| Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval) | 74.73% | Dense FixMatch: a simple semi-supervised learning method for pixel-wise prediction tasks | |
| PCR (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K) | 80.91% | Semi-supervised Semantic Segmentation with Prototype-based Consistency Regularization | |
| U2PL (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K, CutMix) | 80.5% | Semi-Supervised Semantic Segmentation Using Unreliable Pseudo-Labels | |
| GuidedMix-Net(DeepLab v2 with ResNet101, ImageNet pretrained) | 76.5% | GuidedMix-Net: Learning to Improve Pseudo Masks Using Labeled Images as Reference | |
| PCT (DeepLab v3+ with ResNet-50 pretrained on ImageNet-1K) | 77.26% | Learning Pseudo Labels for Semi-and-Weakly Supervised Semantic Segmentation | |
| PS-MT
(DeepLab v3+ with ImageNet-pretrained ResNet-101, single scale inference) | 79.76% | Perturbed and Strict Mean Teachers for Semi-supervised Semantic Segmentation | |
| Dense FixMatch (DeepLabv3+ ResNet-50, over-sampling, single pass eval) | 71.69% | Dense FixMatch: a simple semi-supervised learning method for pixel-wise prediction tasks | |
| GuidedMix-Net(DeepLab v2 with ResNet101, input-size: 512x512 with multi-scale and flip, ImageNet pretrained) | 78.2% | GuidedMix-Net: Learning to Improve Pseudo Masks Using Labeled Images as Reference | |
| AEL (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K) | 80.29% | Semi-Supervised Semantic Segmentation via Adaptive Equalization Learning | |