Variational Conditional GAN for Fine-grained Controllable Image Generation

Mingqi Hu, Deyu Zhou, Yulan He
Proceedings of The Eleventh Asian Conference on Machine Learning, PMLR 101:109-124, 2019.

Abstract

In this paper, we propose a novel variational generator framework for conditional GANs to catch semantic details for improving the generation quality and diversity. Traditional generators in conditional GANs simply concatenate the conditional vector with the noise as the input representation, which is directly employed for upsampling operations. However, the hidden condition information is not fully exploited, especially when the input is a class label. Therefore, we introduce a variational inference into the generator to infer the posterior of latent variable only from the conditional input, which helps achieve a variable augmented representation for image generation. Qualitative and quantitative experimental results show that the proposed method outperforms the state-of-the-art approaches and achieves the realistic controllable images.

Cite this Paper

BibTeX
@InProceedings{pmlr-v101-hu19a, title = {Variational Conditional GAN for Fine-grained Controllable Image Generation}, author = {Hu, Mingqi and Zhou, Deyu and He, Yulan}, booktitle = {Proceedings of The Eleventh Asian Conference on Machine Learning}, pages = {109--124}, year = {2019}, editor = {Lee, Wee Sun and Suzuki, Taiji}, volume = {101}, series = {Proceedings of Machine Learning Research}, month = {17--19 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v101/hu19a/hu19a.pdf}, url = {https://proceedings.mlr.press/v101/hu19a.html}, abstract = {In this paper, we propose a novel variational generator framework for conditional GANs to catch semantic details for improving the generation quality and diversity. Traditional generators in conditional GANs simply concatenate the conditional vector with the noise as the input representation, which is directly employed for upsampling operations. However, the hidden condition information is not fully exploited, especially when the input is a class label. Therefore, we introduce a variational inference into the generator to infer the posterior of latent variable only from the conditional input, which helps achieve a variable augmented representation for image generation. Qualitative and quantitative experimental results show that the proposed method outperforms the state-of-the-art approaches and achieves the realistic controllable images.} }
Endnote
%0 Conference Paper %T Variational Conditional GAN for Fine-grained Controllable Image Generation %A Mingqi Hu %A Deyu Zhou %A Yulan He %B Proceedings of The Eleventh Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Wee Sun Lee %E Taiji Suzuki %F pmlr-v101-hu19a %I PMLR %P 109--124 %U https://proceedings.mlr.press/v101/hu19a.html %V 101 %X In this paper, we propose a novel variational generator framework for conditional GANs to catch semantic details for improving the generation quality and diversity. Traditional generators in conditional GANs simply concatenate the conditional vector with the noise as the input representation, which is directly employed for upsampling operations. However, the hidden condition information is not fully exploited, especially when the input is a class label. Therefore, we introduce a variational inference into the generator to infer the posterior of latent variable only from the conditional input, which helps achieve a variable augmented representation for image generation. Qualitative and quantitative experimental results show that the proposed method outperforms the state-of-the-art approaches and achieves the realistic controllable images.
APA
Hu, M., Zhou, D. & He, Y.. (2019). Variational Conditional GAN for Fine-grained Controllable Image Generation. Proceedings of The Eleventh Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 101:109-124 Available from https://proceedings.mlr.press/v101/hu19a.html.

Related Material