CorrMatch: Label Propagation via Correlation Matching for Semi-Supervised Semantic Segmentation

Benchmark Model Rank Results
semi-supervised-semantic-segmentation-on-1CorrMatch (Deeplabv3+ with ResNet-101)#7Validation mIoU: 79.4%
semi-supervised-semantic-segmentation-on-10CorrMatch (Deeplabv3+ with ResNet-101)#6Validation mIoU: 81.8
semi-supervised-semantic-segmentation-on-2CorrMatch (Deeplabv3+ with ResNet-101)#4Validation mIoU: 78.5%
semi-supervised-semantic-segmentation-on-21CorrMatch (Deeplabv3+ with ResNet-101)#5Validation mIoU: 81.3
semi-supervised-semantic-segmentation-on-22CorrMatch (Deeplabv3+ with ResNet-101)#4Validation mIoU: 77.3
semi-supervised-semantic-segmentation-on-27CorrMatch (Deeplabv3+ with ResNet-101)#8Validation mIoU: 76.4
semi-supervised-semantic-segmentation-on-28CorrMatch (Deeplabv3+ with ResNet-101)#7Validation mIoU: 78.5
semi-supervised-semantic-segmentation-on-29CorrMatch (Deeplabv3+ with ResNet-101)#8Validation mIoU: 79.4
semi-supervised-semantic-segmentation-on-30CorrMatch (Deeplabv3+ with ResNet-101)#7Validation mIoU: 80.6
semi-supervised-semantic-segmentation-on-4CorrMatch (Deeplabv3+ with ResNet-101)#5Validation mIoU: 81.9%
semi-supervised-semantic-segmentation-on-8CorrMatch (Deeplabv3+ with ResNet-101)#5Validation mIoU: 80.4%
semi-supervised-semantic-segmentation-on-9CorrMatch (Deeplabv3+ with ResNet-101)#5Validation mIoU: 80.9