Search for a command to run...
Max Pooling with Vision Transformers reconciles class and shape in weakly supervised semantic segmentation