| DMT (DeepLab v2 MSCOCO/ImageNet pre-trained) | 63.03% | DMT: Dynamic Mutual Training for Semi-Supervised Learning | |
| UniMatch V2 (DINOv2-B) | 84.3% | UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation | |
| PrevMatch (ResNet-101) | 78.9% | Revisiting and Maximizing Temporal Knowledge in Semi-supervised Semantic Segmentation | |
| UniMatch (DeepLab v3+ with ImageNet-pretrained ResNet-101, single scale inference) | 77.92% | Revisiting Weak-to-Strong Consistency in Semi-Supervised Semantic Segmentation | |
| U2PL (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K, AEL) | 76.48% | Semi-Supervised Semantic Segmentation Using Unreliable Pseudo-Labels | |
| Dense FixMatch (DeepLabv3+ ResNet-101, uniform sampling, single pass eval) | 73.91% | Dense FixMatch: a simple semi-supervised learning method for pixel-wise prediction tasks | |
| Error Localization Network (DeeplabV3 with ResNet-50) | 70.33% | Semi-supervised Semantic Segmentation with Error Localization Network | |
| AEL (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K) | 77.9% | Semi-Supervised Semantic Segmentation via Adaptive Equalization Learning | |
| SimpleBaseline(DeeplabV3+ with ImageNet pretrained Xception65, sinle scale inference) | 74.1% | A Simple Baseline for Semi-supervised Semantic Segmentation with Strong Data Augmentation | |
| CutMix (DeepLab v2, ImageNet pre-trained) | 60.34% | Semi-supervised semantic segmentation needs strong, varied perturbations | |
| CW-BASS (DeepLab v3+ with ResNet-50) | 77.20% | Confidence-Weighted Boundary-Aware Learning for Semi-Supervised Semantic Segmentation | - |
| PS-MT (DeepLab v3+ with ImageNet-pretrained ResNet50, single scale inference) | 77.12% | Perturbed and Strict Mean Teachers for Semi-supervised Semantic Segmentation | |
| LaserMix (DeepLab v3+, ImageNet pre-trained ResNet50, single scale inference) | 77.1% | LaserMix for Semi-Supervised LiDAR Semantic Segmentation | |
| GIST and RIST (DeepLabv2 with ResNet101, MSCOCO pre-trained) | 62.57% | The GIST and RIST of Iterative Self-Training for Semi-Supervised Segmentation | - |
| CPS (DeepLab v3+ with ImageNet-pretrained ResNet-101, single scale inference) | 77.62% | Semi-Supervised Semantic Segmentation with Cross Pseudo Supervision | |
| ClassMix (DeepLab v2 MSCOCO pretrained) | 61.35% | ClassMix: Segmentation-Based Data Augmentation for Semi-Supervised Learning | |
| Dense FixMatch (DeepLabv3+ ResNet-50, uniform sampling, single pass eval) | 73.39% | Dense FixMatch: a simple semi-supervised learning method for pixel-wise prediction tasks | |
| SemiVL (ViT-B/16) | 79.4% | SemiVL: Semi-Supervised Semantic Segmentation with Vision-Language Guidance | |
| CPCL (DeepLab v3+ with ResNet-50) | 74.6% | Conservative-Progressive Collaborative Learning for Semi-supervised Semantic Segmentation | |
| s4GAN (DeepLab v2 ImageNet pre-trained) | 59.3% | Semi-Supervised Semantic Segmentation with High- and Low-level Consistency | |