上一条: Zero-Shot Learning via Category-Specific Visual-Semantic Mapping and Label Refinement
下一条: Inharmonious Region Localization