Semantic Segmentation on LoveDA (test)
55.5mIoUBiCoR-Seg
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| BiCoR-Seg2025.12 | 55.5 | 48.1 | 60.9 | 58.7 | 80.8 | 24.4 | 48.7 | 66.8 | — | — | — | |
| SFA-NetVenue=RS 20242025.12 | 54.9 | 48.4 | 60.3 | 59.1 | 81.9 | 24.1 | 46.2 | 64 | — | — | — | |
| Full-tuning# P (M)=86.82026.02 | 54.8 | — | — | — | — | — | — | — | — | — | — | |
| AFENetVenue=TGRS 20252025.12 | 54.8 | 47.4 | 59.2 | 59.2 | 81.6 | 21.4 | 48.7 | 66.4 | — | — | — | |
| AdaRoute# P (M)=5.22026.02 | 54.3 | — | — | — | — | — | — | — | — | — | — | |
| DMA-NetVenue=RS 20252025.12 | 54.2 | 49.3 | 60.1 | 58.7 | 82.3 | 17.3 | 47.1 | 64.8 | — | — | — | |
| LSKNet-S2024.03 | 54 | 46.7 | 59.9 | 58.3 | 80.2 | 24.6 | 46.4 | 61.8 | — | — | — | |
| Mona# P (M)=5.22026.02 | 53.8 | — | — | — | — | — | — | — | — | — | — | |
| LWGANet-L2#P (M)=12.58, Input size=1024x1024, GPU=RTX 30902025.01 | 53.6 | — | — | — | — | — | — | — | — | — | — | |
| LOGCAN++Backbone=ResNet-502024.06 | 53.35 | 47.37 | 58.38 | 56.46 | 80.05 | 18.44 | 47.91 | 64.8 | — | — | — | |
| LSKNet-T2024.03 | 53.2 | 46.4 | 59.5 | 57.1 | 79.9 | 21.8 | 46.6 | 61.4 | — | — | — | |
| LoRand# P (M)=5.92026.02 | 53 | — | — | — | — | — | — | — | — | — | — | |
| SNELL# P (M)=5.82026.02 | 52.9 | — | — | — | — | — | — | — | — | — | — | |
| AdaptFormer# P (M)=5.42026.02 | 52.8 | — | — | — | — | — | — | — | — | — | — | |
| Hi-ResNetBackbone=Hi-ResNet, Pretrain=Y, FLOPS(G)=402, FPS=35, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 52.6 | 46.8 | 58.3 | 55.9 | 80.1 | 17 | 46.7 | 62.7 | — | — | — | |
| RepAdapter# P (M)=5.82026.02 | 52.6 | — | — | — | — | — | — | — | — | — | — | |
| DDPBackbone=Vit-B2024.06 | 52.54 | 46.19 | 57.2 | 58.22 | 80.34 | 14.93 | 46.54 | 64.34 | — | — | — | |
| Hi-ResNet2024.03 | 52.5 | 46.7 | 58.3 | 55.9 | 80.1 | 17 | 46.7 | 62.7 | — | — | — | |
| Hi-ResNetVenue=JSTARS 20242025.12 | 52.5 | 46.7 | 58.3 | 55.9 | 80.1 | 17 | 46.7 | 62.7 | — | — | — | |
| OCRNetBackbone=ResNet-502024.06 | 52.46 | 46.05 | 58.66 | 57.17 | 80.72 | 13.27 | 46.83 | 64.53 | — | — | — | |
| UNetFormer2024.03 | 52.4 | 44.7 | 58.8 | 54.9 | 79.6 | 20.1 | 46 | 62.5 | — | — | — | |
| UperNetBackbone=ViTAE-B + RVSA, Pretrain=Y, FLOPS(G)=413, FPS=11, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 52.4 | 46.7 | 58.1 | 57.1 | 79.6 | 16.5 | 46.4 | 62.4 | — | — | — | |
| AerialFormer-SBackbone=AerialFormer, Pretrain=Y, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 52.4 | 46.6 | 57.4 | 57.3 | 80.5 | 15.6 | 46.8 | 62.8 | — | — | — | |
| UnetFormer#P (M)=11.73, Input size=1024x1024, GPU=RTX 30902025.01 | 52.4 | — | — | — | — | — | — | — | — | — | — | |
| RSSFormer#P (M)=30.82, Input size=1024x1024, GPU=RTX 30902025.01 | 52.4 | — | — | — | — | — | — | — | — | — | — | |
| VPT# P (M)=0.12026.02 | 52.4 | — | — | — | — | — | — | — | — | — | — | |
| UNetFormerVenue=ISPRS 20222025.12 | 52.4 | 44.7 | 58.8 | 54.9 | 79.6 | 20.1 | 46 | 62.5 | — | — | — | |
| RSSFormerVenue=TIP 20232025.12 | 52.4 | 52.4 | 60.7 | 55.2 | 76.3 | 18.7 | 45.4 | 58.3 | — | — | — | |
| PSPNetBackbone=ResNet-502024.06 | 52.31 | 45.81 | 58.21 | 58.37 | 80.23 | 14.19 | 46.55 | 62.79 | — | — | — | |
| RSSFormerBackbone=RSS-B, Pretrain=Y, FLOPS(G)=413, FPS=6, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 52.3 | 52.3 | 60.7 | 55.2 | 76.2 | 18.7 | 45.3 | 58.3 | — | — | — | |
| LoveNAS#P (M)=30.49, Input size=1024x1024, GPU=RTX 30902025.01 | 52.3 | — | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=ResNet-502024.06 | 52.22 | 46.44 | 56.84 | 57.03 | 79.31 | 14.65 | 46.59 | 64.66 | — | — | — | |
| Semantic FPNBackbone=ResNet-502024.06 | 52.19 | 45.45 | 56.84 | 57.54 | 79.43 | 16.78 | 45.89 | 63.43 | — | — | — | |
| LoRA# P (M)=5.42026.02 | 52.1 | — | — | — | — | — | — | — | — | — | — | |
| UperNetBackbone=VIT-B + RVSA, Pretrain=Y, FLOPS(G)=407, FPS=19, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 51.9 | 45.2 | 59.8 | 55.2 | 79.4 | 18.4 | 46.2 | 59.2 | — | — | — | |
| FarSegBackbone=ResNet-502024.06 | 51.78 | 45.18 | 56.55 | 56.92 | 79.78 | 14.96 | 46.59 | 62.5 | — | — | — | |
| DeepLabV3+Backbone=ResNet-502024.06 | 51.18 | 45.31 | 55.92 | 57.09 | 78.33 | 11.82 | 45.87 | 63.89 | — | — | — | |
| UnetFormerBackbone=ResNet-502024.06 | 50.9 | 44.16 | 56.51 | 55.84 | 79 | 13.86 | 44.27 | 62.65 | — | — | — | |
| DC-Swin2024.03 | 50.6 | 41.3 | 54.5 | 56.2 | 78.1 | 14.5 | 47.2 | 62.4 | — | — | — | |
| DC-SwinBackbone=Swin-Tiny, Pretrain=Y, FLOPS(G)=107, FPS=60, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 50.6 | 41.3 | 54.5 | 56.2 | 78.1 | 14.5 | 47.2 | 62.4 | — | — | — | |
| RSSFormerBackbone=RSS-B2024.06 | 50.59 | 44.03 | 57.23 | 55.35 | 79.37 | 10.91 | 45.22 | 62.04 | — | — | — | |
| FarSeg#P (M)=31.37, Input size=1024x1024, GPU=RTX 30902025.01 | 50.1 | — | — | — | — | — | — | — | — | — | — | |
| SwinUperNet2024.03 | 50 | 43.3 | 54.3 | 54.3 | 78.7 | 14.9 | 45.3 | 59.6 | — | — | — | |
| SwinUperNetBackbone=Swin-Tiny, Pretrain=Y, FLOPS(G)=349, FPS=19, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 50 | 43.3 | 54.3 | 54.3 | 78.7 | 14.9 | 45.3 | 59.6 | — | — | — | |
| FactSeg#P (M)=33.44, Input size=1024x1024, GPU=RTX 30902025.01 | 50 | — | — | — | — | — | — | — | — | — | — | |
| HRNet2024.03 | 49.8 | 44.6 | 55.3 | 57.4 | 78 | 11 | 45.3 | 60.9 | — | — | — | |
| HRNet(Baseline)Backbone=HRNet-W48, Pretrain=Y, FLOPS(G)=559, FPS=25, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 49.8 | 44.6 | 55.3 | 57.4 | 78 | 11 | 45.3 | 60.9 | — | — | — | |
| BANet2024.03 | 49.6 | 43.7 | 51.5 | 51.1 | 76.9 | 16.6 | 44.9 | 62.5 | — | — | — | |
| BANetBackbone=ResT-Lite, Pretrain=Y, FLOPS(G)=67, FPS=84, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 49.6 | 43.7 | 51.5 | 51.1 | 76.9 | 16.6 | 44.9 | 62.5 | — | — | — | |
| BANetVenue=GRSL 20222025.12 | 49.6 | 43.7 | 51.5 | 51.1 | 76.9 | 16.6 | 44.9 | 62.5 | — | — | — | |
| SegFormerVenue=NeurIPS 20212025.12 | 49.1 | 42.2 | 56.4 | 50.7 | 78.5 | 17.2 | 45.2 | 53.8 | — | — | — | |
| FactSeg2024.03 | 48.9 | 42.6 | 53.6 | 52.8 | 76.9 | 16.2 | 42.9 | 57.5 | — | — | — | |
| TransUNet2024.03 | 48.9 | 43 | 56.1 | 53.7 | 78 | 9.3 | 44.9 | 56.9 | — | — | — | |
| TransUNetBackbone=ViT-R50, Pretrain=Y, FLOPS(G)=803, FPS=13, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 48.9 | 43 | 56.1 | 53.7 | 78 | 9.3 | 44.9 | 56.9 | — | — | — | |
| FactSegBackbone=ResNet50, Pretrain=Y, FLOPS(G)=267, FPS=46, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 48.9 | 42.6 | 53.6 | 52.8 | 76.9 | 16.2 | 42.9 | 57.5 | — | — | — | |
| TransUNetVenue=MIA 20242025.12 | 48.9 | 43 | 56.1 | 53.7 | 78 | 9.3 | 44.9 | 56.9 | — | — | — | |
| LinkNetVenue=CVPR 20172025.12 | 48.5 | 43.6 | 52.1 | 52.5 | 76.9 | 12.2 | 45.1 | 57.3 | — | — | — | |
| PSPNet2024.03 | 48.3 | 44.4 | 52.1 | 53.5 | 76.5 | 9.7 | 44.1 | 57.9 | — | — | — | |
| PSPNetBackbone=ResNet50, Pretrain=Y, FLOPS(G)=738, FPS=27, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 48.3 | 44.4 | 52.1 | 53.5 | 76.5 | 9.7 | 44.1 | 57.9 | — | — | — | |
| PSPNetVenue=ISPRS 20162025.12 | 48.3 | 44.4 | 52.1 | 53.5 | 76.5 | 9.7 | 44.1 | 57.9 | — | — | — | |
| UNet++2024.03 | 48.2 | 42.9 | 52.6 | 52.8 | 74.5 | 11.4 | 44.4 | 58.8 | — | — | — | |
| SemanticFPN2024.03 | 48.2 | 42.9 | 51.5 | 53.4 | 74.7 | 11.2 | 44.6 | 58.7 | — | — | — | |
| FarSeg2024.03 | 48.2 | 43.1 | 51.5 | 53.9 | 76.6 | 9.8 | 43.3 | 58.9 | — | — | — | |
| SemanticFPNBackbone=ResNet50, Pretrain=Y, FLOPS(G)=589, FPS=37, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 48.2 | 42.9 | 51.5 | 53.4 | 74.7 | 11.2 | 44.6 | 58.7 | — | — | — | |
| FarSegBackbone=ResNet50, Pretrain=Y, FLOPS(G)=350, FPS=47, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 48.2 | 43.1 | 51.5 | 53.9 | 76.6 | 9.8 | 43.3 | 58.9 | — | — | — | |
| UNet++Venue=MICCAI 20182025.12 | 48.2 | 42.9 | 52.6 | 52.8 | 74.5 | 11.4 | 44.4 | 58.8 | — | — | — | |
| SemanticFPNVenue=CVPR 20192025.12 | 48.2 | 42 | 51.5 | 53.4 | 74.7 | 11.2 | 44.6 | 58.7 | — | — | — | |
| FarSegVenue=CVPR 20202025.12 | 48.2 | 43.4 | 51.8 | 53.3 | 76.1 | 10.8 | 43.2 | 58.6 | — | — | — | |
| UNet++Backbone=ResNet50, Pretrain=Y, FLOPS(G)=544, FPS=30, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 48.1 | 42.8 | 52.6 | 52.8 | 74.5 | 11.4 | 44.4 | 58.8 | — | — | — | |
| DC-SwinBackbone=Swin-B2024.06 | 47.81 | 42.34 | 50.17 | 54.83 | 77.13 | 8.16 | 42.76 | 59.28 | — | — | — | |
| U-NetVenue=MICCAI 20152025.12 | 47.8 | 43.1 | 52.7 | 52.8 | 73.1 | 10.3 | 43 | 59.9 | — | — | — | |
| DeepLabV3+2024.03 | 47.6 | 43 | 50.9 | 52 | 74.4 | 10.4 | 44.2 | 58.5 | — | — | — | |
| UNet2024.03 | 47.6 | 43.1 | 52.7 | 52.8 | 73 | 10.3 | 43.1 | 59.9 | — | — | — | |
| DeepLabV3+Backbone=ResNet50, Pretrain=Y, FLOPS(G)=355, FPS=46, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 47.6 | 43 | 50.9 | 52 | 74.4 | 10.4 | 44.2 | 58.5 | — | — | — | |
| DeepLabv3+Venue=ECCV 20182025.12 | 47.6 | 43 | 50.9 | 52 | 74.4 | 10.4 | 44.2 | 58.5 | — | — | — | |
| SegFormer2024.03 | 47.4 | 43.1 | 52.3 | 55 | 70.7 | 10.7 | 43.2 | 56.8 | — | — | — | |
| EfficientViTBackbone=EfficientViT-L22024.06 | 47.12 | 42.92 | 50.98 | 52.78 | 75.65 | 4.27 | 42.04 | 61.19 | — | — | — | |
| Segmenter2024.03 | 47.1 | 38 | 50.7 | 48.7 | 77.4 | 13.3 | 43.5 | 58.2 | — | — | — | |
| SegmenterBackbone=ViT-Tiny, Pretrain=Y, FLOPS(G)=26, FPS=61, Input size=1024x1024, Hardware=NVIDIA GTX 30902023.05 | 47.1 | 38 | 50.7 | 48.7 | 77.4 | 13.3 | 43.5 | 58.2 | — | — | — | |
| FCNVenue=CVPR 20152025.12 | 46.7 | 42.6 | 49.5 | 48.1 | 73.1 | 11.8 | 43.5 | 58.3 | — | — | — | |
| SegmenterBackbone=Swin-B2024.06 | 43.82 | 39.7 | 44.71 | 47.29 | 73.51 | 8.35 | 37.45 | 55.71 | — | — | — | |
| SeeCo+ProxyCLIPLearning Strategy=Training-free2026.04 | 40.3 | — | — | — | — | — | — | — | — | — | — | |
| SeeCo+SegEarth-OVLearning Strategy=Training-free2026.04 | 39.4 | — | — | — | — | — | — | — | — | — | — | |
| CorrCLIPLearning Strategy=Training-free2026.04 | 36.9 | — | — | — | — | — | — | — | — | — | — | |
| SegEarth-OVLearning Strategy=Training-free2026.04 | 36.9 | — | — | — | — | — | — | — | — | — | — | |
| SeeCo+ClearCLIPLearning Strategy=Training-free2026.04 | 36.6 | — | — | — | — | — | — | — | — | — | — | |
| ProxyCLIPLearning Strategy=Training-free2026.04 | 36.4 | — | — | — | — | — | — | — | — | — | — | |
| GEMLearning Strategy=Training-free2026.04 | 31.6 | — | — | — | — | — | — | — | — | — | — | |
| ClearCLIPLearning Strategy=Training-free2026.04 | 31 | — | — | — | — | — | — | — | — | — | — | |
| SCLIPLearning Strategy=Training-free2026.04 | 30.4 | — | — | — | — | — | — | — | — | — | — | |
| MaskCLIPLearning Strategy=Training-free2026.04 | 27.8 | — | — | — | — | — | — | — | — | — | — | |
| CLIPLearning Strategy=Training-free2026.04 | 12.4 | — | — | — | — | — | — | — | — | — | — | |
| AerialFormerInput size=512×5122025.03 | — | — | — | — | — | — | — | — | 113.8 | 26.9 | 126.8 | |
| Dynamic Dictionary LearningInput size=512×5122025.03 | — | — | — | — | — | — | — | — | 90.9 | 3.3 | 89.2 | |
| KTDAInput size=512×5122025.03 | — | — | — | — | — | — | — | — | 258.3 | 170.7 | 566.4 | |
| SFA-NetInput size=512×5122025.03 | — | — | — | — | — | — | — | — | 10.7 | 0.6 | 7.1 |