Multichannel Cross-Scale Semantic Coherent Attention Network for Image Inpainting

Figure 5

L1 error result for different setting.