Random Shuffle Transformer for Image Restoration

Jie Xiao, Xueyang Fu, Man Zhou, Hongjian Liu, Zheng-Jun Zha
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:38039-38058, 2023.

Abstract

Non-local interactions play a vital role in boosting performance for image restoration. However, local window Transformer has been preferred due to its efficiency for processing high-resolution images. The superiority in efficiency comes at the cost of sacrificing the ability to model non-local interactions. In this paper, we present that local window Transformer can also function as modeling non-local interactions. The counterintuitive function is based on the permutation-equivariance of self-attention. The basic principle is quite simple: by randomly shuffling the input, local self-attention also has the potential to model non-local interactions without introducing extra parameters. Our random shuffle strategy enjoys elegant theoretical guarantees in extending the local scope. The resulting Transformer dubbed ShuffleFormer is capable of processing high-resolution images efficiently while modeling non-local interactions. Extensive experiments demonstrate the effectiveness of ShuffleFormer across a variety of image restoration tasks, including image denoising, deraining, and deblurring. Code is available at https://github.com/jiexiaou/ShuffleFormer.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-xiao23a, title = {Random Shuffle Transformer for Image Restoration}, author = {Xiao, Jie and Fu, Xueyang and Zhou, Man and Liu, Hongjian and Zha, Zheng-Jun}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {38039--38058}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/xiao23a/xiao23a.pdf}, url = {https://proceedings.mlr.press/v202/xiao23a.html}, abstract = {Non-local interactions play a vital role in boosting performance for image restoration. However, local window Transformer has been preferred due to its efficiency for processing high-resolution images. The superiority in efficiency comes at the cost of sacrificing the ability to model non-local interactions. In this paper, we present that local window Transformer can also function as modeling non-local interactions. The counterintuitive function is based on the permutation-equivariance of self-attention. The basic principle is quite simple: by randomly shuffling the input, local self-attention also has the potential to model non-local interactions without introducing extra parameters. Our random shuffle strategy enjoys elegant theoretical guarantees in extending the local scope. The resulting Transformer dubbed ShuffleFormer is capable of processing high-resolution images efficiently while modeling non-local interactions. Extensive experiments demonstrate the effectiveness of ShuffleFormer across a variety of image restoration tasks, including image denoising, deraining, and deblurring. Code is available at https://github.com/jiexiaou/ShuffleFormer.} }
Endnote
%0 Conference Paper %T Random Shuffle Transformer for Image Restoration %A Jie Xiao %A Xueyang Fu %A Man Zhou %A Hongjian Liu %A Zheng-Jun Zha %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-xiao23a %I PMLR %P 38039--38058 %U https://proceedings.mlr.press/v202/xiao23a.html %V 202 %X Non-local interactions play a vital role in boosting performance for image restoration. However, local window Transformer has been preferred due to its efficiency for processing high-resolution images. The superiority in efficiency comes at the cost of sacrificing the ability to model non-local interactions. In this paper, we present that local window Transformer can also function as modeling non-local interactions. The counterintuitive function is based on the permutation-equivariance of self-attention. The basic principle is quite simple: by randomly shuffling the input, local self-attention also has the potential to model non-local interactions without introducing extra parameters. Our random shuffle strategy enjoys elegant theoretical guarantees in extending the local scope. The resulting Transformer dubbed ShuffleFormer is capable of processing high-resolution images efficiently while modeling non-local interactions. Extensive experiments demonstrate the effectiveness of ShuffleFormer across a variety of image restoration tasks, including image denoising, deraining, and deblurring. Code is available at https://github.com/jiexiaou/ShuffleFormer.
APA
Xiao, J., Fu, X., Zhou, M., Liu, H. & Zha, Z.. (2023). Random Shuffle Transformer for Image Restoration. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:38039-38058 Available from https://proceedings.mlr.press/v202/xiao23a.html.

Related Material