└── README.md /README.md: -------------------------------------------------------------------------------- 1 | # DATASOURCE for *fAshIon* :thought_balloon: 2 | 3 | This is a **summary**. We reviewed all (to our best knowledge) fashion-related papers in the past decade and recorded the datasets had been used. The numbers to describe the dataset is faithfully followed its original paper. The webpage is organized as: 4 | 5 | The sections are defined according to the types of data, *e.g.* if you want clothing segmentation information, you can see Section 0. parsing to find annotated data. 6 | 7 | If you want to obtain some attributes, you can see Section 1. attribute. We present the type of attributes (*e.g.* brand, review, style, comment, neckline, color *etc*), the number of images, potential tasks, type of images (*e.g.* product image, model image, which view, *etc*) for a quick check. 8 | 9 | :pig: means the dataset can be found. If you find the dataset helpful, please kindly cite it in your paper ("bibtex" is offered for your convenience)~ 10 | 11 | Meanwhile, for consistency, we uniformed the words decribe the fashion concept. 12 | -  **Silhouette** (shape, cut, fit): the shape of a garment, *e.g.* H line, A line *etc*; 13 | -  **Material** (fabric): the material made a garment, *e.g.* chiffon, lace *etc*; 14 | -  **Print** (pattern, texture): the surface design of a garment, *e.g.* checks, dotted *etc*; 15 | -  **Neckline** (collar shape, collar): the design in the neck region of a garment, *e.g.* V-neck, lapel *etc*; 16 | -  **Design details** (structures, style): designs which can be used in anywhere of a garment, *e.g.* frilly, ruffled *etc*; 17 | -  **Opening** (cloth button, fastening): the way designed in the opening of a garment *e.g.* button, zipped *etc*; 18 | -  **Category** (type): type of a garment, *e.g.* dress, top *etc*; 19 | -  **Sub-category**: fine-grained type of a garment, *e.g.* wedding dress, T-shire *etc*; 20 | -  **Styles** (looks): the expressed feeling of a garment of an outfit, *e.g.* lovely, casual *etc*; 21 | -  **Gender** (Persons): Men's wear, women's wear (child, boy, female) *etc*; 22 | -  **Design Attributes**: the attributes used in the process of garment design, *e.g.* shirt cuff, shirt collar *etc*; 23 | -  **Retail Attributes**: the attributes used in the process of retail, *e.g.* parka, windbreaker *etc*. 24 | 25 | If you have problems with a specific dataset shows below, please kindly contact its authors. For a quick check, you can also see my own [memo version](https://drive.google.com/open?id=1ucwMgee0Df1P--cDHQR9XdcPPuMjgaSt) © 26 | 27 | 28 | 29 | - [0.  Parsing](#0-parsing) 30 | - [1.  Keypoints](#1-keypoints) 31 | - [2.  Attribute](#2-attribute) 32 | - [3.  Outfit](#3-outfit) 33 | - [4.  Generation](#4-generation) 34 | - [5.  Others](#5-others) 35 | - [Other Sources](#other-sources) 36 | - [Acknowledge](#acknowledge) 37 | 38 | 39 | 40 | ## 0.  Parsing 41 | For semantic segmentation, object detection, instance segmentation, polygon detection, and *etc*. 42 | ***** 43 | :cherries: **Fashionista 2012** 44 | 45 | (1) **158,235** fashion photos with associated text annotations (tags, comments, and links). 46 | (2) The tags are noisy or incomplete. 47 | (3) **685** photos with good visibility of the full-body with pose annotations for the usual 14 body parts. 48 | (4) There are totally **56** labels (**53** category or sub-category labels, and additional labels for hair, skin, and null (background). 49 | 50 | [[homepage]](http://vision.is.tohoku.ac.jp/~kyamagu/research/clothing_parsing/) [[pdf]](http://vision.is.tohoku.ac.jp/~kyamagu/papers/yamaguchi_cvpr2012.pdf) :pig: 51 | ```bib 52 | @inproceedings{yamaguchi2012parsing, 53 | title={Parsing clothing in fashion photographs}, 54 | author={Yamaguchi, Kota and Kiapour, M Hadi and Ortiz, Luis E and Berg, Tamara L}, 55 | booktitle={2012 IEEE Conference on Computer Vision and Pattern Recognition}, 56 | pages={3570--3577}, 57 | year={2012}, 58 | organization={IEEE} 59 | } 60 | ``` 61 | 62 | 63 | :cherries: **Paperdoll 2013** 64 | 65 | (1) Over **1 million** pictures from chictopia.com with associated metadata tags i.e. color, clothing item, or occasion. 66 | (2) **339,797** pictures weakly annotated with clothing items and estimated pose. 67 | (3) **685** fully parsed images . 68 | 69 | [[homepage]](http://vision.is.tohoku.ac.jp/~kyamagu/research/paperdoll/) [[pdf]](https://www.cv-foundation.org/openaccess/content_iccv_2013/papers/Yamaguchi_Paper_Doll_Parsing_2013_ICCV_paper.pdf) :pig: 70 | ```bib 71 | @inproceedings{yamaguchi2013paper, 72 | title={Paper doll parsing: Retrieving similar styles to parse clothing items}, 73 | author={Yamaguchi, Kota and Hadi Kiapour, M and Berg, Tamara L}, 74 | booktitle={Proceedings of the IEEE international conference on computer vision}, 75 | pages={3519--3526}, 76 | year={2013} 77 | } 78 | ``` 79 | 80 | 81 | :cherries: **CFPD 2013** 82 | 83 | (1) **97,490** images with 292,541 tags from Chictopia.com. 84 | (2) **2,682** images in total, and all the pixels in the images are annotated with both color labels (**13**) and category labels (**23**). 85 | (3) **Weakly supervised setting**, where only image-level tags are available in the training phase. 86 | 87 | [homepage] [[pdf]](https://liusi-group.com/pdf/Fashion%20Parsing%20With%20Weak%20Color-Category%20Labels.pdf) [[github1]](https://github.com/zbxzc35/dataset-CFPD) [[github2]](https://github.com/hrsma2i/dataset-CFPD) :pig: 88 | ```bib 89 | @article{liu2013fashion, 90 | title={Fashion parsing with weak color-category labels}, 91 | author={Liu, Si and Feng, Jiashi and Domokos, Csaba and Xu, Hui and Huang, Junshi and Hu, Zhenzhen and Yan, Shuicheng}, 92 | journal={IEEE Transactions on Multimedia}, 93 | volume={16}, 94 | number={1}, 95 | pages={253--265}, 96 | year={2013}, 97 | publisher={IEEE} 98 | } 99 | ``` 100 | 101 | 102 | :cherries: **CCP 2013** 103 | 104 | (1) It consisting of **2098** high-resolution street fashion photos. 105 | (2) More than **1,000** images are annotated with superpixel-level labeling with a total of **57** tags. 106 | (3) Cross-scenario image pairs, which include about **10,000** product photos and user's photos image pairs. 107 | (4) Each image has **124** fine-grained semantic attributes. 108 | (5) **20** categories, **56** colors, **6** clothing length, **10** silhouette, **25** necklines, and **7** sleeve length. 109 | 110 | [[homepage]](http://www.sysu-hcp.net/clothing-co-parsing-by-joint-image-segmentation-and-labeling/) [[pdf]](http://linliang.net/wp-content/uploads/2017/07/TMM_Clothes.pdf) [[github]](https://github.com/bearpaw/clothing-co-parsing) :pig: 111 | ```bib 112 | @inproceedings{yang2014clothing, 113 | title={Clothing Co-Parsing by Joint Image Segmentation and Labeling}, 114 | author={Yang, Wei and Luo, Ping and Lin, Liang} 115 | booktitle={Computer Vision and Pattern Recognition (CVPR), 2014 IEEE Conference on}, 116 | year={2013}, 117 | organization={IEEE} 118 | } 119 | ``` 120 | 121 | 122 | :cherries: **HCP 2015** 123 | 124 | (1) **7,700** images in total. 125 | (2) Combined Fashionista (685), CFPD (2,682), Daily Photos dataset (2,500). 126 | (3) Crawl another **1,833** challenging images （*e.g.* sitting or occlusion) annotate pixel-level labels. 127 | (4) **18** categories of labels, *e.g.* face, sunglass, hat, scarf *etc*. 128 | 129 | [[homepage]](http://www.sysu-hcp.net/clothing-co-parsing-by-joint-image-segmentation-and-labeling/) [[pdf]](https://arxiv.org/pdf/1503.02391.pdf) :pig: 130 | ```bib 131 | @article{liang2015deep, 132 | title={Deep human parsing with active template regression}, 133 | author={Liang, Xiaodan and Liu, Si and Shen, Xiaohui and Yang, Jianchao and Liu, Luoqi and Dong, Jian and Lin, Liang and Yan, Shuicheng}, 134 | journal={IEEE transactions on pattern analysis and machine intelligence}, 135 | volume={37}, 136 | number={12}, 137 | pages={2402--2414}, 138 | year={2015}, 139 | publisher={IEEE} 140 | } 141 | ``` 142 | 143 | 144 | :cherries: **Fashion Icon 2015** 145 | 146 | (1) **Video dataset** and **Fashion Icon (FI) image dataset**. 147 | (2) Video dataset contains **1, 500** videos. 148 | (3) FI image dataset contains **1, 082** images, **18** categories. 149 | 150 | [homepage] [[pdf]](https://liusi-group.com/pdf/Fashion%20Parsing%20with%20Video%20Context.pdf) 151 | ```bib 152 | @article{liu2015fashion, 153 | title={Fashion parsing with video context}, 154 | author={Liu, Si and Liang, Xiaodan and Liu, Luoqi and Lu, Ke and Lin, Liang and Cao, Xiaochun and Yan, Shuicheng}, 155 | journal={IEEE Transactions on Multimedia}, 156 | volume={17}, 157 | number={8}, 158 | pages={1347--1358}, 159 | year={2015}, 160 | publisher={IEEE} 161 | } 162 | ``` 163 | 164 | 165 | :cherries: **Refined Fashionista 2017** 166 | 167 | (1) Reduces the number of clothing categories from **56** to **25** essential labels. 168 | (2) Manually annotated all the **685** images in the Fashionista dataset. 169 | 170 | [[homepage]](http://vision.is.tohoku.ac.jp/~kyamagu/research/clothing_parsing/) [[pdf]](https://arxiv.org/pdf/1703.01386.pdf) [[github]](https://github.com/pongsate1/fashion-parsing) :pig: 171 | ```bib 172 | @article{tangseng2017looking, 173 | title={Looking at outfit to parse clothing}, 174 | author={Tangseng, Pongsate and Wu, Zhipeng and Yamaguchi, Kota}, 175 | journal={arXiv preprint arXiv:1703.01386}, 176 | year={2017} 177 | } 178 | ``` 179 | 180 | 181 | :cherries: **FASHION8 2018** 182 | 183 | (1) **9,339** fashion images from **8** continuous years are collected. 184 | (2) With human-annotated foreground masks. 185 | 186 | [homepage] [[pdf]](https://arxiv.org/pdf/1803.03415.pdf) 187 | ```bib 188 | @article{zhang2018fusing, 189 | title={Fusing Hierarchical Convolutional Features for Human Body Segmentation and Clothing Fashion Classification}, 190 | author={Zhang, Zheng and Song, Chengfang and Zou, Qin}, 191 | journal={arXiv preprint arXiv:1803.03415}, 192 | year={2018} 193 | } 194 | ``` 195 | 196 | 197 | :cherries: **ModaNet 2018** 198 | 199 | (1) **55, 176** street images, fully annotated with polygons (bounding box, segmentation mask). 200 | (2) Based on 1 million weakly annotated street images in **Paperdoll**. 201 | (3) **13** categories annotated (*e.g.* bag, belt, boots). 202 | 203 | [[homepage]](https://github.com/eBay/modanet) [[pdf]](https://arxiv.org/pdf/1807.01394.pdf) [[github]](https://github.com/hrsma2i/modanet) :pig: 204 | ```bib 205 | @inproceedings{zheng2018modanet, 206 | title={Modanet: A large-scale street fashion dataset with polygon annotations}, 207 | author={Zheng, Shuai and Yang, Fan and Kiapour, M Hadi and Piramuthu, Robinson}, 208 | booktitle={Proceedings of the 26th ACM international conference on Multimedia}, 209 | pages={1670--1678}, 210 | year={2018} 211 | } 212 | ``` 213 | 214 | 215 | ## 1.  Keypoints 216 | For keypoint detection, landmark detection, pose estimation and *etc*. 217 | ***** 218 | :cherries: **FLD 2016** 219 | 220 | (1) Over **120K** images. 221 | (2) Each image is correctly labeled with **8** fashion landmarks along with their visibility. 222 | (3) Different types of clothing items, including upper/lower/full-body clothes. 223 | (4) Different subsets, including normal/medium/large poses and medium/large scales. 224 | 225 | [[homepage]](http://mmlab.ie.cuhk.edu.hk/projects/DeepFashion/LandmarkDetection.html) [[pdf]](https://arxiv.org/pdf/1608.03049.pdf) [[github]](https://github.com/liuziwei7/fashion-landmarks) :pig: 226 | ```bib 227 | @inproceedings{liu2016fashionlandmark, 228 | author = {Ziwei Liu, Sijie Yan, Ping Luo, Xiaogang Wang, and Xiaoou Tang}, 229 | title = {Fashion Landmark Detection in the Wild}, 230 | booktitle = {European Conference on Computer Vision (ECCV)}, 231 | month = {October}, 232 | year = {2016} 233 | } 234 | ``` 235 | 236 | 237 | :cherries: **FASHIONAI Keypoint 2018** 238 | 239 | (1) **24** key points in **324k** image (including armpit, crotch keypoints). 240 | 241 | [[homepage]](https://tianchi.aliyun.com/competition/entrance/231648/introduction?lang=zh-cn) [[pdf]](http://openaccess.thecvf.com/content_CVPRW_2019/papers/FFSS-USAD/Zou_FashionAI_A_Hierarchical_Dataset_for_Fashion_Understanding_CVPRW_2019_paper.pdf) [[TIANCHI]]() :pig: 242 | ```bib 243 | @inproceedings{zou2019fashionai, 244 | title={FashionAI: A Hierarchical Dataset for Fashion Understanding}, 245 | author={Zou, Xingxing and Kong, Xiangheng and Wong, Waikeung and Wang, Congde and Liu, Yuguang and Cao, Yang}, 246 | booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops}, 247 | pages={0--0}, 248 | year={2019} 249 | } 250 | ``` 251 | :cherries: **DeepFashion2 2019** 252 | 253 | (1) **801K** clothing items annotated with style, scale, viewpoint, occlusion, bounding box, dense landmarks, masks. 254 | (2) **873K** Commercial-Consumer clothes pairs. 255 | (3) **13** different definitions of landmarks and poses for **13** different categories. 256 | 257 | [homepage] [[pdf]](https://arxiv.org/pdf/1901.07973.pdf) [[github]](https://github.com/switchablenorms/DeepFashion2) :pig: 258 | ```bib 259 | @article{DeepFashion2, 260 | author = {Yuying Ge and Ruimao Zhang and Lingyun Wu and Xiaogang Wang and Xiaoou Tang and Ping Luo}, 261 | title={A Versatile Benchmark for Detection, Pose Estimation, Segmentation and Re-Identification of Clothing Images}, 262 | journal={CVPR}, 263 | year={2019} 264 | } 265 | ``` 266 | 267 | 268 | ## 2.  Attribute 269 | For style analysis, attribute recognition, trend anaylsis, style anaylsis, multi-task learning, consumer-to-shop clothes retrieval, in-shop clothes retrieval and *etc*. 270 | ***** 271 | :cherries: **Apparel Style 2012** 272 | 273 | (1) **15** categories (*e.g.* Long dress, Coat, Jacket *etc*). 274 | (2) Over **80, 000** images. 275 | (3) Attributes including colors(13), print(15), material(8), design details(4), styles(4+21), gender(5), sleeve length(3). 276 | 277 | [[homepage]](https://data.vision.ee.ethz.ch/cvl/lbossard/accv12/) [[pdf]](https://data.vision.ee.ethz.ch/cvl/lbossard/accv12/accv12_apparel-classification-with-style.pdf) :pig: 278 | ```bib 279 | @inproceedings{bossard2012apparel, 280 | title={Apparel classification with style}, 281 | author={Bossard, Lukas and Dantone, Matthias and Leistner, Christian and Wengert, Christian and Quack, Till and Van Gool, Luc}, 282 | booktitle={Asian conference on computer vision}, 283 | pages={321--335}, 284 | year={2012}, 285 | organization={Springer} 286 | } 287 | ``` 288 | 289 | 290 | :cherries: **WFC 2013** 291 | 292 | (1) **Women’s Coat** Dataset contains **2,092** images total with manuly annotated labels. 293 | (2) **12** coat/jacket categories: cape, military, motorcycle, peacoat, poncho, puffer, trench, *etc*. 294 | (3) **27** (Material(5), Fastener(4), Fastener style(3), clothing length(3), silhouette(2), pocket(2), neckline(8)) binary attributes. 295 | 296 | [homepage] [[pdf]](https://www.cv-foundation.org/openaccess/content_cvpr_workshops_2013/W03/papers/Di_Style_Finder_Fine-Grained_2013_CVPR_paper.pdf) 297 | ```bib 298 | @inproceedings{di2013style, 299 | title={Style finder: Fine-grained clothing style detection and retrieval}, 300 | author={Di, Wei and Wah, Catherine and Bhardwaj, Anurag and Piramuthu, Robinson and Sundaresan, Neel}, 301 | booktitle={Proceedings of the IEEE Conference on computer vision and pattern recognition workshops}, 302 | pages={8--13}, 303 | year={2013} 304 | } 305 | ``` 306 | 307 | 308 | :cherries: **ZOZOTOWN 2013** 309 | 310 | (1) **12,719** photos on ZOZOTOWN. 311 | (2) Have several meta-data, gender information, wearing items information, and photo date. 312 | 313 | [homepage] [[pdf]](https://www.researchgate.net/profile/Kiyoharu_Aizawa/publication/261438969_SNAPPER_Fashion_coordinate_image_retrieval_system/links/0f31753472d15b5839000000/SNAPPER-Fashion-coordinate-image-retrieval-system.pdf) 314 | ```bib 315 | @inproceedings{miura2013snapper, 316 | title={SNAPPER: fashion coordinate image retrieval system}, 317 | author={Miura, Shinya and Yamasaki, Toshihiko and Aizawa, Kiyoharu}, 318 | booktitle={2013 International Conference on Signal-Image Technology \& Internet-Based Systems}, 319 | pages={784--789}, 320 | year={2013}, 321 | organization={IEEE} 322 | } 323 | ``` 324 | 325 | 326 | :cherries: **Fashion136K 2013** 327 | 328 | (1) **135,893** street fashion images with annotations by fashionistas, brand, demographics. 329 | (2) **3-4** annotations per image. 330 | 331 | [homepage] [[pdf]](https://dl.acm.org/doi/pdf/10.1145/2623330.2623332) 332 | ```bib 333 | @inproceedings{jagadeesh2014large, 334 | title={Large scale visual recommendations from street fashion images}, 335 | author={Jagadeesh, Vignesh and Piramuthu, Robinson and Bhardwaj, Anurag and Di, Wei and Sundaresan, Neel}, 336 | booktitle={Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining}, 337 | pages={1925--1934}, 338 | year={2014} 339 | } 340 | ``` 341 | 342 | 343 | :cherries: **UT-Zap50K 2014** 344 | 345 | (1) **50,000 shoe** images with fine-grained attributes. 346 | (2) **4** relative attributes: “open”, “pointy at the toe”, “sporty”, and “comfortable". 347 | (3) **12,000** total pairs, 3,000 per attribute. 348 | 349 | [[homepage]](http://vision.cs.utexas.edu/projects/finegrained/utzap50k/) [[pdf]](http://aronyu.io/vision/papers/cvpr14/aron-cvpr14.pdf) :pig: 350 | ```bib 351 | @InProceedings{finegrained, 352 | author = {A. Yu and K. Grauman}, 353 | title = {Fine-Grained Visual Comparisons with Local Learning}, 354 | booktitle = {Computer Vision and Pattern Recognition (CVPR)}, 355 | month = {Jun}, 356 | year = {2014} 357 | } 358 | ``` 359 | 360 | 361 | :cherries: **Fashion 10000 2014** 362 | 363 | (1) **32,398** photos, with their associated metadata, distributed in **262** different fashion categories. 364 | 365 | [[homepage]](http://traces.cs.umass.edu/index.php/Mmsys/Mmsys) [[pdf]](https://dl.acm.org/doi/pdf/10.1145/2557642.2563675) :pig: 366 | ```bib 367 | @inproceedings{loni2014fashion, 368 | title={Fashion 10000: an enriched social image dataset for fashion and clothing}, 369 | author={Loni, Babak and Cheung, Lei Yen and Riegler, Michael and Bozzon, Alessandro and Gottlieb, Luke and Larson, Martha}, 370 | booktitle={Proceedings of the 5th ACM Multimedia Systems Conference}, 371 | pages={41--46}, 372 | year={2014} 373 | } 374 | ``` 375 | 376 | 377 | :cherries: **Hipster Wars 2014** 378 | 379 | (1) **1,893** images labeled with **5 style** categories: hipster, bohemian, pinup, preppy, and goth. 380 | 381 | [homepage] [[pdf]](https://projet.liris.cnrs.fr/imagine/pub/proceedings/ECCV-2014/papers/8689/86890472.pdf) 382 | ```bib 383 | @inproceedings{kiapour2014hipster, 384 | title={Hipster wars: Discovering elements of fashion styles}, 385 | author={Kiapour, M Hadi and Yamaguchi, Kota and Berg, Alexander C and Berg, Tamara L}, 386 | booktitle={European conference on computer vision}, 387 | pages={472--488}, 388 | year={2014}, 389 | organization={Springer} 390 | } 391 | ``` 392 | 393 | 394 | :cherries: **Fashion144K 2015** 395 | 396 | (1) **144,169** user posts containing diverse images, textual, and meta information. 397 | (2) Labels like location, comments, votes *etc*. 398 | 399 | [[homepage]](https://esslab.jp/~ess/en/data/fashion144k_stylenet/) [[pdf]](https://www.cs.toronto.edu/~urtasun/publications/simo_etal_cvpr15.pdf) :pig: 400 | ```bib 401 | @InProceedings{SimoSerraCVPR2015, 402 | author = {Edgar Simo-Serra and Sanja Fidler and Francesc Moreno-Noguer and Raquel Urtasun}, 403 | title = {{Neuroaesthetics in Fashion: Modeling the Perception of Fashionability}}, 404 | booktitle = "Proceedings of the Conference on Computer Vision and Pattern Recognition (CVPR)", 405 | year = 2015, 406 | } 407 | ``` 408 | 409 | 410 | :cherries: **WITB(Exact Street2Shop) 2015** 411 | 412 | (1) **404,683** shop photos and **20,357** street photos. 413 | (2) Providing a total of **39,479** clothing item matches. 414 | 415 | [[homepage]](http://www.tamaraberg.com/street2shop/) [[pdf]](http://www.tamaraberg.com/papers/street2shop.pdf) :pig: 416 | ```bib 417 | @inproceedings{hadi2015buy, 418 | title={Where to buy it: Matching street clothing photos in online shops}, 419 | author={Hadi Kiapour, M and Han, Xufeng and Lazebnik, Svetlana and Berg, Alexander C and Berg, Tamara L}, 420 | booktitle={Proceedings of the IEEE international conference on computer vision}, 421 | pages={3343--3351}, 422 | year={2015} 423 | } 424 | ``` 425 | 426 | :cherries: **DARN 2015** 427 | 428 | (1) **453,983** online upper-clothing images with **179** attributes in high-resolution. 429 | (2) Each image contains a single frontal-view person. 430 | (3) Opening(12), Category(20), Color(56), Cloth length(6), Print(27), Silhouette(10), Neckline(25), Sleeve length(7), Sleeve shape(16). 431 | 432 | [homepage] [[pdf]](https://www.cv-foundation.org/openaccess/content_iccv_2015/papers/Huang_Cross-Domain_Image_Retrieval_ICCV_2015_paper.pdf) 433 | ```bib 434 | @inproceedings{huang2015cross, 435 | title={Cross-domain image retrieval with a dual attribute-aware ranking network}, 436 | author={Huang, Junshi and Feris, Rogerio S and Chen, Qiang and Yan, Shuicheng}, 437 | booktitle={Proceedings of the IEEE international conference on computer vision}, 438 | pages={1062--1070}, 439 | year={2015} 440 | } 441 | ``` 442 | 443 | 444 | :cherries: **Clothing1M 2015** 445 | 446 | (1) **1, 000, 000** clothing images with 14 class labels: T-shirt, Shirt, Knitwear, Chiffon, Sweater, Hoodie *etc*. 447 | (2) Each image is automatically assigned with a noisy label according to the keywords in its surrounding text. 448 | (3) Manually refine **72, 409** image labels, which constitute a clean sub-dataset. 449 | 450 | [homepage] [[pdf]](https://www.ee.cuhk.edu.hk/~xgwang/papers/xiaoXYHWcvpr15.pdf) [[github]](https://github.com/Cysu/noisy_label) :pig: 451 | ```bib 452 | @inproceedings{xiao2015learning, 453 | title={Learning from Massive Noisy Labeled Data for Image Classification}, 454 | author={Xiao, Tong and Xia, Tian and Yang, Yi and Huang, Chang and Wang, Xiaogang}, 455 | booktitle={CVPR}, 456 | year={2015} 457 | } 458 | ``` 459 | 460 | 461 | :cherries: **YahooClothing 2015** 462 | 463 | (1) **161,234** fashion images in the Yahoo shopping dataset. 464 | (2) It labeled with category, gender, and sub-category(15) such as Top, Dress, Coat *etc*. 465 | 466 | [homepage] [[pdf]](https://dl.acm.org/doi/pdf/10.1145/2671188.2749318) 467 | ```bib 468 | @inproceedings{lin2015rapid, 469 | title={Rapid clothing retrieval via deep learning of binary codes and hierarchical search}, 470 | author={Lin, Kevin and Yang, Huei-Fang and Liu, Kuan-Hsien and Hsiao, Jen-Hao and Chen, Chu-Song}, 471 | booktitle={Proceedings of the 5th ACM on International Conference on Multimedia Retrieval}, 472 | pages={499--502}, 473 | year={2015} 474 | } 475 | ``` 476 | 477 | 478 | :cherries: **Chitopia 2015** 479 | 480 | (1) Chictopia dataset has **26,8124** usable images, each image has **2** clothing-keywords under **18** categories. 481 | (2) Dress dataset consists of **712** images with total of **58** attributes. 482 | 483 | [homepage] [[pdf]](http://www.bmva.org/bmvc/2015/papers/paper051/paper051.pdf) 484 | ```bib 485 | @inproceedings{yamaguchi2015mix, 486 | title={Mix and Match: Joint Model for Clothing and Attribute Recognition.}, 487 | author={Yamaguchi, Kota and Okatani, Takayuki and Sudo, Kyoko and Murasaki, Kazuhiko and Taniguchi, Yukinobu}, 488 | booktitle={BMVC}, 489 | volume={1}, 490 | number={2}, 491 | pages={4}, 492 | year={2015} 493 | } 494 | ``` 495 | 496 | 497 | :cherries: **Etsy | Wear 2016** 498 | 499 | (1) Etsy dataset has **173,175** clothing products. 500 | (2) Wear dataset has **212,129** images associated shots from different views, list of items, blog text, tags, and other metadata. 501 | 502 | [[homepage]](http://vision.is.tohoku.ac.jp/~kyamagu/research/etsy-dataset/) [[pdf]](https://arxiv.org/pdf/1607.07262.pdf) :pig: 503 | ```bib 504 | @inproceedings{vittayakorn2016automatic, 505 | title={Automatic attribute discovery with neural activations}, 506 | author={Vittayakorn, Sirion and Umeda, Takayuki and Murasaki, Kazuhiko and Sudo, Kyoko and Okatani, Takayuki and Yamaguchi, Kota}, 507 | booktitle={European Conference on Computer Vision}, 508 | pages={252--268}, 509 | year={2016}, 510 | organization={Springer} 511 | } 512 | ``` 513 | 514 | 515 | :cherries: **DeepFashion 2016** 516 | 517 | (1) Over **800,000** images, which are richly annotated with massive attributes, clothing landmarks, and correspondence of images. 518 | (2) **50** fine-grained categories and **1, 000 descriptive attributes**. 519 | (3) Attributes including Category, Print, Material, Silhouette, Part, Style. 520 | 521 | [[homepage]](http://mmlab.ie.cuhk.edu.hk/projects/DeepFashion.html) [[pdf]](https://www.ee.cuhk.edu.hk/~xgwang/papers/liuLQWTcvpr16.pdf) :pig: 522 | ```bib 523 | @inproceedings{liuLQWTcvpr16DeepFashion, 524 | author = {Liu, Ziwei and Luo, Ping and Qiu, Shi and Wang, Xiaogang and Tang, Xiaoou}, 525 | title = {DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations}, 526 | booktitle = {Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}, 527 | month = {June}, 528 | year = {2016} 529 | } 530 | ``` 531 | 532 | 533 | :cherries: **MVC 2016** 534 | 535 | (1) **37,499** items and **161,638** clothing images, **264** attributes(gender, category, sub-catgegory, design attributes). 536 | (2) Multi-view(front, back, right, left) with high-resolution. 537 | 538 | [[homepage]](http://mvc-datasets.github.io/MVC/) [[pdf]](https://www.iis.sinica.edu.tw/papers/song/19692-F.pdf) [[github]](https://github.com/MVC-Datasets/MVC) :pig: 539 | ```bib 540 | @inproceedings{liu2016mvc, 541 | title={Mvc: A dataset for view-invariant clothing retrieval and attribute prediction}, 542 | author={Liu, Kuan-Hsien and Chen, Ting-Yen and Chen, Chu-Song}, 543 | booktitle={Proceedings of the 2016 ACM on International Conference on Multimedia Retrieval}, 544 | pages={313--316}, 545 | year={2016} 546 | } 547 | ``` 548 | 549 | 550 | :cherries: **StreetStyle27K 2017** 551 | 552 | (1) **27K** images, each with **12** clothing attributes. 553 | (2) A first-of-its-kind analysis of global and per-city fashion choices and trends. 554 | (3) 7 binary attributes(Wearing Jacket, Wearing Scarf *etc*), 13 colors, 7 categories, 3 sleeve length, 3 neckline, 6 print. 555 | 556 | [[homepage]](http://streetstyle.cs.cornell.edu/#dataset) [[pdf]](https://arxiv.org/pdf/1706.01869.pdf) :pig: 557 | ```bib 558 | @article{StreetStyle2017, 559 | title={{StreetStyle}: {E}xploring world-wide clothing styles from millions of photos}, 560 | author={Kevin Matzen and Kavita Bala and Noah Snavely}, 561 | journal={arXiv preprint arXiv:1706.01869}, 562 | year={2017} 563 | } 564 | ``` 565 | 566 | 567 | :cherries: **Fashionstyle14 2017** 568 | 569 | (1) **13,126** images labeled with **14** different fashion styles (conservative, dressy, ethnic, fairy, feminine *etc*). 570 | 571 | [[homepage]](https://esslab.jp/~ess/en/data/fashionstyle14/) [[pdf]](http://openaccess.thecvf.com/content_ICCV_2017_workshops/papers/w32/Takagi_What_Makes_a_ICCV_2017_paper.pdf) :pig: 572 | ```bib 573 | @inproceedings{takagi2017makes, 574 | title={What makes a style: Experimental analysis of fashion prediction}, 575 | author={Takagi, Moeko and Simo-Serra, Edgar and Iizuka, Satoshi and Ishikawa, Hiroshi}, 576 | booktitle={Proceedings of the IEEE International Conference on Computer Vision Workshops}, 577 | pages={2247--2253}, 578 | year={2017} 579 | } 580 | ``` 581 | 582 | 583 | :cherries: **Fashion200K 2017** 584 | 585 | (1) Over **200,000** images of five categories (dress, top, pants, skirt, and jacket) and their product descriptions. 586 | (2) After dealing with product description, **4,404** attributes(words) have remained. 587 | 588 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_ICCV_2017/papers/Han_Automatic_Spatially-Aware_Fashion_ICCV_2017_paper.pdf) [[github]](https://github.com/xthan/fashion-200k/) :pig: 589 | ```bib 590 | @inproceedings{han2017automatic, 591 | title = {Automatic Spatially-aware Fashion Concept Discovery}, 592 | author = {Han, Xintong and Wu, Zuxuan and Huang, Phoenix X. and Zhang, Xiao and Zhu, Menglong and Li, Yuan and Zhao, Yang and Davis, Larry S.}, 593 | booktitle = {ICCV}, 594 | year = {2017}, 595 | } 596 | ``` 597 | 598 | 599 | :cherries: **Fashion550K 2017** 600 | 601 | (1) **weakly-labeled** image dataset consists of **550,661** images that includes **5,300** human-annotated images. 602 | (2) **66** binary labels, 26 colors, 22 categories, 7 shoes, 11 accessories. 603 | 604 | [[homepage]](https://esslab.jp/~ess/en/data/fashion550k/) [[pdf]](http://openaccess.thecvf.com/content_ICCV_2017_workshops/papers/w32/Inoue_Multi-Label_Fashion_Image_ICCV_2017_paper.pdf) :pig: 605 | ```bib 606 | @InProceedings{InoueICCVW2017, 607 | author = {Naoto Inoue and Edgar Simo-Serra and Toshihiko Yamasaki and Hiroshi Ishikawa}, 608 | title = {{Multi-Label Fashion Image Classification with Minimal Human Supervision}}, 609 | booktitle = "Proceedings of the International Conference on Computer Vision Workshops (ICCVW)", 610 | year = 2017, 611 | } 612 | ``` 613 | 614 | 615 | :cherries: **Amazon Dress** 616 | 617 | (1) **53 689** images of dresses and their product descriptions. 618 | (2) Different categories, such as bridesmaid, casual, mother of the bride, night out and cocktail, and wedding. 619 | (3) The product descriptions consist of the surrounding natural language text, like the title, features, and editorial content. 620 | 621 | [homepage] [[pdf]](https://kddfashion2017.mybluemix.net/final_submissions/ML4Fashion_paper_7.pdf) 622 | ```bib 623 | @inproceedings{laenen2017cross, 624 | title={Cross-modal search for fashion attributes}, 625 | author={Laenen, Katrien and Zoghbi, Susana and Moens, Marie-Francine}, 626 | booktitle={Proceedings of the KDD 2017 Workshop on Machine Learning Meets Fashion}, 627 | volume={2017}, 628 | pages={1--10}, 629 | year={2017}, 630 | organization={ACM} 631 | } 632 | ``` 633 | 634 | 635 | :cherries: **SFS 2017** 636 | 637 | (1) **293,105** posts, each image consists of weakly-labeled multi-task ground-truth. 638 | (2) Labels including season, occasion, fashion style, garment categories, geographical and year information. 639 | 640 | [[homepage]](https://zenodo.org/record/833051#.XqfGivmWa70) [[pdf]](http://doi.org/10.1145/3123266.3123441) :pig: 641 | ```bib 642 | @inproceedings{gu2017understanding, 643 | title={Understanding fashion trends from street photos via neighbor-constrained embedding learning}, 644 | author={Gu, Xiaoling and Wong, Yongkang and Peng, Pai and Shou, Lidan and Chen, Gang and Kankanhalli, Mohan S}, 645 | booktitle={Proceedings of the 25th ACM international conference on Multimedia}, 646 | pages={190--198}, 647 | year={2017} 648 | } 649 | ``` 650 | 651 | 652 | :cherries: **Video2Shop 2017** 653 | 654 | (1) **26,352** clothing trajectories extracted from 526 videos and **85,677** clothing shopping images. 655 | (2) **14** categories of clothes are manually labeled. 656 | 657 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_cvpr_2017/papers/Cheng_Video2Shop_Exact_Matching_CVPR_2017_paper.pdf) 658 | ```bib 659 | @inproceedings{cheng2017video2shop, 660 | title={Video2shop: Exact matching clothes in videos to online shopping images}, 661 | author={Cheng, Zhi-Qi and Wu, Xiao and Liu, Yang and Hua, Xian-Sheng}, 662 | booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition}, 663 | pages={4048--4056}, 664 | year={2017} 665 | } 666 | ``` 667 | 668 | 669 | :cherries: **RFS | PFS 2018** 670 | 671 | 672 | 673 | [homepage] [[pdf]] 674 | ```bib 675 | @article{gu2018multi, 676 | title={Multi-Modal and Multi-Domain Embedding Learning for Fashion Retrieval and Analysis}, 677 | author={Gu, Xiaoling and Wong, Yongkang and Shou, Lidan and Peng, Pai and Chen, Gang and Kankanhalli, Mohan S}, 678 | journal={IEEE Transactions on Multimedia}, 679 | volume={21}, 680 | number={6}, 681 | pages={1524--1537}, 682 | year={2018}, 683 | publisher={IEEE} 684 | } 685 | ``` 686 | 687 | 688 | :cherries: **BrandFashion 2018** 689 | 690 | (1) **10K** clothing images with distinctive logos from **15** brands. 691 | (2) Label with **16** clothing categories and **32** semantic attributes. 692 | 693 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_ECCVW_2018/papers/11131/Manandhar_Tiered_Deep_Similarity_Search_for_Fashion_ECCVW_2018_paper.pdf) 694 | ```bib 695 | @inproceedings{manandhar2018tiered, 696 | title={Tiered Deep Similarity Search for Fashion}, 697 | author={Manandhar, Dipu and Bastan, Muhammet and Yap, Kim-Hui}, 698 | booktitle={Proceedings of the European Conference on Computer Vision (ECCV)}, 699 | pages={0--0}, 700 | year={2018} 701 | } 702 | ``` 703 | 704 | :cherries: **FashionBrand 2018** 705 | 706 | (1) **3,828,735** clothing product images from **1,219** brands. 707 | (2) Attributes including 5 categories, 13 sub-categories. 708 | 709 | [homepage] [[pdf]](https://arxiv.org/pdf/1810.09941v1.pdf) 710 | ```bib 711 | @inproceedings{hadi2018brand, 712 | title={Brand> Logo: Visual Analysis of Fashion Brands}, 713 | author={Hadi Kiapour, M and Piramuthu, Robinson}, 714 | booktitle={Proceedings of the European Conference on Computer Vision (ECCV)}, 715 | pages={0--0}, 716 | year={2018} 717 | } 718 | ``` 719 | :cherries: **Fashion60 2018** 720 | 721 | (1) **539,704** images for training and **30, 000** images for testing. 722 | (2) **60** fine-grained fashion attributes categorized into 5 coarse-grained groups. 723 | 724 | [homepage] [[pdf]](https://ieeexplore.ieee.org/document/8396968) 725 | ```bib 726 | @inproceedings{kuang2018ontology, 727 | title={Ontology-driven hierarchical deep learning for fashion recognition}, 728 | author={Kuang, Zhenzhong and Yu, Jun and Yu, Zhou and Fan, Jianping}, 729 | booktitle={2018 IEEE Conference on Multimedia Information Processing and Retrieval (MIPR)}, 730 | pages={19--24}, 731 | year={2018}, 732 | organization={IEEE} 733 | } 734 | ``` 735 | :cherries: **X-domain 2018** 736 | 737 | (1) **245,467** shop images. Each image is annotated by **9** attribute labels. 738 | (2) Imbalanced with an **imbalance-ratio** of 1:4,162 (20 : 204,177). 739 | (3) **178** distinctive attributes over the 9 labels, including 6 sleeve-length, 55 colors. 740 | 741 | [homepage] [[pdf]](https://arxiv.org/pdf/1804.10851.pdf) 742 | ```bib 743 | @article{dong2018imbalanced, 744 | title={Imbalanced deep learning by minority class incremental rectification}, 745 | author={Dong, Qi and Gong, Shaogang and Zhu, Xiatian}, 746 | journal={IEEE transactions on pattern analysis and machine intelligence}, 747 | volume={41}, 748 | number={6}, 749 | pages={1367--1381}, 750 | year={2018}, 751 | publisher={IEEE} 752 | } 753 | ``` 754 | 755 | 756 | :cherries: **Women | Men Video 2018** 757 | 758 | (1) Video dataset **for unsupervised learning**. 759 | (2) Two new clothing image datasets are annotated with **10** pre-defined clothing attributes. 760 | (3) **18,737** woman clothing videos and **21,224** man clothing videos. 761 | 762 | [homepage] [[pdf]](https://www.researchgate.net/profile/Sanyi_Zhang/publication/321785076_Watch_Fashion_Shows_to_Tell_Clothing_Attributes/links/5cdf0e3c92851c4eabaa3e07/Watch-Fashion-Shows-to-Tell-Clothing-Attributes.pdf) 763 | ```bib 764 | @article{zhang2018watch, 765 | title={Watch fashion shows to tell clothing attributes}, 766 | author={Zhang, Sanyi and Liu, Si and Cao, Xiaochun and Song, Zhanjie and Zhou, Jie}, 767 | journal={Neurocomputing}, 768 | volume={282}, 769 | pages={98--110}, 770 | year={2018}, 771 | publisher={Elsevier} 772 | } 773 | ``` 774 | :cherries: **FASHIONAI Attributes 2018** 775 | 776 | (1) **324k** images with **245** labels that cover 6 categories of women’s clothing, and a total of 41 subcategories (single labeled). 777 | [[homepage]](https://tianchi.aliyun.com/competition/entrance/231649/information) [[pdf]](http://openaccess.thecvf.com/content_CVPRW_2019/papers/FFSS-USAD/Zou_FashionAI_A_Hierarchical_Dataset_for_Fashion_Understanding_CVPRW_2019_paper.pdf) [[TIANCHI]]() :pig: 778 | ```bib 779 | @inproceedings{zou2019fashionai, 780 | title={FashionAI: A Hierarchical Dataset for Fashion Understanding}, 781 | author={Zou, Xingxing and Kong, Xiangheng and Wong, Waikeung and Wang, Congde and Liu, Yuguang and Cao, Yang}, 782 | booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops}, 783 | pages={0--0}, 784 | year={2019} 785 | } 786 | ``` 787 | 788 | 789 | :cherries: **Feidegger 2018** 790 | 791 | (1) **8,700** fashion items, each with a high-resolution image and **5** independently collected textual descriptions. 792 | (2) Restrict the domain to images of dresses, and German-language visual descriptions. 793 | 794 | [[homepage]](https://research.zalando.com/welcome/mission/research-projects/feidegger-dataset/) [[pdf]](http://www.lrec-conf.org/proceedings/lrec2018/pdf/319.pdf) 795 | ```bib 796 | @inproceedings{lefakis2018feidegger, 797 | title={FEIDEGGER: A Multi-modal Corpus of Fashion Images and Descriptions in German}, 798 | author={Lefakis, Leonidas and Akbik, Alan and Vollgraf, Roland}, 799 | booktitle={Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)}, 800 | year={2018} 801 | } 802 | ``` 803 | 804 | 805 | :cherries: **Studio2Shop 2018** 806 | 807 | (1) **1.15 million** query images with neutral backgrounds. 808 | (2) Category(7), Color(82), Print(19), Cloth length (12), sleeve length (9), shirt collar(27), Neckline (12), Material (14), Trouser rise (3). 809 | 810 | [homepage] [[pdf]](https://www.scitepress.org/Papers/2018/65445/65445.pdf) 811 | ```bib 812 | @article{lasserre2018studio2shop, 813 | title={Studio2shop: from studio photo shoots to fashion articles}, 814 | author={Lasserre, Julia and Rasch, Katharina and Vollgraf, Roland}, 815 | journal={arXiv preprint arXiv:1807.00556}, 816 | year={2018} 817 | } 818 | ``` 819 | 820 | 821 | :cherries: **Shopping 100k 2018** 822 | 823 | (1) **101,021** images that consist of pure clothing items. 824 | (2) Category(16), Neckline(17+11), Color(19), Material(14), Opening(9), Silhouette(15), Gender(2), Cloth length(7), Print(15), Pocket(7), Sleeve length(9), Sport(15). 825 | 826 | [homepage] [[pdf]](https://www.researchgate.net/profile/Kenan_Ak/publication/324728522_Efficient_Multi-Attribute_Similarity_Learning_Towards_Attribute-based_Fashion_Search/links/5adf5078aca272fdaf89c65a/Efficient-Multi-Attribute-Similarity-Learning-Towards-Attribute-based-Fashion-Search.pdf) 827 | ```bib 828 | @inproceedings{ak2018efficient, 829 | title={Efficient multi-attribute similarity learning towards attribute-based fashion search}, 830 | author={Ak, Kenan E and Lim, Joo Hwee and Tham, Jo Yew and Kassim, Ashraf A}, 831 | booktitle={2018 IEEE Winter Conference on Applications of Computer Vision (WACV)}, 832 | pages={1671--1679}, 833 | year={2018}, 834 | organization={IEEE} 835 | } 836 | ``` 837 | 838 | 839 | :cherries: **Footwear 2018** 840 | 841 | (1) **1,000** object, 12 category footwear dataset, each object captured from 4 different poses. 842 | 843 | [[homepage]](https://www.cse.iitd.ac.in/~chetan/projects.html) [[pdf]](https://www.cse.iitd.ac.in/~chetan/papers/icip18-fgvc.pdf) 844 | ```bib 845 | @inproceedings{mahajan2018pose, 846 | title={Pose Aware Fine-Grained Visual Classification Using Pose Experts}, 847 | author={Mahajan, Kushagra and Khurana, Tarasha and Chopra, Ayush and Gupta, Isha and Arora, Chetan and Rai, Atul}, 848 | booktitle={2018 25th IEEE International Conference on Image Processing (ICIP)}, 849 | pages={2381--2385}, 850 | year={2018}, 851 | organization={IEEE} 852 | } 853 | ``` 854 | :cherries: **iMaterialist 2019** 855 | 856 | (1) **1M+** fashion images with **228** fine-grained attributes in total (multi-labeled). 857 | (2) Category(105), Color(21), Gender(3), Material(34), Neckline(11), Print(28), Sleeve length(5), Design details(21). 858 | 859 | [[homepage]](https://www.kaggle.com/c/imaterialist-fashion-2019-FGVC6) [[pdf]](http://openaccess.thecvf.com/content_ICCVW_2019/papers/CVFAD/Guo_The_iMaterialist_Fashion_Attribute_Dataset_ICCVW_2019_paper.pdf) [[github]](https://github.com/visipedia/imat_fashion_comp) :pig: 860 | ```bib 861 | @inproceedings{guo2019imaterialist, 862 | title={The imaterialist fashion attribute dataset}, 863 | author={Guo, Sheng and Huang, Weilin and Zhang, Xiao and Srikhanta, Prasanna and Cui, Yin and Li, Yuan and Adam, Hartwig and Scott, Matthew R and Belongie, Serge}, 864 | booktitle={Proceedings of the IEEE International Conference on Computer Vision Workshops}, 865 | pages={0--0}, 866 | year={2019} 867 | } 868 | ``` 869 | 870 | 871 | :cherries: **Atlas 2019** 872 | 873 | (1) **186,150** images under clothing category with **3** levels and **52** leaf nodes in the taxonomy. 874 | 875 | [[homepage]](https://github.com/vumaasha/atlas) [[pdf]](https://arxiv.org/pdf/1908.08984.pdf) :pig: 876 | ```bib 877 | @article{umaashankar2019atlas, 878 | title={Atlas: A Dataset and Benchmark for E-commerce Clothing Product Categorization}, 879 | author={Umaashankar, Venkatesh and Prakash, Aditi and others}, 880 | journal={arXiv preprint arXiv:1908.08984}, 881 | year={2019} 882 | } 883 | ``` 884 | 885 | 886 | :cherries: **DeepShoe 2019** 887 | 888 | (1) **14,314** and **31,048** images from the street and online shop scenario in multiple viewpoints. 889 | (2) Attributes including 22 Colors, 4 Toe shape, 7 Heel shape. 890 | 891 | [homepage] [[pdf]](http://alumni.media.mit.edu/~shiboxin/files/Zhan_CVIU19.pdf) 892 | ```bib 893 | @article{zhan2019deepshoe, 894 | title={DeepShoe: An improved Multi-Task View-invariant CNN for street-to-shop shoe retrieval}, 895 | author={Zhan, Huijing and Shi, Boxin and Duan, Ling-Yu and Kot, Alex C}, 896 | journal={Computer Vision and Image Understanding}, 897 | volume={180}, 898 | pages={23--33}, 899 | year={2019}, 900 | publisher={Elsevier} 901 | } 902 | ``` 903 | 904 | 905 | :cherries: **FindFashion 2019 combine DeepFashion & Street2Shop** 906 | 907 | (1) Customer-to-shop clothes retrieval dataset consists of **565,041** images and **382,230** pairs. 908 | (2) Labeled **3** attributes (i.e., occlusions, views, and cropping). 909 | (3) Divided the benchmark into 4 levels. i.e., Easy, Hard-Cropping, Hard-Occlusion, and Hard-View. 910 | 911 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_ICCV_2019/papers/Kuang_Fashion_Retrieval_via_Graph_Reasoning_Networks_on_a_Similarity_Pyramid_ICCV_2019_paper.pdf) 912 | ```bib 913 | @inproceedings{kuang2019fashion, 914 | title={Fashion Retrieval via Graph Reasoning Networks on a Similarity Pyramid}, 915 | author={Kuang, Zhanghui and Gao, Yiming and Li, Guanbin and Luo, Ping and Chen, Yimin and Lin, Liang and Zhang, Wayne}, 916 | booktitle={Proceedings of the IEEE International Conference on Computer Vision}, 917 | pages={3066--3075}, 918 | year={2019} 919 | } 920 | ``` 921 | 922 | 923 | :cherries: **GarmentSet 2020** 924 | 925 | (1) **9,636** images with collar part annotations and **8,616** images with shoulder and sleeve annotations. 926 | (2) With annotation of landmarks of collars and sleeves on clean garment images. 927 | 928 | [homepage] [[pdf]](https://arxiv.org/pdf/2001.06427.pdf) 929 | ```bib 930 | @inproceedings{chen2020tailorgan, 931 | title={TailorGAN: Making User-Defined Fashion Designs}, 932 | author={Chen, Lele and Tian, Justin and Li, Guo and Wu, Cheng-Haw and King, Erh-Kan and Chen, Kuan-Ting and Hsieh, Shao-Hang and Xu, Chenliang}, 933 | booktitle={The IEEE Winter Conference on Applications of Computer Vision}, 934 | pages={3241--3250}, 935 | year={2020} 936 | } 937 | ``` 938 | 939 | 940 | 941 | ## 3.  Outfit 942 | For outfit generation, recommendation, evaluation, comnpatibility learning *etc*. 943 | ****** 944 | 945 | :cherries: **WoW 2012** 946 | 947 | (1) **24,417** clothing images that are fully annotated. 948 | (2) **7** multi-value clothing attributes and **10** occasion categories. 949 | (3) **9,469** images with visible full-body. **8,421** images with only upper-body. **6,527** images with lower-body clothing. 950 | (4) Color(11), Material(6), Print(6), Neckline(6), Sleeve length(3), Bottom length(3). 951 | (5) Labeled with **10** common occasions. 952 | 953 | [homepage] [[pdf]](https://dl.acm.org/doi/pdf/10.1145/2393347.2393433) 954 | ```bib 955 | @inproceedings{liu2012hi, 956 | title={Hi, magic closet, tell me what to wear!}, 957 | author={Liu, Si and Feng, Jiashi and Song, Zheng and Zhang, Tianzhu and Lu, Hanqing and Xu, Changsheng and Yan, Shuicheng}, 958 | booktitle={Proceedings of the 20th ACM international conference on Multimedia}, 959 | pages={619--628}, 960 | year={2012} 961 | } 962 | ``` 963 | 964 | 965 | :cherries: **Stylatrix** 966 | 967 | [homepage] [[pdf]] 968 | ```bib 969 | @article{sunstylatrix, 970 | title={Stylatrix: an interactive model-based system for fashion exploration and outfit discovery}, 971 | author={Sun, Will J and Gajos, Krzysztof Z} 972 | } 973 | ``` 974 | 975 | 976 | :cherries: **Edge2Garment 2016** 977 | 978 | [[homepage]](https://phillipi.github.io/pix2pix/) [[pdf]](https://arxiv.org/pdf/1810.09941v1.pdf) [[github]](https://github.com/phillipi/pix2pix) [[datalink]](https://people.eecs.berkeley.edu/~tinghuiz/projects/pix2pix/datasets/) :pig: 979 | ```bib 980 | @article{pix2pix2016, 981 | title={Image-to-Image Translation with Conditional Adversarial Networks}, 982 | author={Isola, Phillip and Zhu, Jun-Yan and Zhou, Tinghui and Efros, Alexei A}, 983 | journal={arxiv}, 984 | year={2016} 985 | } 986 | ``` 987 | 988 | 989 | :cherries: **FashionVC 2017** 990 | 991 | (1) **20,726** outfits with **14,871** tops and **13,663** bottoms. 992 | (2) Visual image, categories and title description are collected. 993 | 994 | [homepage] [[pdf]](https://dl.acm.org/doi/pdf/10.1145/3123266.3123314) 995 | ```bib 996 | @inproceedings{song2017neurostylist, 997 | title={Neurostylist: Neural compatibility modeling for clothing matching}, 998 | author={Song, Xuemeng and Feng, Fuli and Liu, Jinhuan and Li, Zekun and Nie, Liqiang and Ma, Jun}, 999 | booktitle={Proceedings of the 25th ACM international conference on Multimedia}, 1000 | pages={753--761}, 1001 | year={2017} 1002 | } 1003 | ``` 1004 | 1005 | 1006 | :cherries: **Fashion409K 2017** 1007 | 1008 | (1) **409,776** sets of clothing items from **644,192** unique items. 1009 | 1010 | [[homepage]](http://vision.is.tohoku.ac.jp/~tangseng/smart_closet_project.html) [[pdf]](http://openaccess.thecvf.com/content_ICCV_2017_workshops/papers/w32/Tangseng_Recommending_Outfits_From_ICCV_2017_paper.pdf) :pig: 1011 | ```bib 1012 | @InProceedings{Tangseng_2017_ICCV, 1013 | author = {Tangseng, Pongsate and Yamaguchi, Kota and Okatani, Takayuki}, 1014 | title = {Recommending Outfits From Personal Closet}, 1015 | booktitle = {The IEEE International Conference on Computer Vision (ICCV) Workshops}, 1016 | month = {Oct}, 1017 | year = {2017} 1018 | } 1019 | ``` 1020 | 1021 | 1022 | :cherries: **Polyvore 2017** 1023 | 1024 | (1) **21,889** outfits and **164,379** items. 1025 | (2) Keep the first **8** for simplicity, the average number of fashion items in an outfit is **6.5**. 1026 | (3) To clean the text descriptions, words appearing fewer than 30 times are deleted, leading to a vocabulary of size **2,757**. 1027 | 1028 | [homepage] [[pdf]](https://arxiv.org/pdf/1707.05691.pdf) [[github]](https://github.com/xthan/polyvore) :pig: 1029 | ```bib 1030 | @inproceedings{han2017learning, 1031 | author = {Han, Xintong and Wu, Zuxuan and Jiang, Yu-Gang and Davis, Larry S}, 1032 | title = {Learning Fashion Compatibility with Bidirectional LSTMs}, 1033 | booktitle = {ACM Multimedia}, 1034 | year = {2017}, 1035 | } 1036 | ``` 1037 | 1038 | 1039 | :cherries: **AVA 2017** 1040 | 1041 | (1) Photo.net dataset contains **20,278** images with at least 10 score ratings per image. 1042 | (2) CUHK-PhotoQuality (CUHK-PQ) dataset t contains **17,690** images. All images are given a binary aesthetic label. 1043 | (3) Aesthetic Visual Analysis (AVA) dataset contains **250k** images. Each image receives 78 ∼ 549 votes of score ranging from 1 to 10. 1044 | 1045 | [[homepage]](http://personal.ie.cuhk.edu.hk/~dy015/ImageAesthetics/Image_Aesthetic_Assessment.html) [[pdf]](https://arxiv.org/pdf/1610.00838.pdf) :pig: 1046 | ```bib 1047 | @article{deng2017image, 1048 | author = {Deng, Yubin and Loy, Chen Change and Tang, Xiaoou}, 1049 | title = {Image Aesthetic Assessment: An Experimental Survey}, 1050 | journal={IEEE Signal Processing Magazine}, 1051 | volume={34}, 1052 | number={4}, 1053 | pages={80--106}, 1054 | year={2017}, 1055 | publisher={IEEE} 1056 | } 1057 | ``` 1058 | 1059 | 1060 | :cherries: **UIUC 2018** 1061 | 1062 | (1) **68,306** outfits and **365,054** items. 1063 | (2) **19** max items, which has semantic category labeled. 1064 | 1065 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_ECCV_2018/papers/Mariya_Vasileva_Learning_Type-Aware_Embeddings_ECCV_2018_paper.pdf) [[github]](https://github.com/mvasil/fashion-compatibility) 1066 | ```bib 1067 | @inproceedings{VasilevaECCV18FasionCompatibility, 1068 | Author = {Mariya I. Vasileva and Bryan A. Plummer and Krishna Dusad and Shreya Rajpal and Ranjitha Kumar and David Forsyth}, 1069 | Title = {Learning Type-Aware Embeddings for Fashion Compatibility}, 1070 | booktitle = {ECCV}, 1071 | Year = {2018} 1072 | } 1073 | ``` 1074 | 1075 | 1076 | :cherries: **IQON 2018** 1077 | 1078 | (1) **164,837** items of clothing grouped in **21,889** outfits. 1079 | 1080 | [homepage] [[pdf]](https://arxiv.org/pdf/1807.03133.pdf) [[github]](https://github.com/Cherrybruin/iqon-dataset) 1081 | ```bib 1082 | @article{nakamura2018outfit, 1083 | title={Outfit generation and style extraction via bidirectional lstm and autoencoder}, 1084 | author={Nakamura, Takuma and Goto, Ryosuke}, 1085 | journal={arXiv preprint arXiv:1807.03133}, 1086 | year={2018} 1087 | } 1088 | ``` 1089 | 1090 | 1091 | :cherries: **Style4BodyShape 2018** 1092 | 1093 | (1) **3,150** female celebrities annotated with the corresponding types of body shapes. 1094 | (2) **349,298** images of **270** stylish celebrities annotated with the types of clothing items. 1095 | 1096 | [homepage] [[pdf]](https://dl.acm.org/doi/pdf/10.1145/3240508.3240546) 1097 | ```bib 1098 | @inproceedings{hidayati2018dress, 1099 | title={What dress fits me best? fashion recommendation on the clothing style for personal body shape}, 1100 | author={Hidayati, Shintami Chusnul and Hsu, Cheng-Chun and Chang, Yu-Ting and Hua, Kai-Lung and Fu, Jianlong and Cheng, Wen-Huang}, 1101 | booktitle={Proceedings of the 26th ACM international conference on Multimedia}, 1102 | pages={438--446}, 1103 | year={2018} 1104 | } 1105 | ``` 1106 | 1107 | 1108 | :cherries: **Lookastic 2019** 1109 | 1110 | (1) **30,790** fashionable outfits from the Lookastic. 1111 | (2) **124,665** matched pairs for men with **5,069** items. 1112 | (3) **158,755** matched pairs for women with **10,016** items. 1113 | (4) 65 Colors, 38 Materials, 40 Print, 253 fine-grained Categories, 11 Styles, and 114 Sub-categories. 1114 | 1115 | [homepage] [[pdf]](http://staff.ustc.edu.cn/~hexn/papers/sigir19-fashion.pdf) 1116 | ```bib 1117 | @inproceedings{yang2019interpretable, 1118 | title={Interpretable Fashion Matching with Rich Attributes}, 1119 | author={Yang, Xun and He, Xiangnan and Wang, Xiang and Ma, Yunshan and Feng, Fuli and Wang, Meng and Chua, Tat-Seng}, 1120 | booktitle={Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval}, 1121 | pages={775--784}, 1122 | year={2019} 1123 | } 1124 | ``` 1125 | 1126 | 1127 | :cherries: **POG 2019** 1128 | 1129 | (1) **1.01 million** outfits, **583K** fashion items, with context information. 1130 | (2) **0.28 billion** user click actions from **3.57 million** users. 1131 | 1132 | [homepage] [[pdf]](https://arxiv.org/pdf/1905.01866.pdf) [[github]](https://github.com/wenyuer/POG) :pig: 1133 | ```bib 1134 | @inproceedings{chen2019pog, 1135 | title={POG: Personalized Outfit Generation for Fashion Recommendation at Alibaba iFashion}, 1136 | author={Chen, Wen and Huang, Pipei and Xu, Jiaming and Guo, Xin and Guo, Cheng and Sun, Fei and Li, Chao and Pfadler, Andreas and Zhao, Huan and Zhao, Binqiang}, 1137 | booktitle={Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery \& Data Mining}, 1138 | pages={2662--2670}, 1139 | year={2019} 1140 | } 1141 | ``` 1142 | 1143 | 1144 | :cherries: **Shop the Look 2019** 1145 | 1146 | (1) Based on Shop the Look (STL) task, and covert STL data into a format that can be used for our Complete the Look (CTL) task. 1147 | 1148 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_CVPR_2019/papers/Kang_Complete_the_Look_Scene-Based_Complementary_Product_Recommendation_CVPR_2019_paper.pdf)[[github]](https://github.com/kang205/STL-Dataset) 1149 | ```bib 1150 | @inproceedings{kang2019complete, 1151 | title={Complete the Look: Scene-based Complementary Product Recommendation}, 1152 | author={Kang, Wang-Cheng and Kim, Eric and Leskovec, Jure and Rosenberg, Charles and McAuley, Julian}, 1153 | booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition}, 1154 | pages={10532--10541}, 1155 | year={2019} 1156 | } 1157 | ``` 1158 | 1159 | 1160 | :cherries: **ExpFashion 2019** 1161 | 1162 | (1) FashionVC+, consisting of both the visual and textual metadata of fashion items (i.e.,the tops, bottoms and shoes) on Polyvore. 1163 | (2) **20,726** outfits with **14,870** tops, **13,662** bottoms and **14,093** pairs of shoes, respectively. 1164 | (3) ExpFashion dataset consists of **893,991** outfits with **168,682** tops and **117,668** bottom. 1165 | 1166 | [homepage] [[pdf]](https://xuemengsong.github.io/Neuro_2019.pdf) 1167 | ```bib 1168 | @article{liu2019neural, 1169 | title={Neural fashion experts: I know how to make the complementary clothing matching}, 1170 | author={Liu, Jinhuan and Song, Xuemeng and Chen, Zhumin and Ma, Jun}, 1171 | journal={Neurocomputing}, 1172 | volume={359}, 1173 | pages={249--263}, 1174 | year={2019}, 1175 | publisher={Elsevier} 1176 | } 1177 | ``` 1178 | 1179 | 1180 | :cherries: **ASOS outfits 2019** 1181 | 1182 | (1) **586,320** fashion outfits (images and textual descriptions) composed by ASOS stylists. 1183 | (2) Each containing between 2 and 5 items. 1184 | (3) **591,725** unique items representing **18** different womenswear product types and **22** different menswear product type. 1185 | 1186 | [homepage] [[pdf]](https://arxiv.org/pdf/1904.00741.pdf) 1187 | ```bib 1188 | @article{bettaney2019fashion, 1189 | title={Fashion Outfit Generation for E-commerce}, 1190 | author={Bettaney, Elaine M and Hardwick, Stephen R and Zisimopoulos, Odysseas and Chamberlain, Benjamin Paul}, 1191 | journal={arXiv preprint arXiv:1904.00741}, 1192 | year={2019} 1193 | } 1194 | ``` 1195 | 1196 | 1197 | :cherries: **Chuanda 2020** 1198 | 1199 | (1) **3,557** outfits covering **67** basic fashion styles. 1200 | (2) Labeled with **1,879** distinct fashion related attributes that belong to 5 types: Gender, Season, Style, Material, and Function. 1201 | 1202 | [homepage] [[pdf]](https://arxiv.org/pdf/2004.06229.pdf) 1203 | ```bib 1204 | @article{liu2020imitation, 1205 | title={Imitation Learning for Fashion Style Based on Hierarchical Multimodal Representation}, 1206 | author={Liu, Shizhu and Yang, Shanglin and Zhou, Hui}, 1207 | journal={arXiv preprint arXiv:2004.06229}, 1208 | year={2020} 1209 | } 1210 | ``` 1211 | 1212 | 1213 | ## 4.  Generation 1214 | For 3D generation, VTON *etc*. 1215 | ******* 1216 | 1217 | :cherries: **MPI Dynamic FAUST | BUFF 2017** 1218 | 1219 | (1) **40,000** raw and aligned meshes. 1220 | 1221 | [[homepage]](http://buff.is.tue.mpg.de/) [[pdf]](http://openaccess.thecvf.com/content_cvpr_2017/papers/Bogo_Dynamic_FAUST_Registering_CVPR_2017_paper.pdf) :pig: 1222 | ```bib 1223 | @inproceedings{dfaust:CVPR:2017, 1224 | title = {Dynamic {FAUST}: {R}egistering Human Bodies in Motion}, 1225 | author = {Bogo, Federica and Romero, Javier and Pons-Moll, Gerard and Black, Michael J.}, 1226 | booktitle = {IEEE Conf. on Computer Vision and Pattern Recognition (CVPR)}, 1227 | month = jul, 1228 | year = {2017}, 1229 | month_numeric = {7} 1230 | } 1231 | ``` 1232 | 1233 | 1234 | :cherries: **FashionGAN 2017** 1235 | 1236 | (1) Extended the DeepFashion by collecting sentence descriptions for **79K** images. 1237 | 1238 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_ICCV_2017/papers/Zhu_Be_Your_Own_ICCV_2017_paper.pdf) [[github]](https://github.com/zhusz/ICCV17-fashionGAN) :pig: 1239 | ```bib 1240 | @inproceedings{zhu2017be, 1241 | title={Be Your Own Prada: Fashion Synthesis with Structural Coherence}, 1242 | author={Zhu, Shizhan and Fidler, Sanja and Urtasun, Raquel and Lin, Dahua and Chen, Change Loy}, 1243 | booktitle={Proceedings of the IEEE Conference on International Conference on Computer Vision}, 1244 | year={2017} 1245 | } 1246 | ``` 1247 | 1248 | 1249 | :cherries: **BeautyGAN 2018** 1250 | 1251 | (1) Facial makeup dataset consists of **3,834** female images. 1252 | 1253 | [[homepage]](http://liusi-group.com/projects/BeautyGAN) [[pdf]](https://dl.acm.org/doi/pdf/10.1145/3240508.3240618) :pig: 1254 | ```bib 1255 | @inproceedings{li2018beautygan, 1256 | title={Beautygan: Instance-level facial makeup transfer with deep generative adversarial network}, 1257 | author={Li, Tingting and Qian, Ruihe and Dong, Chao and Liu, Si and Yan, Qiong and Zhu, Wenwu and Lin, Liang}, 1258 | booktitle={Proceedings of the 26th ACM international conference on Multimedia}, 1259 | pages={645--653}, 1260 | year={2018} 1261 | } 1262 | ``` 1263 | 1264 | 1265 | :cherries: **VTON 2018** 1266 | 1267 | (1) Collected dataset from Zalando. 1268 | (2) **19,000** frontal-view woman and top2 image pairs (removed noisy images with no parsing results), yielding **16,253** pairs. 1269 | 1270 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_cvpr_2018/papers/Han_VITON_An_Image-Based_CVPR_2018_paper.pdf) [[github]](https://github.com/xthan/VITON) 1271 | ```bib 1272 | @inproceedings{han2017viton, 1273 | title = {VITON: An Image-based Virtual Try-on Network}, 1274 | author = {Han, Xintong and Wu, Zuxuan and Wu, Zhe and Yu, Ruichi and Davis, Larry S}, 1275 | booktitle = {CVPR}, 1276 | year = {2018}, 1277 | } 1278 | ``` 1279 | 1280 | 1281 | :cherries: **DeepWear 2018** 1282 | 1283 | (1) A specific brand clothes dataset. 1284 | 1285 | [homepage] [[pdf]](https://digitalnature.slis.tsukuba.ac.jp/wp-content/uploads/2018/03/deepwear.pdf) 1286 | ```bib 1287 | @inproceedings{kato2018deepwear, 1288 | title={DeepWear: a Case Study of Collaborative Design between Human and Artificial Intelligence}, 1289 | author={Kato, Natsumi and Osone, Hiroyuki and Sato, Daitetsu and Muramatsu, Naoya and Ochiai, Yoichi}, 1290 | booktitle={Proceedings of the Twelfth International Conference on Tangible, Embedded, and Embodied Interaction}, 1291 | pages={529--536}, 1292 | year={2018} 1293 | } 1294 | ``` 1295 | 1296 | 1297 | :cherries: **FashionGEN 2018** 1298 | 1299 | (1) **293,008** high-resolution fashion images paired with item descriptions provided by professional stylists. 1300 | (2) All fashion items are photographed from 1 to 6 different angles depending on the category of the item. 1301 | (3) **48** main categories, and **121** fine-grained sub-categories. 1302 | (4) Paired with paragraph-length descriptive captions sourced from experts (professional designers). 1303 | (5) Provide metadata such as stylist recommended matched items, the fashion season, designer and the brand. 1304 | (6) Provide the distribution of colors extracted from the text description. 1305 | 1306 | [[homepage]](https://fashion-gen.com/) [[pdf]](https://arxiv.org/pdf/1806.08317.pdf) 1307 | ```bib 1308 | @article{rostamzadeh2018fashion, 1309 | title={Fashion-gen: The generative fashion dataset and challenge}, 1310 | author={Rostamzadeh, Negar and Hosseini, Seyedarian and Boquet, Thomas and Stokowiec, Wojciech and Zhang, Ying and Jauvin, Christian and Pal, Chris}, 1311 | journal={arXiv preprint arXiv:1806.08317}, 1312 | year={2018} 1313 | } 1314 | ``` 1315 | 1316 | 1317 | :cherries: **ZalandoGAN 2018** 1318 | 1319 | (1) Over **120,000** images of dresses that are downloaded from Zalando’s website. 1320 | 1321 | [[homepage]](https://research.zalando.com/welcome/mission/research-projects/generative-fashion-design/) [[pdf]](https://arxiv.org/pdf/1806.07819.pdf) [[github]](https://github.com/zalandoresearch/disentangling_conditional_gans) 1322 | ```bib 1323 | @article{yildirim2018disentangling, 1324 | title={Disentangling multiple conditional inputs in gans}, 1325 | author={Yildirim, G{\"o}khan and Seward, Calvin and Bergmann, Urs}, 1326 | journal={arXiv preprint arXiv:1806.07819}, 1327 | year={2018} 1328 | } 1329 | ``` 1330 | 1331 | 1332 | :cherries: **RTW 2018** 1333 | 1334 | (1) Augment the dataset of **4,157** images by a factor 5 by jittering images with random scaling and translations. 1335 | (2) 7 categories: jackets, coats, shirts, tops, t-shirts, dresses and pullovers. 1336 | (3) 7 print: plain(uniform), plain(tiled?), striped, animal print(animal skin), dotted, graphic print(print and graphical pattern). 1337 | 1338 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_ECCVW_2018/papers/11131/Sbai_DesIGN_Design_Inspiration_from_Generative_Networks_ECCVW_2018_paper.pdf) 1339 | ```bib 1340 | @inproceedings{sbai2018design, 1341 | title={Design: Design inspiration from generative networks}, 1342 | author={Sbai, Othman and Elhoseiny, Mohamed and Bordes, Antoine and LeCun, Yann and Couprie, Camille}, 1343 | booktitle={Proceedings of the European Conference on Computer Vision (ECCV)}, 1344 | pages={0--0}, 1345 | year={2018} 1346 | } 1347 | ``` 1348 | 1349 | 1350 | :cherries: **SMPL 2018** 1351 | 1352 | (1) Comprises a pair of jeans, a T-shirt and a sweater worn by **600** bodies in various poses. 1353 | 1354 | [[homepage]](https://www.epfl.ch/labs/cvlab/research/garment-simulation/garnet/) [[pdf]](http://openaccess.thecvf.com/content_ICCV_2019/papers/Gundogdu_GarNet_A_Two-Stream_Network_for_Fast_and_Accurate_3D_Cloth_ICCV_2019_paper.pdf) 1355 | ```bib 1356 | @inproceedings{gundogdu19garnet, 1357 | title = {Garnet: A Two-stream Network for Fast and Accurate 3D Cloth Draping}, 1358 | author = {Gundogdu, Erhan and Constantin, Victor and Seifoddini, Amrollah and Dang, Minh and Salzmann, Mathieu and Fua, Pascal}, 1359 | booktitle = {{IEEE} International Conference on Computer Vision ({ICCV})}, 1360 | month = {oct}, 1361 | organization = {{IEEE}}, 1362 | year = {2019}, 1363 | } 1364 | ``` 1365 | 1366 | 1367 | :cherries: **Fashion Takes 2019** 1368 | 1369 | (1) Over **18,000** images with meta-data including clothing category. 1370 | (2) Manual shape annotation indicating whether the person’s shape is above average or average. 1371 | (3) The data comprises 181 different users. 1372 | (4) Allowed to study ** the relationship between clothing categories and body shape**. 1373 | 1374 | [homepage] [[pdf]](https://virtualhumans.mpi-inf.mpg.de/papers/sattar2019WACV/sattar2019WACV.pdf) 1375 | ```bib 1376 | @inproceedings{sattar2019fashion, 1377 | title={Fashion is taking shape: Understanding clothing preference based on body shape from online sources}, 1378 | author={Sattar, Hosnieh and Pons-Moll, Gerard and Fritz, Mario}, 1379 | booktitle={2019 IEEE Winter Conference on Applications of Computer Vision (WACV)}, 1380 | pages={968--977}, 1381 | year={2019}, 1382 | organization={IEEE} 1383 | } 1384 | ``` 1385 | 1386 | 1387 | :cherries: **StyleGAN 2019** 1388 | 1389 | (1) Based on a proprietary image dataset with around **380k** entries with high-resolution. 1390 | (2) An outfit is composed of a set of the maximum of **6** articles. 1391 | 1392 | [homepage] [[pdf]](http://openaccess.thecvf.com/content_ICCVW_2019/papers/CVFAD/Yildirim_Generating_High-Resolution_Fashion_Model_Images_Wearing_Custom_Outfits_ICCVW_2019_paper.pdf) 1393 | ```bib 1394 | @inproceedings{yildirim2019generating, 1395 | title={Generating High-Resolution Fashion Model Images Wearing Custom Outfits}, 1396 | author={Yildirim, Gokhan and Jetchev, Nikolay and Vollgraf, Roland and Bergmann, Urs}, 1397 | booktitle={Proceedings of the IEEE International Conference on Computer Vision Workshops}, 1398 | pages={0--0}, 1399 | year={2019} 1400 | } 1401 | ``` 1402 | 1403 | 1404 | :cherries: **Deep Fashion3D 2020** 1405 | 1406 | (1) **2,078** models reconstructed from real garments, which covers **10** different categories and **563** garment instances. 1407 | (2) Multi-view stereo, multi-view real images, 3D feature lines, 3D body pose. 1408 | 1409 | [homepage] [[pdf]](https://arxiv.org/pdf/2003.12753.pdf) 1410 | ```bib 1411 | @article{zhu2020deep, 1412 | title={Deep Fashion3D: A Dataset and Benchmark for 3D Garment Reconstruction from Single Images}, 1413 | author={Zhu, Heming and Cao, Yu and Jin, Hang and Chen, Weikai and Du, Dong and Wang, Zhangye and Cui, Shuguang and Han, Xiaoguang}, 1414 | journal={arXiv preprint arXiv:2003.12753}, 1415 | year={2020} 1416 | } 1417 | ``` 1418 | 1419 | ## 5.  Others 1420 | :cherries: **Amazon Reviews 2015** 1421 | 1422 | (1) Based on the **Amazon** web store. 1423 | (2) Over **180 million** relationships between a pool of almost 6 million objects. 1424 | 1425 | [[homepage]](https://nijianmo.github.io/amazon/index.html) [[pdf]](https://dl.acm.org/doi/pdf/10.1145/2766462.2767755) 1426 | ```bib 1427 | @inproceedings{mcauley2015image, 1428 | title={Image-based recommendations on styles and substitutes}, 1429 | author={McAuley, Julian and Targett, Christopher and Shi, Qinfeng and Van Den Hengel, Anton}, 1430 | booktitle={Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval}, 1431 | pages={43--52}, 1432 | year={2015} 1433 | } 1434 | ``` 1435 | 1436 | 1437 | :cherries: **Tradesy 2016** 1438 | 1439 | (1) **19,823** users, **166,526** items, **410,186** feedback. 1440 | (2) Contains users’ purchase histories and ‘thumbs-up’. 1441 | 1442 | [[homepage]](http://jmcauley.ucsd.edu/data/tradesy/) [[pdf]](http://cseweb.ucsd.edu/~jmcauley/pdfs/aaai16.pdf) :pig: 1443 | ```bib 1444 | @inproceedings{he2016vbpr, 1445 | title={VBPR: visual bayesian personalized ranking from implicit feedback}, 1446 | author={He, Ruining and McAuley, Julian}, 1447 | booktitle={Thirtieth AAAI Conference on Artificial Intelligence}, 1448 | year={2016} 1449 | } 1450 | ``` 1451 | 1452 | 1453 | :cherries: **Fashion-MNIST 2017** 1454 | 1455 | (1) Comprising of 28 × 28 grayscale images of **70,000** fashion products from 10 categories. 1456 | 1457 | [[homepage]](https://research.zalando.com/welcome/mission/research-projects/fashion-mnist/) [[pdf]](https://arxiv.org/pdf/1708.07747.pdf) [[github]](https://github.com/zalandoresearch/fashion-mnist) :pig: 1458 | ```bib 1459 | @online{xiao2017/online, 1460 | author = {Han Xiao and Kashif Rasul and Roland Vollgraf}, 1461 | title = {Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms}, 1462 | date = {2017-08-28}, 1463 | year = {2017}, 1464 | eprintclass = {cs.LG}, 1465 | eprinttype = {arXiv}, 1466 | eprint = {cs.LG/1708.07747}, 1467 | } 1468 | ``` 1469 | 1470 | 1471 | :cherries: **Shoe 2018** 1472 | 1473 | (1) Supports further research on the task of relative **image captioning**. 1474 | (2) Pairing **3,600** captions that were discriminative with additional dissimilar images. 1475 | 1476 | [homepage] [[pdf]](http://papers.nips.cc/paper/7348-dialog-based-interactive-image-retrieval.pdf) 1477 | ```bib 1478 | @inproceedings{guo2018dialog, 1479 | title={Dialog-based interactive image retrieval}, 1480 | author={Guo, Xiaoxiao and Wu, Hui and Cheng, Yu and Rennie, Steven and Tesauro, Gerald and Feris, Rogerio}, 1481 | booktitle={Advances in Neural Information Processing Systems}, 1482 | pages={678--688}, 1483 | year={2018} 1484 | } 1485 | ``` 1486 | 1487 | 1488 | :cherries: **Flickr30k 2018** 1489 | 1490 | (1) **300k** posts with **5M** comments. 1491 | (2) Each image is paired with user comment. The maximum number of comments is 427, average per image is 14. 1492 | 1493 | [homepage] [[pdf]](https://dl.acm.org/doi/pdf/10.1145/3184558.3186354) 1494 | ```bib 1495 | @inproceedings{lin2018netizen, 1496 | title={Netizen-style commenting on fashion photos: dataset and diversity measures}, 1497 | author={Lin, Wen Hua and Chen, Kuan-Ting and Chiang, Hung Yueh and Hsu, Winston}, 1498 | booktitle={Companion Proceedings of the The Web Conference 2018}, 1499 | pages={395--402}, 1500 | year={2018} 1501 | } 1502 | ``` 1503 | :cherries: **FCDB 2019** 1504 | 1505 | (1) **100 million** Flickr images which focus on 21 global cities based on city perception. 1506 | (2) **25,707,690** clothing images for **trend analysis**. 1507 | 1508 | [[homepage]](http://hirokatsukataoka.net/) [[pdf]](http://openaccess.thecvf.com/content_CVPRW_2019/papers/FFSS-USAD/Kataoka_Ten-Million-Order_Human_Database_for_World-Wide_Fashion_Culture_Analysis_CVPRW_2019_paper.pdf) [[github]](https://github.com/cvpaperchallenge/FashionCultureDataBase_DLoader) :pig: 1509 | ```bib 1510 | @InProceedings{Kataoka_2019_CVPR_Workshops, 1511 | author = {Kataoka, Hirokatsu and Satoh, Yutaka and Abe, Kaori and Minoguchi, Munetaka and Nakamura, Akio}, 1512 | title = {Ten-Million-Order Human Database for World-Wide Fashion Culture Analysis}, 1513 | booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops}, 1514 | month = {June}, 1515 | year = {2019} 1516 | } 1517 | ``` 1518 | 1519 | 1520 | :cherries: **Fashion IQ 2019** 1521 | 1522 | (1) Dataset for **natural language based fashion image retrieval**. 1523 | (2) Each image is crawled from Amazon.com and extracted corresponding product information, when available. 1524 | 1525 | [homepage] [[pdf]](https://arxiv.org/pdf/1905.12794.pdf) [[github]](https://github.com/XiaoxiaoGuo/fashion-iq) [[github]](https://github.com/XiaoxiaoGuo/fashion-iq) :pig: 1526 | ```bib 1527 | @article{guo2019fashion, 1528 | title={The Fashion IQ Dataset: Retrieving Images by Combining Side Information and Relative Natural Language Feedback}, 1529 | author={Guo, Xiaoxiao and Wu, Hui and Gao, Yupeng and Rennie, Steven and Feris, Rogerio}, 1530 | journal={arXiv preprint arXiv:1905.12794}, 1531 | year={2019} 1532 | } 1533 | ``` 1534 | 1535 | 1536 | :cherries: **TFCD 2019** 1537 | 1538 | (1) A dataset containing data from real user sessions on a major European e-commerce fashion website. 1539 | 1540 | [homepage] [[pdf]](https://arxiv.org/ftp/arxiv/papers/1907/1907.00400.pdf) 1541 | ```bib 1542 | @article{bigon2019prediction, 1543 | title={Prediction is very hard, especially about conversion. Predicting user purchases from clickstream data in fashion e-commerce}, 1544 | author={Bigon, Luca and Cassani, Giovanni and Greco, Ciro and Lacasa, Lucas and Pavoni, Mattia and Polonioli, Andrea and Tagliabue, Jacopo}, 1545 | journal={arXiv preprint arXiv:1907.00400}, 1546 | year={2019} 1547 | } 1548 | ``` 1549 | 1550 | ## Other Sources 1551 | - MPV (Multi-Pose Virtual try on) dataset [[link]](http://sysu-hcp.net/lip/overview.php) 1552 | - Clothing Attributes Dataset [[link]](https://exhibits.stanford.edu/data/catalog/tb980qz1002) 1553 | - Clothing Detection Dataset [[link]](https://github.com/seralexger/clothing-detection-dataset) 1554 | - Clothing Size Recommendation [[link]](https://github.com/NeverInAsh/fit-recommendation) 1555 | - Fashion Product Image [[link]](https://www.kaggle.com/paramaggarwal/fashion-product-images-dataset) 1556 | - Dresses_Attribute_Sales Data Set [[link]](https://archive.ics.uci.edu/ml/datasets/Dresses_Attribute_Sales) 1557 | - HICP-Clothing [[link]](https://data.europa.eu/euodp/en/data/dataset/sE1cgO8hyGVp2RxD9iafA) 1558 | - Paper Doll Raw Dataset [[link]](https://github.com/kyamagu/paperdoll/tree/master/data/chictopia) 1559 | - Fashion Toolbox [[link]](https://github.com/open-mmlab/mmfashion) 1560 | - https://data.world/datasets/fashion 1561 | - https://tianchi.aliyun.com/dataset/ 1562 | ## Acknowledge 1563 | Here I would like to thank Miss Po Yee(Boey), PANG, Miss Wai Lee(Selene), CHONG, for their hard work on collecting the datasource information. 1564 | --------------------------------------------------------------------------------