Dual-path attention network for single image super-resolution
作者：
Highlights：
• A novel dual-path attention network is proposed for single image super-resolution.
• The model focus on the high-frequency features because of the dual skip connections.
• A dual-path topology is introduced to build our dual-path blocks.
• The path attention fusion module adaptively integrates the output of the two paths.
• Performance of the proposed method is evaluated on several public benchmark datasets.
摘要
•A novel dual-path attention network is proposed for single image super-resolution.•The model focus on the high-frequency features because of the dual skip connections.•A dual-path topology is introduced to build our dual-path blocks.•The path attention fusion module adaptively integrates the output of the two paths.•Performance of the proposed method is evaluated on several public benchmark datasets.
论文关键词：Super-resolution,Image restoration,Deep learning,Dual-path network,Attention
论文评审过程：Received 21 August 2020, Revised 2 December 2020, Accepted 2 December 2020, Available online 7 December 2020, Version of Record 24 December 2020.
论文官网地址：https://doi.org/10.1016/j.eswa.2020.114450
