Wide Activation for Efficient and Accurate Image Super-Resolution

By Jiahui Yu, Yuchen Fan, Jianchao Yang, Ning Xu, Zhaowen Wang, Xinchao Wang, Thomas S. Huang. 2018

381 citations recorded by the source corpus.

In this report we demonstrate that with same parameters and computational budgets, models with wider features before ReLU activation have significantly better performance for single image super-resolution (SISR). The resulted SR residual network has a slim identity mapping pathway with wider (\(2\times\) to \(4\times\)) channels before activation in each residual block. To further widen activation (\(6\times\) to \(9\times\)) without computational overhead, we introduce linear low-rank convolution into SR networks and achieve even better accuracy-efficiency tradeoffs. In addition, compared with batch normalization or no normalization, we find training with weight normalization leads to better accuracy for deep super-resolution networks. Our proposed SR network \textit{WDSR} achieves better results on large-scale DIV2K image super-resolution benchmark in terms of PSNR with same or lower computational complexity. Based on WDSR, our method also won 1st places in NTIRE 2018 Challenge on Single Image Super-Resolution in all three realistic tracks. Experiments and ablation studies support the importance of wide activation for image super-resolution. Code is released at: this https URL

Subfields: Computer Vision and Pattern Recognition, Media Technology, Efficiency & Compression

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

In Efficiency & Compression

Citations in the corpus

Scaling Autoregressive Models for Content-Rich Text-to-Image Generation (2022, methodology): “Stacked convolutional layers with residual connections are used as the super-resolution network module following WDSR [31] (12 residual blocks with 128 channels).”
Blind Super-Resolution Kernel Estimation using an Internal-GAN (2019, methodology): “7190 Blind-SR WDSR [36] - 1st in NTIRE track2 21.”
AIM 2019 Challenge on Constrained Super-Resolution: Methods and Results (2019, methodology): “Moreover, the wide-activate residual unit in WDSR [28] shown in Figure 8(a) is employed as the basic residual unit (Basic RU).”
Revisiting Stereo Depth Estimation From a Sequence-to-Sequence Perspective with Transformers (2020, background): “The expansion before the ReLU is to encourage better information flow [42].”
HINet: Half Instance Normalization Network for Image Restoration (2021, background): “As the disharmony between image restoration tasks and Batch Normalization (BN) discussed in [51], image restoration tasks commonly use small image patches and small mini-batch size to train the network, which causes the statistics of BN unstable.”
NTIRE 2019 Challenge on Video Deblurring and Super-Resolution: Dataset and Study (2019, methodology): “They built, WDVR architecture inspired by WDSR [40, 7].”