GGD-GAN: Gradient-Guided dual-Branch adversarial networks for relic sketch generation

Jun Wang, Erlei Zhang, Shan Cui, Jiaxin Wang, Qunxi Zhang, Jianping Fan, Jinye Peng

Abstract: Highlights•The first end-to-end relics sketch generation framework based on GAN.•Gradient guided dual-branch GANs for learning different and complement characteristics to make the network pay more attention to the main shape, details, and noise.•Develop a feature transmission module and a fusion block to accomplish feature learning between two branches.