RCBSR: Re-Parameterization Convolution Block for Super-Resolution
Abstract
Super resolution(SR) with high efficiency and low power consumption is highly demanded in the actual application scenes. In this paper, We designed a super light-weight SR network with strong feature expression. The network we proposed is named RCBSR. Based on the novel technique of re-parameterization, we adopt a block with multiple paths structure in the training stage and merge multiple paths structure into one single 3 $\times $ × 3 convolution in the inference stage. And then the neural architecture search(NAS) method is adopted to determine amounts of block M and amounts of channel C. Finally, the proposed SR network achieves a fairly good result of PSNR(27.52 dB) with power consumption(0.1 W@30 fps) on the MediaTek Dimensity 9000 platform in the challenge testing stage.
Cite
Text
Gao et al. "RCBSR: Re-Parameterization Convolution Block for Super-Resolution." European Conference on Computer Vision Workshops, 2022. doi:10.1007/978-3-031-25063-7_33Markdown
[Gao et al. "RCBSR: Re-Parameterization Convolution Block for Super-Resolution." European Conference on Computer Vision Workshops, 2022.](https://mlanthology.org/eccvw/2022/gao2022eccvw-rcbsr/) doi:10.1007/978-3-031-25063-7_33BibTeX
@inproceedings{gao2022eccvw-rcbsr,
title = {{RCBSR: Re-Parameterization Convolution Block for Super-Resolution}},
author = {Gao, Si and Zheng, Chengjian and Zhang, Xiaofeng and Liu, Shaoli and Wu, Biao and Lu, Kaidi and Zhang, Diankai and Wang, Ning},
booktitle = {European Conference on Computer Vision Workshops},
year = {2022},
pages = {540-548},
doi = {10.1007/978-3-031-25063-7_33},
url = {https://mlanthology.org/eccvw/2022/gao2022eccvw-rcbsr/}
}