1 paper
Zhaotong Yang, Yi Chen, Yanying Li +5
Recent deep models for image shadow removal often rely on attention-based architectures to capture long-range dependencies. However, their fixed attention patterns tend to mix illu…