Multi-width Activation and Multiple Receptive Field Networks for Dynamic Scene Deblurring

Jinkai Cui, Weihong Li, Wei Guo, Weiguo Gong
Proceedings of The Eleventh Asian Conference on Machine Learning, PMLR 101:852-867, 2019.

Abstract

In this paper, we propose an end-to-end multi-width activation and multiple receptive field networks for the large-scale and complicated dynamic scene deblurring. Firstly, we design a multi-width activation feature extraction module, in which a multi-width activation residual block is proposed for making the activation function learn more the nonlinear information and extracting wider nonlinear features. Secondly, we design a multiple receptive field (RF) feature extraction module, in which a multiple RF residual block is proposed for enlarging the RF efficiently and capturing more nonlinear information from distant locations. And then, we design the multi-scale feature fusion module, where a learning fusion structure is designed to adaptively fuse the multi-scale features and complicated blur information from the different modules. Finally, we use a multi-component loss function to jointly optimize our networks. Extensive experimental results demonstrate that the proposed method outperforms the recent state-of-the-art deblurring methods, both quantitatively and qualitatively.

Cite this Paper

BibTeX
@InProceedings{pmlr-v101-cui19b, title = {Multi-width Activation and Multiple Receptive Field Networks for Dynamic Scene Deblurring}, author = {Cui, Jinkai and Li, Weihong and Guo, Wei and Gong, Weiguo}, booktitle = {Proceedings of The Eleventh Asian Conference on Machine Learning}, pages = {852--867}, 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/cui19b/cui19b.pdf}, url = {https://proceedings.mlr.press/v101/cui19b.html}, abstract = {In this paper, we propose an end-to-end multi-width activation and multiple receptive field networks for the large-scale and complicated dynamic scene deblurring. Firstly, we design a multi-width activation feature extraction module, in which a multi-width activation residual block is proposed for making the activation function learn more the nonlinear information and extracting wider nonlinear features. Secondly, we design a multiple receptive field (RF) feature extraction module, in which a multiple RF residual block is proposed for enlarging the RF efficiently and capturing more nonlinear information from distant locations. And then, we design the multi-scale feature fusion module, where a learning fusion structure is designed to adaptively fuse the multi-scale features and complicated blur information from the different modules. Finally, we use a multi-component loss function to jointly optimize our networks. Extensive experimental results demonstrate that the proposed method outperforms the recent state-of-the-art deblurring methods, both quantitatively and qualitatively.} }
Endnote
%0 Conference Paper %T Multi-width Activation and Multiple Receptive Field Networks for Dynamic Scene Deblurring %A Jinkai Cui %A Weihong Li %A Wei Guo %A Weiguo Gong %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-cui19b %I PMLR %P 852--867 %U https://proceedings.mlr.press/v101/cui19b.html %V 101 %X In this paper, we propose an end-to-end multi-width activation and multiple receptive field networks for the large-scale and complicated dynamic scene deblurring. Firstly, we design a multi-width activation feature extraction module, in which a multi-width activation residual block is proposed for making the activation function learn more the nonlinear information and extracting wider nonlinear features. Secondly, we design a multiple receptive field (RF) feature extraction module, in which a multiple RF residual block is proposed for enlarging the RF efficiently and capturing more nonlinear information from distant locations. And then, we design the multi-scale feature fusion module, where a learning fusion structure is designed to adaptively fuse the multi-scale features and complicated blur information from the different modules. Finally, we use a multi-component loss function to jointly optimize our networks. Extensive experimental results demonstrate that the proposed method outperforms the recent state-of-the-art deblurring methods, both quantitatively and qualitatively.
APA
Cui, J., Li, W., Guo, W. & Gong, W.. (2019). Multi-width Activation and Multiple Receptive Field Networks for Dynamic Scene Deblurring. Proceedings of The Eleventh Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 101:852-867 Available from https://proceedings.mlr.press/v101/cui19b.html.

Related Material