Pseudo-mask Matters in Weakly-supervised Semantic Segmentation

Open paper
Benchmark Model Rank Results
weakly-supervised-semantic-segmentation-onPMM(Res2Net101, no saliency, no RW)#45Mean IoU: 70.0
weakly-supervised-semantic-segmentation-onPMM(ResNet38)#54Mean IoU: 68.5
weakly-supervised-semantic-segmentation-onPMM(ResNet38, no saliency, no RW)#55Mean IoU: 68.5
weakly-supervised-semantic-segmentation-on-1PMM(Res2Net101, no saliency, no RW)#41Mean IoU: 70.5
weakly-supervised-semantic-segmentation-on-1PMM(ResNet38, no saliency, no RW)#50Mean IoU: 69.0
weakly-supervised-semantic-segmentation-on-4PMM(ScaleNet101, no saliency, no RW)#28mIoU: 40.2
weakly-supervised-semantic-segmentation-on-4PMM(ResNet38, no saliency, no RW)#30mIoU: 36.7

Rank counts only results with a code link.