,0.11623299,-0... | 7 | 176.11029 | 1 | ... | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb26p082.jpg | 0.8175719,0.48385304,-0.73929733,1.0291914,1.4682066,1.225123,-0.43518582,-0.4896673,-0.7537334,-0.42715576,0.72162604,-0.72562873,-0.662768,-0.6266761,-0.55470747,-0.57515705,-0.5467274,-0.6520286,2.099903,-0.6497359,-0.64108604,-0.55193967,1.3572993,-0.6779002,1.3149762,0.8512635,0.81857395,-0.5982413,0.056738913,-0.45211998,2.0946088,-0.67007643,-0.52171326,-0.4592305,-0.600525,1.1100725,-0.73933,-0.57338566,-0.24140868,0.9004108,-0.61049384,0.4742828,-0.6189488,-0.6682474,0.335356,0.40166396,0.7790335,-0.6001429,-0.651804,-0.4778452,0.7360641,-0.66552746,-0.6482407,-0.13845,-0.7311095,... | 1 | 157.938324 | 1 | 196.849426 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb69p101-color.jpg | art/sb69p101-color.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | -0.23259398,-0.5714155,-0.73929733,-0.6553513,0.34417403,1.046661,-0.42288247,-0.4896673,0.39017296,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.93892807,-0.57515705,-0.5222921,-0.6520286,-0.6898426,-0.6497359,-0.6366938,-0.55193967,-0.5446661,-0.6779002,-0.46687827,-0.6690378,-0.4518578,-0.5982413,0.7252437,-0.45211998,1.8137504,-0.67007643,-0.52171326,-0.36457688,-0.600525,0.8178272,-0.73933,0.23574609,-0.65925777,-0.6179389,-0.61049384,-0.15158218,-0.41958112,-0.6682474,-0.75794554,0.4943592,0.8210626,-0.6001429,-0.651804,-0.080083996,0.48374963,-0.42743486,-0.6482407,-0.4... | 7 | 187.45346 | 1 | ... | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb69p101-color.jpg | 0.14908737,-0.5714155,-0.73929733,0.44245207,-0.6254566,1.4611686,-0.7267234,-0.4896673,0.87923527,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.4918049,-0.57515705,-0.5467274,-0.6520286,-0.6898426,-0.2860955,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.69009775,-0.6690378,-0.4518578,0.18292433,-0.004632473,-0.45211998,1.7543458,-0.67007643,-0.52171326,-0.41669255,-0.600525,1.4654884,0.5742926,-0.57338566,-0.65925777,-0.6179389,-0.52983093,0.26484245,-0.25630984,-0.6682474,0.19898295,0.95297796,0.38033813,-0.6001429,-0.651804,-0.5315948,1.3264011,0.31547034,-0.6482407,-0.47... | 1 | 167.278229 | 1 | 136.638000 |
5 rows × 25 columns
show_all_clusters(
clusters_df=watercolors_df,
clusters=watercolors_cropped_I,
title="watercolors_cropped_emb",
)
fig = plt.scatter(
x=watercolors_cropped_X_tsne[:, 0],
y=watercolors_cropped_X_tsne[:, 1],
c=watercolors_df.watercolors_cropped_cluster_id,
s=8.0,
)
side quest: varying the numbers of centroids for drawings
drawings_cluster16
ncentroids = 16
niter = 20
verbose = False
drawings_cluster16_kmeans = faiss.Kmeans(
drawings_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cluster16_kmeans.train(drawings_emb)
drawings_cluster16_D, drawings_cluster16_I = drawings_index.search(
drawings_cluster16_kmeans.centroids, 16
)
drawings_cluster16_kD, drawings_cluster16_kI = drawings_cluster16_kmeans.index.search(
drawings_emb, 1
)
drawings_df.loc[:, "drawings_cluster16_id"] = drawings_cluster16_kI
drawings_df.loc[:, "drawings_cluster16_dist"] = drawings_cluster16_kD
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | handlabeled_abs_path | emb_csv_cropped | drawings_cluster_id | drawings_cluster_dist | drawings_cropped_cluster_id | drawings_cropped_cluster_dist | drawings_cluster32_id | drawings_cluster32_dist | drawings_cluster16_id | drawings_cluster16_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... | 3 | 217.249588 | 5 | 332.983734 | 15 | 166.716202 | 3 | 169.585052 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... | 2 | 169.947693 | 5 | 126.699768 | 18 | 137.920044 | 8 | 141.980621 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... | 3 | 162.882431 | 5 | 257.762512 | 8 | 154.434723 | 14 | 161.993011 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... | 4 | 94.257965 | 2 | 145.937225 | 28 | 85.081177 | 15 | 92.158417 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... | 0 | 173.559937 | 5 | 197.142792 | 25 | 167.685181 | 0 | 166.856781 |
5 rows × 29 columns
show_all_clusters(
clusters_df=drawings_df,
clusters=drawings_cluster16_I,
title="drawings_cluster16_emb",
)
fig = plt.scatter(
x=drawings_X_tsne[:, 0],
y=drawings_X_tsne[:, 1],
c=drawings_df.drawings_cluster16_id,
s=2.0,
)
drawings_cluster32
ncentroids = 32
niter = 20
verbose = False
drawings_cluster32_kmeans = faiss.Kmeans(
drawings_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cluster32_kmeans.train(drawings_emb)
drawings_cluster32_D, drawings_cluster32_I = drawings_index.search(
drawings_cluster32_kmeans.centroids, 16
)
drawings_cluster32_kD, drawings_cluster32_kI = drawings_cluster32_kmeans.index.search(
drawings_emb, 1
)
drawings_df.loc[:, "drawings_cluster32_id"] = drawings_cluster32_kI
drawings_df.loc[:, "drawings_cluster32_dist"] = drawings_cluster32_kD
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | drawings_cluster_id | drawings_cluster_dist | drawings_cropped_cluster_id | drawings_cropped_cluster_dist | drawings_cluster32_id | drawings_cluster32_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | False | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p043.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p043.jpg | -0.6198182,-0.5714155,0.2937309,0.2691822,-0.6254566,-0.46855158,-0.7267234,-0.4896673,-0.7537334,-0.42715576,2.241211,-0.013144195,-0.30681813,-0.6266761,0.43516117,-0.57515705,-0.5467274,-0.5245168,1.3334948,2.1400952,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.38178188,-0.5982413,-0.50063276,-0.21540937,0.4948988,-0.67007643,-0.52171326,-0.4592305,1.8570362,-0.6186755,0.94023055,-0.57338566,0.5270257,-0.6179389,2.302543,1.1004333,0.15528035,-0.6682474,1.890626,-0.59257406,1.331521,-0.6001429,1.5220151,0.93749076,-0.59497476,-0.66552746,-0.6482407,-0.4732165,0.... | 3 | 217.249588 | 5 | 332.983734 | 15 | 166.716202 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb67p021.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb67p021.jpg | -0.20756221,-0.24843562,0.29454154,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.9697293,-0.72562873,-0.662768,-0.6266761,-0.23836556,0.4814458,-0.5467274,-0.6520286,0.838902,-0.6497359,-0.57338166,-0.55193967,-0.5446661,-0.28419846,1.6012006,-0.6690378,-0.4518578,-0.0104218125,-0.73986673,-0.3002332,-0.68509877,-0.67007643,-0.36297536,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,-0.65925777,1.6904012,-0.61049384,-0.58021957,0.508085,0.14767355,0.8082305,-0.59257406,-0.6157326,-0.6001429,-0.651804,-0.30467254,0.17396247,-0.66552746,-0.6482407,-0.0... | 2 | 169.947693 | 5 | 126.699768 | 18 | 137.920044 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb25p227.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb25p227.jpg | -0.6198182,-0.428226,-0.73929733,-0.6553513,-0.49369657,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.28293163,-0.62038505,-0.32824886,2.5504894,-0.6266761,0.8397537,0.85716057,-0.5467274,-0.058725297,0.64954805,-0.6497359,-0.64108604,-0.36876833,-0.5446661,-0.29563427,0.08626467,-0.6690378,-0.4518578,0.64337015,-0.73986673,0.018805295,-0.68509877,-0.67007643,1.5876667,-0.4592305,-0.600525,-0.6186755,0.28688914,-0.57338566,-0.65925777,1.6810645,-0.61049384,1.8564332,-0.6189488,-0.6682474,-0.75794554,-0.59257406,-0.6157326,-0.6001429,-0.63579994,-0.45296612,-0.29925343,-0.0263654,-0.64824... | 3 | 162.882431 | 5 | 257.762512 | 8 | 154.434723 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb48p057.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb48p057.jpg | -0.34353483,-0.5714155,2.2665114,-0.6553513,-0.42076486,-0.6254998,-0.7267234,-0.4896673,-0.7537334,-0.42715576,0.6137707,-0.72562873,0.2611376,1.0485492,0.91806084,-0.452704,-0.5467274,-0.6520286,0.621349,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.7967064,0.011958599,-0.4518578,-0.25104153,-0.6493137,-0.45211998,-0.02894944,-0.67007643,1.3037661,-0.4592305,-0.600525,-0.6186755,-0.072172046,-0.57338566,0.42140257,0.8994249,-0.61049384,-0.35844505,-0.03452295,-0.40064916,0.1334514,-0.59257406,-0.06328797,-0.6001429,-0.651804,-0.5315948,0.55393887,-0.66552746,-0.6482407,0.165... | 4 | 94.257965 | 2 | 145.937225 | 28 | 85.081177 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/0_drawings/sb77p044.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/0_drawings/sb77p044.jpg | -0.3477007,-0.5714155,-0.73929733,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14662892,-0.43905735,-0.64813715,0.059202433,-0.6266761,-0.51404697,-0.20505324,-0.5467274,-0.6520286,1.2549256,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.41643643,-0.6690378,-0.22824319,-0.5982413,-0.73986673,-0.3338361,-0.68509877,0.10866904,-0.1451351,-0.4592305,0.5446466,0.7924262,-0.73933,-0.57338566,-0.45801556,-0.11409092,-0.4252534,0.4120332,-0.6189488,-0.28175777,2.210379,-0.59257406,-0.070759535,-0.6001429,-0.651804,0.012511611,0.26820052,-0.66552746,0.9221995,... | 0 | 173.559937 | 5 | 197.142792 | 25 | 167.685181 |
5 rows × 27 columns
show_all_clusters(
clusters_df=drawings_df,
clusters=drawings_cluster32_I,
title="drawings_cluster32_emb",
)
fig = plt.scatter(
x=drawings_X_tsne[:, 0],
y=drawings_X_tsne[:, 1],
c=drawings_df.drawings_cluster32_id,
s=2.0,
)drawings_cluster48
ncentroids = 48
niter = 20
verbose = False
drawings_cluster48_kmeans = faiss.Kmeans(
drawings_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cluster48_kmeans.train(drawings_emb)
drawings_cluster48_D, drawings_cluster48_I = drawings_index.search(
drawings_cluster48_kmeans.centroids, 16
)
drawings_cluster48_kD, drawings_cluster48_kI = drawings_cluster48_kmeans.index.search(
drawings_emb, 1
)
drawings_df.loc[:, "drawings_cluster48_id"] = drawings_cluster48_kI
drawings_df.loc[:, "drawings_cluster48_dist"] = drawings_cluster48_kD
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | drawings_cropped_cluster_id | drawings_cropped_cluster_dist | drawings_cluster32_id | drawings_cluster32_dist | drawings_cluster16_id | drawings_cluster16_dist | drawings_cluster64_id | drawings_cluster64_dist | drawings_cluster48_id | drawings_cluster48_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 5 | 332.983734 | 15 | 166.716202 | 3 | 169.585052 | 58 | 145.230499 | 15 | 166.337051 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 5 | 126.699768 | 18 | 137.920044 | 8 | 141.980621 | 45 | 136.906982 | 45 | 137.201172 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 5 | 257.762512 | 8 | 154.434723 | 14 | 161.993011 | 61 | 150.458466 | 8 | 148.634705 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 2 | 145.937225 | 28 | 85.081177 | 15 | 92.158417 | 28 | 87.053833 | 28 | 83.963623 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 5 | 197.142792 | 25 | 167.685181 | 0 | 166.856781 | 21 | 162.981720 | 39 | 169.478943 |
5 rows × 33 columns
show_all_clusters(
clusters_df=drawings_df,
clusters=drawings_cluster48_I,
title="drawings_cluster48_emb",
)
fig = plt.scatter(
x=drawings_X_tsne[:, 0],
y=drawings_X_tsne[:, 1],
c=drawings_df.drawings_cluster48_id,
s=2.0,
)
drawings_cluster64
ncentroids = 64
niter = 20
verbose = False
drawings_cluster64_kmeans = faiss.Kmeans(
drawings_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cluster64_kmeans.train(drawings_emb)
drawings_cluster64_D, drawings_cluster64_I = drawings_index.search(
drawings_cluster64_kmeans.centroids, 16
)
drawings_cluster64_kD, drawings_cluster64_kI = drawings_cluster64_kmeans.index.search(
drawings_emb, 1
)
drawings_df.loc[:, "drawings_cluster64_id"] = drawings_cluster64_kI
drawings_df.loc[:, "drawings_cluster64_dist"] = drawings_cluster64_kD
drawings_df.head()WARNING clustering 1923 points to 64 centroids: please provide at least 2496 training points| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | drawings_cluster_id | drawings_cluster_dist | drawings_cropped_cluster_id | drawings_cropped_cluster_dist | drawings_cluster32_id | drawings_cluster32_dist | drawings_cluster16_id | drawings_cluster16_dist | drawings_cluster64_id | drawings_cluster64_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 3 | 217.249588 | 5 | 332.983734 | 15 | 166.716202 | 3 | 169.585052 | 58 | 145.230499 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 2 | 169.947693 | 5 | 126.699768 | 18 | 137.920044 | 8 | 141.980621 | 45 | 136.906982 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 3 | 162.882431 | 5 | 257.762512 | 8 | 154.434723 | 14 | 161.993011 | 61 | 150.458466 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 4 | 94.257965 | 2 | 145.937225 | 28 | 85.081177 | 15 | 92.158417 | 28 | 87.053833 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 0 | 173.559937 | 5 | 197.142792 | 25 | 167.685181 | 0 | 166.856781 | 21 | 162.981720 |
5 rows × 31 columns
# show_all_clusters(
# clusters_df=drawings_df,
# clusters=drawings_cluster64_I,
# title="drawings_cluster64_emb",
# )fig = plt.scatter(
x=drawings_X_tsne[:, 0],
y=drawings_X_tsne[:, 1],
c=drawings_df.drawings_cluster64_id,
s=2.0,
)
drawings_cropped variations
drawings_cropped_cluster16
ncentroids = 16
niter = 20
verbose = False
drawings_cropped_cluster16_kmeans = faiss.Kmeans(
drawings_cropped_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cropped_cluster16_kmeans.train(drawings_cropped_emb)
(
drawings_cropped_cluster16_D,
drawings_cropped_cluster16_I,
) = drawings_cropped_index.search(drawings_cropped_cluster16_kmeans.centroids, 16)
(
drawings_cropped_cluster16_kD,
drawings_cropped_cluster16_kI,
) = drawings_cropped_cluster16_kmeans.index.search(drawings_cropped_emb, 1)
drawings_df.loc[:, "drawings_cropped_cluster16_id"] = drawings_cropped_cluster16_kI
drawings_df.loc[:, "drawings_cropped_cluster16_dist"] = drawings_cropped_cluster16_kD
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | drawings_cluster32_id | drawings_cluster32_dist | drawings_cluster16_id | drawings_cluster16_dist | drawings_cluster64_id | drawings_cluster64_dist | drawings_cluster48_id | drawings_cluster48_dist | drawings_cropped_cluster16_id | drawings_cropped_cluster16_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 15 | 166.716202 | 3 | 169.585052 | 58 | 145.230499 | 15 | 166.337051 | 15 | 265.844086 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 18 | 137.920044 | 8 | 141.980621 | 45 | 136.906982 | 45 | 137.201172 | 14 | 121.146545 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 8 | 154.434723 | 14 | 161.993011 | 61 | 150.458466 | 8 | 148.634705 | 10 | 220.298950 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 28 | 85.081177 | 15 | 92.158417 | 28 | 87.053833 | 28 | 83.963623 | 2 | 144.254028 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 25 | 167.685181 | 0 | 166.856781 | 21 | 162.981720 | 39 | 169.478943 | 15 | 151.277313 |
5 rows × 35 columns
# show_all_clusters(
# clusters_df=drawings_df,
# clusters=drawings_cropped_cluster16_I,
# title="drawings_cropped_cluster16_emb",
# )fig = plt.scatter(
x=drawings_cropped_X_tsne[:, 0],
y=drawings_cropped_X_tsne[:, 1],
c=drawings_df.drawings_cropped_cluster16_id,
s=2.0,
)
drawings_cropped_cluster32
ncentroids = 32
niter = 20
verbose = False
drawings_cropped_cluster32_kmeans = faiss.Kmeans(
drawings_cropped_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cropped_cluster32_kmeans.train(drawings_cropped_emb)
(
drawings_cropped_cluster32_D,
drawings_cropped_cluster32_I,
) = drawings_cropped_index.search(drawings_cropped_cluster32_kmeans.centroids, 16)
(
drawings_cropped_cluster32_kD,
drawings_cropped_cluster32_kI,
) = drawings_cropped_cluster32_kmeans.index.search(drawings_cropped_emb, 1)
drawings_df.loc[:, "drawings_cropped_cluster32_id"] = drawings_cropped_cluster32_kI
drawings_df.loc[:, "drawings_cropped_cluster32_dist"] = drawings_cropped_cluster32_kD
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | drawings_cluster16_id | drawings_cluster16_dist | drawings_cluster64_id | drawings_cluster64_dist | drawings_cluster48_id | drawings_cluster48_dist | drawings_cropped_cluster16_id | drawings_cropped_cluster16_dist | drawings_cropped_cluster32_id | drawings_cropped_cluster32_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 3 | 169.585052 | 58 | 145.230499 | 15 | 166.337051 | 15 | 265.844086 | 15 | 257.350555 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 8 | 141.980621 | 45 | 136.906982 | 45 | 137.201172 | 14 | 121.146545 | 31 | 122.172424 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 14 | 161.993011 | 61 | 150.458466 | 8 | 148.634705 | 10 | 220.298950 | 1 | 217.335358 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 15 | 92.158417 | 28 | 87.053833 | 28 | 83.963623 | 2 | 144.254028 | 21 | 129.951965 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 0 | 166.856781 | 21 | 162.981720 | 39 | 169.478943 | 15 | 151.277313 | 15 | 141.268982 |
5 rows × 37 columns
# show_all_clusters(
# clusters_df=drawings_df,
# clusters=drawings_cropped_cluster32_I,
# title="drawings_cropped_cluster32_emb",
# )fig = plt.scatter(
x=drawings_cropped_X_tsne[:, 0],
y=drawings_cropped_X_tsne[:, 1],
c=drawings_df.drawings_cropped_cluster32_id,
s=2.0,
)
drawings_cropped_cluster48
ncentroids = 48
niter = 20
verbose = False
drawings_cropped_cluster48_kmeans = faiss.Kmeans(
drawings_cropped_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cropped_cluster48_kmeans.train(drawings_cropped_emb)
(
drawings_cropped_cluster48_D,
drawings_cropped_cluster48_I,
) = drawings_cropped_index.search(drawings_cropped_cluster48_kmeans.centroids, 16)
(
drawings_cropped_cluster48_kD,
drawings_cropped_cluster48_kI,
) = drawings_cropped_cluster48_kmeans.index.search(drawings_cropped_emb, 1)
drawings_df.loc[:, "drawings_cropped_cluster48_id"] = drawings_cropped_cluster48_kI
drawings_df.loc[:, "drawings_cropped_cluster48_dist"] = drawings_cropped_cluster48_kD
drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | drawings_cluster64_id | drawings_cluster64_dist | drawings_cluster48_id | drawings_cluster48_dist | drawings_cropped_cluster16_id | drawings_cropped_cluster16_dist | drawings_cropped_cluster32_id | drawings_cropped_cluster32_dist | drawings_cropped_cluster48_id | drawings_cropped_cluster48_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 58 | 145.230499 | 15 | 166.337051 | 15 | 265.844086 | 15 | 257.350555 | 15 | 268.672974 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 45 | 136.906982 | 45 | 137.201172 | 14 | 121.146545 | 31 | 122.172424 | 31 | 116.409210 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 61 | 150.458466 | 8 | 148.634705 | 10 | 220.298950 | 1 | 217.335358 | 1 | 213.251678 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 28 | 87.053833 | 28 | 83.963623 | 2 | 144.254028 | 21 | 129.951965 | 21 | 140.072540 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 21 | 162.981720 | 39 | 169.478943 | 15 | 151.277313 | 15 | 141.268982 | 15 | 145.393372 |
5 rows × 39 columns
# show_all_clusters(
# clusters_df=drawings_df,
# clusters=drawings_cropped_cluster48_I,
# title="drawings_cropped_cluster48_emb",
# )fig = plt.scatter(
x=drawings_cropped_X_tsne[:, 0],
y=drawings_cropped_X_tsne[:, 1],
c=drawings_df.drawings_cropped_cluster48_id,
s=2.0,
)
drawings_cropped_cluster64
ncentroids = 64
niter = 20
verbose = False
drawings_cropped_cluster64_kmeans = faiss.Kmeans(
drawings_cropped_emb.shape[1], ncentroids, niter=niter, verbose=verbose, gpu=True
)
drawings_cropped_cluster64_kmeans.train(drawings_cropped_emb)
(
drawings_cropped_cluster64_D,
drawings_cropped_cluster64_I,
) = drawings_cropped_index.search(drawings_cropped_cluster64_kmeans.centroids, 16)
(
drawings_cropped_cluster64_kD,
drawings_cropped_cluster64_kI,
) = drawings_cropped_cluster64_kmeans.index.search(drawings_cropped_emb, 1)
drawings_df.loc[:, "drawings_cropped_cluster64_id"] = drawings_cropped_cluster64_kI
drawings_df.loc[:, "drawings_cropped_cluster64_dist"] = drawings_cropped_cluster64_kD
drawings_df.head()WARNING clustering 1923 points to 64 centroids: please provide at least 2496 training points| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | drawings_cluster48_id | drawings_cluster48_dist | drawings_cropped_cluster16_id | drawings_cropped_cluster16_dist | drawings_cropped_cluster32_id | drawings_cropped_cluster32_dist | drawings_cropped_cluster48_id | drawings_cropped_cluster48_dist | drawings_cropped_cluster64_id | drawings_cropped_cluster64_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 15 | 166.337051 | 15 | 265.844086 | 15 | 257.350555 | 15 | 268.672974 | 58 | 276.820343 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 45 | 137.201172 | 14 | 121.146545 | 31 | 122.172424 | 31 | 116.409210 | 34 | 122.042419 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 8 | 148.634705 | 10 | 220.298950 | 1 | 217.335358 | 1 | 213.251678 | 1 | 191.671631 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 28 | 83.963623 | 2 | 144.254028 | 21 | 129.951965 | 21 | 140.072540 | 21 | 144.775696 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 39 | 169.478943 | 15 | 151.277313 | 15 | 141.268982 | 15 | 145.393372 | 58 | 151.138092 |
5 rows × 41 columns
# show_all_clusters(
# clusters_df=drawings_df,
# clusters=drawings_cropped_cluster64_I,
# title="drawings_cropped_cluster64_emb",
# )fig = plt.scatter(
x=drawings_cropped_X_tsne[:, 0],
y=drawings_cropped_X_tsne[:, 1],
c=drawings_df.drawings_cropped_cluster64_id,
s=2.0,
)
saving cluster outputs
save the TSV’s with cluster ID’s
preprocessing_dir'/Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP'subclusters_dir = (
singleline_data_home() / "raster/epoch-20231214/04_SUBCLUSTERS"
)
if not os.path.isdir(subclusters_dir):
os.makedirs(subclusters_dir)drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | drawings_cluster48_id | drawings_cluster48_dist | drawings_cropped_cluster16_id | drawings_cropped_cluster16_dist | drawings_cropped_cluster32_id | drawings_cropped_cluster32_dist | drawings_cropped_cluster48_id | drawings_cropped_cluster48_dist | drawings_cropped_cluster64_id | drawings_cropped_cluster64_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 15 | 166.337051 | 15 | 265.844086 | 15 | 257.350555 | 15 | 268.672974 | 58 | 276.820343 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 45 | 137.201172 | 14 | 121.146545 | 31 | 122.172424 | 31 | 116.409210 | 34 | 122.042419 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 8 | 148.634705 | 10 | 220.298950 | 1 | 217.335358 | 1 | 213.251678 | 1 | 191.671631 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 28 | 83.963623 | 2 | 144.254028 | 21 | 129.951965 | 21 | 140.072540 | 21 | 144.775696 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 39 | 169.478943 | 15 | 151.277313 | 15 | 141.268982 | 15 | 145.393372 | 58 | 151.138092 |
5 rows × 41 columns
clustered_drawings_tsv_fname = f"{subclusters_dir}/clustered_drawings.tsv"
drawings_df.to_csv(
clustered_drawings_tsv_fname, index=True, index_label="idx", sep="\t", header=True
)
clustered_drawings_df = pd.read_csv(
clustered_drawings_tsv_fname, delimiter="\t", index_col="idx"
)
clustered_drawings_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | drawings_cluster48_id | drawings_cluster48_dist | drawings_cropped_cluster16_id | drawings_cropped_cluster16_dist | drawings_cropped_cluster32_id | drawings_cropped_cluster32_dist | drawings_cropped_cluster48_id | drawings_cropped_cluster48_dist | drawings_cropped_cluster64_id | drawings_cropped_cluster64_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| idx | |||||||||||||||||||||
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p043.jpg | art/sb77p043.jpg | art | xtra | 4 | 0.168532,0.008473,0.008068,0.016940,0.797986 | -0.6198182,-0.5714155,-0.6871516,-0.6553513,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.42715576,1.4904574,-0.72562873,-0.662768,-0.6266761,0.6913696,0.054546177,-0.5467274,-0.6520286,0.117179275,0.94330966,0.5815062,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.4584825,-0.3818994,-0.5750927,-0.73986673,-0.45211998,-0.6141643,-0.67007643,-0.52171326,-0.4592305,1.4788612,-0.6186755,0.799198,-0.57338566,-0.17778617,0.10336053,0.093812525,0.17924881,-0.6189488,-0.6682474,2.1040225,-0.59257406,1.6553459,-0.6001429,0.5109105,0.19352531,-0.2517513,-0.66552746,-0.6482407,-0.4732... | 11 | 199.22961 | 1 | ... | 15 | 166.33705 | 15 | 265.844100 | 15 | 257.350560 | 15 | 268.67297 | 58 | 276.82034 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb67p021.jpg | art/sb67p021.jpg | art | xtra | 4 | 0.170976,0.008673,0.008480,0.017028,0.794843 | 1.1073987,0.3512016,-0.73929733,-0.6553513,1.7477593,-0.65341413,0.32179314,-0.4896673,-0.1797334,-0.42715576,1.3163843,-0.72562873,-0.61387855,-0.6266761,-0.35361218,0.37611246,-0.08771938,-0.5262803,0.9261699,-0.6497359,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.32559583,-0.6690378,-0.4518578,0.51828945,-0.4238917,-0.45211998,-0.16527033,-0.67007643,1.0136865,-0.4592305,-0.600525,-0.060860336,-0.19446748,-0.24127388,2.2282472,1.8003348,-0.61049384,0.39795423,0.2626753,-0.6682474,-0.75794554,-0.59257406,0.80257696,-0.6001429,-0.651804,-0.028328001,0.53641737,-0.6501347,-0.6482407,0.... | 3 | 132.26111 | 0 | ... | 45 | 137.20117 | 14 | 121.146545 | 31 | 122.172424 | 31 | 116.40921 | 34 | 122.04242 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb25p227.jpg | art/sb25p227.jpg | art | xtra | 4 | 0.158396,0.008277,0.007424,0.016584,0.809319 | 1.3776337,-0.5714155,-0.2587232,-0.6553513,1.3532461,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.25077122,0.050992966,-0.72562873,0.1406585,-0.6266761,1.0653365,0.96198404,-0.5467274,0.30261123,0.6327609,-0.6497359,-0.64108604,-0.28434965,-0.35651857,-0.6779002,-0.28245115,-0.6690378,-0.4518578,0.030088186,-0.73986673,-0.18931103,-0.041653574,-0.67007643,-0.52171326,-0.29881266,-0.600525,-0.23231092,-0.73933,-0.57338566,-0.65925777,1.8833572,-0.61049384,0.44560486,0.14075392,-0.6682474,-0.75794554,-0.59257406,-0.07511848,-0.6001429,-0.651804,-0.5315948,0.023931146,0.723624,-0.6482407,-... | 8 | 188.58075 | 0 | ... | 8 | 148.63470 | 10 | 220.298950 | 1 | 217.335360 | 1 | 213.25168 | 1 | 191.67163 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb48p057.jpg | art/sb48p057.jpg | art | xtra | 4 | 0.165322,0.008453,0.008010,0.016587,0.801628 | 2.4936543,0.17563903,-0.6522293,-0.55642915,1.1084181,-0.5028205,-0.046447277,-0.4896673,-0.7537334,-0.42715576,0.84124434,-0.72562873,-0.52445275,-0.5557985,-0.21369287,2.2372599,-0.26381853,-0.6520286,-0.6898426,-0.6497359,-0.64108604,0.43080842,-0.23286673,-0.6779002,0.7969176,-0.6690378,-0.4518578,0.747658,0.04896885,-0.45211998,-0.68509877,-0.2462106,0.7319356,-0.4592305,-0.600525,-0.6186755,-0.5338096,-0.57338566,1.1886733,0.04046011,-0.61049384,0.45326668,0.72983754,-0.6682474,-0.10000485,0.034278154,-0.33160114,-0.6001429,-0.651804,-0.5315948,-0.20898154,-0.66552746,1.2346795,-0.47... | 14 | 90.26291 | 0 | ... | 28 | 83.96362 | 2 | 144.254030 | 21 | 129.951970 | 21 | 140.07254 | 21 | 144.77570 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb77p044.jpg | art/sb77p044.jpg | art | xtra | 4 | 0.171638,0.008182,0.008061,0.016469,0.795650 | 2.170152,-0.01916939,-0.47766963,-0.6553513,-0.44795185,-0.65341413,1.5596724,-0.4896673,-0.7537334,-0.42715576,1.0671424,-0.72562873,1.8207353,-0.6266761,-0.6194963,1.1724913,-0.5467274,0.12343103,1.5514618,-0.6497359,0.23890108,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.33040416,-0.4518578,0.39869583,0.120936334,-0.45211998,-0.68509877,0.2514736,0.17836922,-0.4592305,-0.600525,-0.6186755,-0.73933,-0.57338566,1.3668395,0.6006557,1.2428443,1.0844817,0.36636358,0.48543411,-0.5954572,-0.59257406,-0.30289415,1.2783914,0.5800507,-0.23473695,-0.59497476,-0.66552746,2.2758756,-0.4732165,-0.... | 4 | 176.29102 | 0 | ... | 39 | 169.47894 | 15 | 151.277310 | 15 | 141.268980 | 15 | 145.39337 | 58 | 151.13809 |
5 rows × 41 columns
clustered_watercolors_tsv_fname = f"{subclusters_dir}/clustered_watercolors.tsv"
watercolors_df.to_csv(
clustered_watercolors_tsv_fname,
index=True,
index_label="idx",
sep="\t",
header=True,
)
clustered_watercolors_df = pd.read_csv(
clustered_watercolors_tsv_fname, delimiter="\t", index_col="idx"
)
clustered_watercolors_df.head()| abs_fname | rel_fname | label | pred_label | pred_idx | pred_probs | emb_csv | cluster | cluster_dist | metacluster | ... | handlabeled_metacluster_name | handlabeled_metacluster_id | handlabeled_metacluster_was_correct | preprocessed_abs_path | handlabeled_abs_path | emb_csv_cropped | watercolors_cluster_id | watercolors_cluster_dist | watercolors_cropped_cluster_id | watercolors_cropped_cluster_dist | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| idx | |||||||||||||||||||||
| 0 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb07p104.jpg | art/sb07p104.jpg | art | xtra | 4 | 0.177406,0.008799,0.008245,0.017177,0.788373 | -0.39105213,0.96309596,-0.73929733,0.4489988,0.8297614,-0.3937308,-0.7267234,-0.32461396,-0.7537334,-0.42715576,-0.62038505,-0.72562873,0.3037846,-0.6266761,-0.5213815,-0.31669965,-0.5467274,-0.6520286,-0.22544897,-0.6497359,-0.64108604,0.6374447,-0.5446661,-0.6779002,-0.14143991,-0.6690378,-0.44753593,0.66945183,-0.73986673,-0.45211998,-0.05996394,-0.21525535,0.82928157,-0.04412651,-0.600525,-0.04148197,-0.73825717,-0.57338566,-0.25227422,2.3566113,-0.61049384,1.0027854,-0.40134108,1.3733256,-0.75794554,-0.59257406,-0.4395525,-0.6001429,1.0358506,1.1911299,0.8162154,0.032713354,-0.6482407... | 5 | 238.33170 | 1 | ... | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb07p104.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb07p104.jpg | -0.6198182,0.8895926,-0.73929733,0.6418346,0.46164584,1.0085131,-0.7267234,-0.4896673,-0.7537334,-0.42715576,-0.62038505,-0.72562873,1.4399182,-0.6266761,0.474914,-0.57515705,-0.5467274,-0.6520286,1.2274435,-0.6497359,-0.27622125,-0.45870692,-0.5446661,-0.6779002,1.3343718,-0.6690378,-0.4518578,0.42277813,-0.73986673,-0.45211998,1.6291834,-0.67007643,-0.09817025,-0.00090405345,-0.600525,-0.6186755,-0.73933,-0.57338566,0.006386578,1.034553,-0.61049384,-0.58021957,-0.09483403,0.78779536,-0.73080033,0.10947651,0.3367887,-0.5382227,-0.26548737,0.5630663,2.15088,0.2266624,-0.6482407,0.049348265... | 7 | 215.02927 | 7 | 206.64117 |
| 1 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb31p017.jpg | art/sb31p017.jpg | art | xtra | 4 | 0.172119,0.008924,0.008226,0.016950,0.793781 | -0.05104202,-0.3935787,-0.73929733,0.86878717,-0.6254566,-0.65341413,-0.7267234,-0.4896673,-0.7537334,-0.14799604,-0.62038505,-0.72562873,-0.2654881,-0.6266761,-0.68991846,-0.57515705,-0.41532665,-0.6520286,-0.6898426,-0.34099144,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.2674706,-0.53598946,-0.13334358,-0.5982413,-0.73986673,-0.45211998,-0.68509877,0.033780396,-0.52171326,0.054827005,-0.20721096,0.3301326,-0.73933,-0.57338566,0.34032136,-0.6179389,-0.20170414,0.16225064,-0.44437563,0.44581896,-0.069196045,0.20976627,-0.07846111,-0.6001429,0.38122594,-0.09738222,-0.59497476,-0.665527... | 11 | 191.66959 | 1 | ... | 1_watercolors | 1 | True | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb31p017.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb31p017.jpg | 1.491735,-0.5714155,-0.73929733,0.05867392,0.5154146,0.0936929,-0.7267234,-0.4896673,0.4246894,-0.42715576,-0.33477074,-0.72562873,1.2274611,-0.6266761,-0.68991846,-0.57515705,-0.5467274,1.1312869,-0.06054777,-0.028933167,-0.64108604,-0.55193967,-0.5446661,-0.6779002,-0.57759243,0.46393585,0.4817328,0.22738433,-0.73986673,-0.45211998,0.0057709217,-0.67007643,-0.52171326,0.68283725,0.060967267,0.7253895,-0.73933,-0.57338566,-0.65925777,-0.6179389,-0.32987997,0.86336035,-0.6189488,1.3613329,1.2250904,-0.59257406,-0.5857397,-0.6001429,-0.5687719,-0.14814761,0.5325973,-0.66552746,-0.6482407,0.... | 4 | 199.03098 | 4 | 195.97546 |
| 2 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb05p079.jpg | art/sb05p079.jpg | art | xtra | 4 | 0.172158,0.008873,0.008536,0.017017,0.793416 | 1.8089037,-0.5714155,-0.73929733,-0.10773122,-0.22753507,0.34884918,0.124596715,-0.4896673,-0.27447522,-0.42715576,0.61659455,-0.03226477,-0.662768,-0.6266761,-0.68991846,-0.57515705,-0.39897114,-0.6520286,-0.6898426,-0.6497359,0.16184372,-0.55193967,-0.5446661,-0.6779002,0.04647112,-0.6690378,-0.4518578,0.9576235,0.29423904,-0.45211998,-0.68509877,0.09260088,-0.3948297,-0.4592305,-0.600525,-0.6186755,-0.64046603,-0.57338566,0.08902073,0.09568572,-0.61049384,0.7515181,-0.15892029,-0.4098586,1.8483992,-0.59257406,-0.6157326,-0.56642026,-0.651804,-0.44185895,-0.19970179,0.19587982,1.3608505,... | 11 | 140.32997 | 1 | ... | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb05p079.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb05p079.jpg | -0.6013173,-0.5714155,-0.7368871,-0.6553513,-0.6254566,-0.65341413,0.071327925,-0.4896673,-0.7537334,-0.29406685,1.2854458,-0.72562873,1.4279991,-0.2084853,0.4439276,-0.57515705,-0.5467274,-0.2426523,0.6567551,-0.6497359,0.284827,-0.55193967,-0.5446661,-0.6779002,-0.57759243,-0.6690378,0.39760548,-0.5982413,-0.16403872,0.82081807,-0.68509877,-0.67007643,-0.16007048,-0.23898152,0.4340614,-0.6186755,-0.73933,-0.57338566,-0.65925777,-0.60700464,-0.22504222,1.2887287,0.17025876,-0.5827259,3.245521,-0.59257406,1.607188,-0.6001429,-0.1800242,-0.5315948,0.2059763,0.33526373,0.109642506,-0.0013180... | 4 | 142.76602 | 4 | 248.37231 |
| 3 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb26p082.jpg | art/sb26p082.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | 1.8392665,0.3796271,-0.73929733,0.82189727,1.6830277,0.9302285,-0.5532687,-0.4896673,-0.30764756,-0.42715576,-0.44450876,-0.72562873,-0.662768,0.54702616,-0.68991846,-0.57515705,-0.31611928,-0.59368926,1.2819774,-0.6497359,-0.64108604,-0.55193967,0.84984857,-0.6779002,-0.06536412,-0.6690378,0.34403068,-0.16627067,0.3990196,-0.45211998,0.65170527,-0.67007643,0.6049255,-0.4367706,-0.600525,-0.0046679974,-0.73933,-0.109048486,-0.65925777,0.65358853,-0.61049384,0.70374626,-0.6189488,-0.6682474,-0.0177418,-0.59257406,1.1068444,-0.29651338,-0.651804,-0.5315948,1.8200746,-0.66552746,0.11623299,-0... | 7 | 176.11029 | 1 | ... | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb26p082.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb26p082.jpg | 0.8175719,0.48385304,-0.73929733,1.0291914,1.4682066,1.225123,-0.43518582,-0.4896673,-0.7537334,-0.42715576,0.72162604,-0.72562873,-0.662768,-0.6266761,-0.55470747,-0.57515705,-0.5467274,-0.6520286,2.099903,-0.6497359,-0.64108604,-0.55193967,1.3572993,-0.6779002,1.3149762,0.8512635,0.81857395,-0.5982413,0.056738913,-0.45211998,2.0946088,-0.67007643,-0.52171326,-0.4592305,-0.600525,1.1100725,-0.73933,-0.57338566,-0.24140868,0.9004108,-0.61049384,0.4742828,-0.6189488,-0.6682474,0.335356,0.40166396,0.7790335,-0.6001429,-0.651804,-0.4778452,0.7360641,-0.66552746,-0.6482407,-0.13845,-0.7311095,... | 1 | 157.93832 | 1 | 196.84943 |
| 4 | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/01_FLAT/art/sb69p101-color.jpg | art/sb69p101-color.jpg | art | xtra | 4 | 0.179525,0.008828,0.008282,0.017196,0.786169 | -0.23259398,-0.5714155,-0.73929733,-0.6553513,0.34417403,1.046661,-0.42288247,-0.4896673,0.39017296,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.93892807,-0.57515705,-0.5222921,-0.6520286,-0.6898426,-0.6497359,-0.6366938,-0.55193967,-0.5446661,-0.6779002,-0.46687827,-0.6690378,-0.4518578,-0.5982413,0.7252437,-0.45211998,1.8137504,-0.67007643,-0.52171326,-0.36457688,-0.600525,0.8178272,-0.73933,0.23574609,-0.65925777,-0.6179389,-0.61049384,-0.15158218,-0.41958112,-0.6682474,-0.75794554,0.4943592,0.8210626,-0.6001429,-0.651804,-0.080083996,0.48374963,-0.42743486,-0.6482407,-0.4... | 7 | 187.45346 | 1 | ... | 1_watercolors | 1 | NaN | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/10_SIMPLE_CROP/1_watercolors/sb69p101-color.jpg | /Users/al/Dropbox/2-Areas/2-Sketchbooks/datasets/full-v2/03_VISUAL_HAND_LABELED/1_watercolors/sb69p101-color.jpg | 0.14908737,-0.5714155,-0.73929733,0.44245207,-0.6254566,1.4611686,-0.7267234,-0.4896673,0.87923527,-0.42715576,-0.62038505,-0.72562873,-0.662768,-0.6266761,0.4918049,-0.57515705,-0.5467274,-0.6520286,-0.6898426,-0.2860955,-0.64108604,-0.55193967,-0.5446661,-0.6779002,0.69009775,-0.6690378,-0.4518578,0.18292433,-0.004632473,-0.45211998,1.7543458,-0.67007643,-0.52171326,-0.41669255,-0.600525,1.4654884,0.5742926,-0.57338566,-0.65925777,-0.6179389,-0.52983093,0.26484245,-0.25630984,-0.6682474,0.19898295,0.95297796,0.38033813,-0.6001429,-0.651804,-0.5315948,1.3264011,0.31547034,-0.6482407,-0.47... | 1 | 167.27823 | 1 | 136.63800 |
5 rows × 25 columns
drawing_centroids = {
"drawings_emdb": {
"drawings_cluster8": (drawings_D.tolist(), drawings_I.tolist()),
"drawings_cluster16": (
drawings_cluster16_D.tolist(),
drawings_cluster16_I.tolist(),
),
"drawings_cluster32": (
drawings_cluster32_D.tolist(),
drawings_cluster32_I.tolist(),
),
"drawings_cluster48": (
drawings_cluster48_D.tolist(),
drawings_cluster48_I.tolist(),
),
"drawings_cluster64": (
drawings_cluster64_D.tolist(),
drawings_cluster64_I.tolist(),
),
},
"drawings_cropped_emdb": {
"drawings_cropped_cluster8": (
drawings_cropped_D.tolist(),
drawings_cropped_I.tolist(),
),
"drawings_cropped_cluster16": (
drawings_cropped_cluster16_D.tolist(),
drawings_cropped_cluster16_I.tolist(),
),
"drawings_cropped_cluster32": (
drawings_cropped_cluster32_D.tolist(),
drawings_cropped_cluster32_I.tolist(),
),
"drawings_cropped_cluster48": (
drawings_cropped_cluster48_D.tolist(),
drawings_cropped_cluster48_I.tolist(),
),
"drawings_cropped_cluster64": (
drawings_cropped_cluster64_D.tolist(),
drawings_cropped_cluster64_I.tolist(),
),
},
}# see also: https://github.com/facebookresearch/faiss/issues/2087
drawing_centroids_fname = f"{subclusters_dir}/centroids_drawings.json"
with open(drawing_centroids_fname, "w") as outfile:
json.dump(drawing_centroids, outfile, indent=2)watercolor_centroids = {
"watercolors_emdb": {
"drawings_cluster8": (watercolors_D.tolist(), watercolors_I.tolist()),
},
"watercolors_cropped_emdb": {
"drawings_cropped_cluster8": (
watercolors_cropped_D.tolist(),
watercolors_cropped_I.tolist(),
),
},
}watercolor_centroids_fname = f"{subclusters_dir}/centroids_watercolors.json"
with open(watercolor_centroids_fname, "w") as outfile:
json.dump(watercolor_centroids, outfile, indent=2)from pprint import pprint
pprint(watercolor_centroids)# finish saving