Home

Awesome

Diversity-Sensitive Conditional GANs

Created by Dingdong Yang, Seunghoon Hong, Yunseok Jang, Tianchen Zhao, Honglak Lee

Paper | Project page

Introduction

This repository contains the implementations of Diversity-Sensitive Conditional Generative Adversarial Networks (ICLR2019), which proposes a simple regularization term that can be nicely plugged into any cGAN framework to promote diversity in the generator outputs.

@inproceedings{dsganICLR2019,
  title={Diversity-Sensitive Conditional Generative Adversarial Networks},
  author={Yang, Dingdong and Hong, Seunghoon and Jang, Yunseok and Zhao, Tianchen and Lee, Honglak},
  booktitle={Proceedings of the International Conference on Learning Representations},
  year={2019}
}

Example Results

Image-to-Image Translation

<img src='Image2ImageTranslation/example/im2im.png'>

Image Inpainting

Video Prediction

<img src="VideoPrediction/example/kth_1.gif" width="100%" alt="KTH_1"> <img src="VideoPrediction/example/kth_2.gif" width="100%" alt="KTH_2"> <img src="VideoPrediction/example/bair_1.gif" width="100%" alt="BAIR_1"> <img src="VideoPrediction/example/bair_2.gif" width="100%" alt="BAIR_2">