Threshold Matters in WSSS: Manipulating the Activation for the Robust and Accurate Segmentation Model Against Thresholds

Benchmark Model Rank Results
weakly-supervised-semantic-segmentation-onAMN (DeepLabV2-ResNet101, MS-COCO-pretrained weights)#35Mean IoU: 70.7
weakly-supervised-semantic-segmentation-onAMN (DeepLabV2-ResNet101)#47Mean IoU: 69.5
weakly-supervised-semantic-segmentation-on-1AMN (DeepLabV2-ResNet101, MS-COCO-pretrained weights)#39Mean IoU: 70.6
weakly-supervised-semantic-segmentation-on-1AMN (DeepLabV2-ResNet101)#47Mean IoU: 69.6
weakly-supervised-semantic-segmentation-on-4AMN (DeepLabV2-ResNet101)#18mIoU: 44.7