A Generate Adversarial Network with Structural Branch Assistance for Image Inpainting

In digital image inpainting tasks, existing deep-learning-based methods have achieved remarkable staged results by introducing structural prior information into the network. However, corresponding relationship between texture and structure is not fully considered, inconsistency appears in of current method. this paper, we propose a dual-branch network with branch assistance, which decouples low-frequency high-frequency utilizing parallel branches. The feature fusion (FF) module introduced to integrate from two branches, effectively ensures consistency image. attention (FA) extract long-distance information, enhances local features overall gives more detailed texture. Experiments on Paris StreetView CelebA-HQ datasets prove effectiveness superiority our