`handbag_data/github_datasets/FANCY/README.md`

38 lines
# FANCY datasets


# Descriptions
We contribute the ADM and the FANCY database, a large-scale clothes database, which have appealing properties:

## ADM  ([Download](https://drive.google.com/drive/folders/1rvd5D0vv2MrMEaZZpmeJ2ny6Tn44DyA-?usp=sharing))
- The ADM dataset contains 25,470 diverse fashion images ranging from well-posed shop images to unconstrained consumer photos, which consists of FashionStyle14 dataset (12,190 images) and the ones crawled from Google Images (13,280 images).
- 146 fashion attributes were multi-labeled for each image through collaborating with fashion professionals.
- We built the ADM dataset for developing our object detection model. 
![FIg001](https://user-images.githubusercontent.com/30140635/107113079-d8b8a380-689f-11eb-9d7e-99d18c15b5e4.jpeg)


## FANCY  ([Download](https://drive.google.com/drive/folders/1abIiasmgCSdvNpEDJM9-2iaURlTQh8PY?usp=sharing))
- The FANCY dataset contains 302,772 images with 146 attributes and 25 style groups.
A total of 302,772 images from 8,121 fashion shows were collected from Vogue US between 2010 and 2019. The fashion images covers 987 brand names ranging from mega couture (e.g., Gucci, Chanel) to high street brands (e.g., JCrew, Topshop).
- 25 style groups were labelled, considering fashion professionals’ knowledge and a computational method.


![Fig05 001](https://user-images.githubusercontent.com/30140635/107113084-e2420b80-689f-11eb-89ab-7251ecdf02e5.jpeg)


For more details of the datasets, please refer to the paper, FANCY: Deep Learning-based Framework for Fashion Style Analysis, WWW’21. 
 
# Agreement
The FANCY and ADM datasets are available for non-commercial research purposes only.
You agree not to reproduce, duplicate, copy, sell, trade, resell or exploit for any commercial purposes, any portion of the images and any portion of derived data.
The AjouHCI lab reserves the right to terminate your access to the FANCY and ADM datasets at any time.

# Citation
Please cite the following papers if you use the dataset.
@inproceedings{Jeonwww21fancy,
 author = {Youngseung Jeon, Seungwan Jin, Kyungsik Han},
 title = {FANCY:Deep Learning-based Framework for Fashion Style Analysis} }

### Contact
Please contact Youngseung Jeon and Seungwan Jin for questions about the datasets.