Efficient Progressive High Dynamic Range Image Restoration via Attention and Alignment Network
This work addresses the problem of efficient HDR image restoration for computational photography, presenting an incremental improvement with specific performance gains.
The paper tackles the challenge of high dynamic range (HDR) image restoration by proposing a lightweight neural network, EAPNet, which achieves about 20 times compression on MAccs with improved mu-PSNR and PSNR compared to state-of-the-art methods, securing second place in the NTIRE 2022 HDR Track 1 and Track 2.
HDR is an important part of computational photography technology. In this paper, we propose a lightweight neural network called Efficient Attention-and-alignment-guided Progressive Network (EAPNet) for the challenge NTIRE 2022 HDR Track 1 and Track 2. We introduce a multi-dimensional lightweight encoding module to extract features. Besides, we propose Progressive Dilated U-shape Block (PDUB) that can be a progressive plug-and-play module for dynamically tuning MAccs and PSNR. Finally, we use fast and low-power feature-align module to deal with misalignment problem in place of the time-consuming Deformable Convolutional Network (DCN). The experiments show that our method achieves about 20 times compression on MAccs with better mu-PSNR and PSNR compared to the state-of-the-art method. We got the second place of both two tracks during the testing phase. Figure1. shows the visualized result of NTIRE 2022 HDR challenge.