Publications

Detailed Information

ContraGAN: Contrastive learning for conditional image generation

Cited 0 time in Web of Science Cited 83 time in Scopus
Authors

Kang, Minguk; Park, Jaesik

Issue Date
2020
Publisher
Neural information processing systems foundation
Citation
Advances in Neural Information Processing Systems, Vol.2020-December
Abstract
Conditional image generation is the task of generating diverse images using class label information. Although many conditional Generative Adversarial Networks (GAN) have shown realistic results, such methods consider pairwise relations between the embedding of an image and the embedding of the corresponding label (data-to-class relations) as the conditioning losses. In this paper, we propose ContraGAN that considers relations between multiple image embeddings in the same batch (data-to-data relations) as well as the data-to-class relations by using a conditional contrastive loss. The discriminator of ContraGAN discriminates the authenticity of given samples and minimizes a contrastive objective to learn the relations between training images. Simultaneously, the generator tries to generate realistic images that deceive the authenticity and have a low contrastive loss. The experimental results show that ContraGAN outperforms state-of-the-art-models by 7.3% and 7.7% on Tiny ImageNet and ImageNet datasets, respectively. Besides, we experimentally demonstrate that contrastive learning helps to relieve the overfitting of the discriminator. For a fair comparison, we re-implement twelve state-of-the-art GANs using the PyTorch library. The software package is available at https://github.com/POSTECH-CVLab/PyTorch-StudioGAN.
ISSN
1049-5258
URI
https://hdl.handle.net/10371/201305
DOI
https://doi.org/10.48550/arXiv.2006.12681
Files in This Item:
There are no files associated with this item.
Appears in Collections:

Related Researcher

  • College of Engineering
  • Dept. of Computer Science and Engineering
Research Area Computer Graphics, Computer Vision, Machine Learning

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share