Learning Pseudo Labels for Semi-and-Weakly Supervised Semantic Segmentation

Benchmark Model Rank Results
semi-supervised-semantic-segmentation-on-15PCT (DeepLab v3+ with ResNet-50 pretrained on ImageNet-1K)#8Validation mIoU: 77.26%
semi-supervised-semantic-segmentation-on-21PCT (DeepLab v3+ with ResNet-50 pretrained on ImageNet-1K)#15Validation mIoU: 71.35
semi-supervised-semantic-segmentation-on-4PCT (DeepLab v3+ with ResNet-50 pretrained on ImageNet-1K)#19Validation mIoU: 75.52%
semi-supervised-semantic-segmentation-on-9PCT (DeepLab v3+ with ResNet-50 pretrained on ImageNet-1K)#18Validation mIoU: 76.47