A more efficient all-in-one deep learning network for image dehazing
GScan OÜ, Peng Zhang, Ting-quan He, R. X. Li, Xinrong Liu, Yuhui Wei, Jianping Xu, Zhibin Yu · International Journal of Parallel Emergent and Distributed Systems · 2025
In this paper, based on the AOD-Net, we proposed a more efficient all-in-one dehazing deep learning network (MEAODNet) model to improving the quality of image data. Compared to the AOD-net, the Feature attention (FA) module with the channel attention and pixel attention mechanisms in MEAOD-Net is introduced to capture the channel-level and pixel-level information, which improves the representation ability of the MEAOD-Net. Meanwhile, in MEAOD-Net, a pyramid pooling module is introduced to expand the improved network's receptive field and fuse features from different scales. Experimental results shown that our method has better performance than the traditional methods.