animetimm/danbooru-wdtagger-v4-w640-ws-30k

Danbooru WDTagger V4 Webdataset 30k Sub-Dataset

This is the webdataset subset dataset for animetimm/danbooru-wdtagger-v4-w640.

Metadata here are cleaned by @SmilingWolf, can be used for multi-label classification training.

Images here are resized to min(width, height) <= 640.

How to Use It

from datasets import load_dataset

dataset = load_dataset('animetimm/danbooru-wdtagger-v4-w640-ws-30k')
print(dataset["train"][0])

Images

29983 images in total.

Split	Image Count	Total Size
train	27499	1.96 GB
test	1274	89.5 MB
val	1210	86.4 MB

Tags

322 tags in total.

Rating Tags

4 selected rating tags (4 of them shown) in total.

tag_id	name	category	count	selected_count	train_count	test_count	val_count
9999999	general	9	2782276	5622	5132	241	249
9999998	sensitive	9	4478213	14416	13176	633	607
9999997	questionable	9	1032970	4936	4522	224	190
9999996	explicit	9	900939	5009	4669	176	164

General Tags

318 selected general tags (50 of them shown) in total.

tag_id	name	category	count	selected_count	train_count	test_count	val_count
13197	long_hair	0	4677442	17469	16057	723	689
470575	1girl	0	6421421	17244	15847	716	681
8601	breasts	0	3696520	15521	14322	626	573
3389	blush	0	3145456	13395	12365	520	510
11906	open_mouth	0	2563264	12946	11928	517	501
1815	smile	0	3103828	12670	11695	479	496
469576	looking_at_viewer	0	3591347	12043	11121	477	445
566835	multiple_girls	0	1638110	11221	10269	479	473
12590	shirt	0	2015187	9992	9231	405	356
15080	short_hair	0	2413622	9931	9139	396	396
429	skirt	0	1665177	9623	8833	411	379
375387	holding	0	1511512	9573	8838	380	355
1882	thighhighs	0	1238300	8803	8130	351	322
5735	gloves	0	1462455	8567	7896	329	342
212816	solo	0	5341234	8388	7731	347	310
383159	long_sleeves	0	1715960	7873	7255	313	305
2241	dress	0	1409604	7830	7192	301	337
10959	blue_eyes	0	1887739	7704	7061	336	307
380350	hair_ornament	0	1531545	7641	7035	311	295
87788	blonde_hair	0	1660630	7529	6910	322	297
4563	bow	0	1265406	7028	6515	254	259
13200	black_hair	0	1629597	6734	6219	263	252
12289	sitting	0	1009335	6714	6167	285	262
1821	2girls	0	1096035	6622	6061	282	279
2373	hat	0	1291706	6607	6080	284	243
540830	1boy	0	1534705	6551	6038	250	263
6126	animal_ears	0	1266631	6368	5838	273	257
16613	jewelry	0	1160701	6302	5821	252	229
16867	brown_hair	0	1628876	6300	5797	255	248
403247	large_breasts	0	1695376	6250	5773	246	231
8526	red_eyes	0	1366138	6050	5532	249	269
658573	hair_between_eyes	0	1271873	6029	5611	227	191
464575	ribbon	0	1156999	5815	5378	220	217
9294	navel	0	1229063	5713	5258	242	213
464906	underwear	0	833693	5526	5115	204	207
376054	bare_shoulders	0	1042437	5475	5085	188	202
15674	tail	0	848270	5417	4979	223	215
1709	twintails	0	961709	5176	4777	210	189
376766	white_shirt	0	947457	5139	4752	199	188
572080	closed_mouth	0	1309112	5138	4765	188	185
15675	standing	0	980391	5061	4696	186	179
3843	jacket	0	1040203	4962	4570	207	185
417660	very_long_hair	0	1034958	4936	4577	185	174
608813	full_body	0	895769	4843	4468	183	192
391	panties	0	678753	4666	4322	172	172
10953	blue_hair	0	922475	4545	4169	187	189
667868	medium_breasts	0	942601	4496	4145	178	173
465619	closed_eyes	0	766933	4486	4156	191	139
390728	multicolored_hair	0	773028	4319	3983	174	162
5827	boots	0	517650	4222	3887	167	168

Citation

@inproceedings{danbooruwdtaggerv4webdataset30ksubdataset,
  author = {narugo1992 and Smilingwolf},
  title = {Danbooru WDTagger V4 Webdataset 30k Sub-Dataset},
  year = {2025},
  url = {https://huggingface.co/narugo and https://huggingface.co/SmilingWolf},
  note = {Dataset available at https://huggingface.co/datasets/animetimm/danbooru-wdtagger-v4-w640-ws-30k}
}