Fashion Synthesis Benchmark facilitates the studies of generating new clothing images. It includes 78,979 images selected from the In-shop Clothes Benchmark. Each image is associated with several sentences as captions and a segmentation map. If you use the images, captions, and segmentations, please appropriately cite the papers of DeepFashion and FashionGAN.

  • The 78,979 images can be downloaded from Google Drive or Baidu Drive.

  • The captions and segmentations for these images can be downloaded from FashionGAN.

  • Images with high resolutions are released upon request.

Download Instructions

  • Some image data are encrypted to prevent unauthorized access. Please download the DeepFashion dataset Release Agreement.
  • Read it carefully, complete and sign it appropriately. This is an example.
  • Please send the completed form to Ziwei Liu ( and cc to Ping Luo (pluo(at) using institutional email address. The email Subject Title is "DeepFashion Agreement". We will verify your request and contact you with the passwords to unzip the image data.


 author = {Liu, Ziwei and Luo, Ping and Qiu, Shi and Wang, Xiaogang and Tang, Xiaoou},
 title = {DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations},
 booktitle = {Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
 month = {June},
 year = {2016}
 author = {Zhu, Shizhan and Fidler, Sanja and Urtasun, Raquel and Lin, Dahua and Loy, Chen Change},
 title = {Be Your Own Prada: Fashion Synthesis with Structural Coherence},
 booktitle = {International Conference on Computer Vision (ICCV)},
 month = {October},
 year = {2017}


Please contact Ziwei Liu and Ping Luo for questions about the dataset.