Fully 11 Convolutional Network for Lightweight Image Super-Resolution

Code

• github.com/aitical/scnet
  OfficialIn paperpytorch★ 100

Abstract

Deep models have achieved significant process on single image super-resolution (SISR) tasks, in particular large models with large kernel (33 or more). However, the heavy computational footprint of such models prevents their deployment in real-time, resource-constrained environments. Conversely, 11 convolutions bring substantial computational efficiency, but struggle with aggregating local spatial representations, an essential capability to SISR models. In response to this dichotomy, we propose to harmonize the merits of both 33 and 11 kernels, and exploit a great potential for lightweight SISR tasks. Specifically, we propose a simple yet effective fully 11 convolutional network, named Shift-Conv-based Network (SCNet). By incorporating a parameter-free spatial-shift operation, it equips the fully 11 convolutional network with powerful representation capability while impressive computational efficiency. Extensive experiments demonstrate that SCNets, despite its fully 11 convolutional structure, consistently matches or even surpasses the performance of existing lightweight SR models that employ regular convolutions. The code and pre-trained models can be found at https://github.com/Aitical/SCNet.

Tasks

Reproductions