Max Pooling with Vision Transformers reconciles class and shape in weakly supervised semantic segmentation

Benchmark Model Rank Results
weakly-supervised-semantic-segmentation-on-pascal-voc-2012-trainViT-PCM#4Mean IoU: 71.4
weakly-supervised-semantic-segmentation-on-coco-2014-valViT-PCM#16mIoU: 45.0
weakly-supervised-semantic-segmentation-on-pascal-voc-2012-testViT-PCM#30Mean IoU: 70.9
weakly-supervised-semantic-segmentation-on-pascal-voc-2012-valViT-PCM#40Mean IoU: 70.3