It is collected by the OCSC and invited to conduct an bechmark of super-resolution.
This paper uses group convolutions and residual operations to enhance deep and wide correlations of different channels to implement an efficient SR network.
ESRGCNN_out.mp4
Absract
CNNs with strong learning abilities are widely chosen to resolve super-resolution problem. However, CNNs depend on deeper network architectures to improve performance of image super-resolution, which may increase computational cost in general. In this paper, we present an enhanced super-resolution group CNN (ESRGCNN) with a shallow architecture by fully fusing deep and wide channel features to extract more accurate low-frequency information in terms of correlations of different channels in single image super-resolution (SISR). Also, a signal enhancement operation in the ESRGCNN is useful to inherit more long-distance contextual information for resolving long-term dependency. An adaptive up-sampling operation is gathered into a CNN to obtain an image super-resolution model with low-resolution images of different sizes. Extensive experiments report that our ESRGCNN surpasses the state-of-the-arts in terms of SISR performance, complexity, execution speed, image quality evaluation and visual effect in SISR. Code is found at https://github.com/hellloxiaotian/ESRGCNN.
2. A parallel upsampling operation for training a blind SR model.
3. A upsampling operation for testing a blind SR model.
4. ESRGCNN for x2, x3 and x4 on Set5.
5. ESRGCNN for x2, x3 and x4 on Set14.
6. ESRGCNN for x2, x3 and x4 on B100.
7. ESRGCNN for x2, x3 and x4 on U100.
8. ESRGCNN for x2 on B100
9. Running time of different methods on hr images of size 256x256, 512x512 and 1024x1024 for x2.
10. Complexities of different methods for x2.
11. ESRGCNN for x2, x3 and x4 on B100 about FSIM
12. Visual results of U100 for x3.
13. Visual results of B100 for x2.
If you want to cite this paper, please refer to the following formats:
1. Tian C, Yuan Y, Zhang S, et al. Image Super-resolution with An Enhanced Group Convolutional Neural Network[J]. arXiv preprint arXiv:2205.14548, 2022.
2. @article{tian2022image,
title={Image Super-resolution with An Enhanced Group Convolutional Neural Network},
author={Tian, Chunwei and Yuan, Yixuan and Zhang, Shichao and Lin, Chia-Wen and Zuo, Wangmeng and Zhang, David},