Few-Shot Segmentation on Pascal-5i (mIoU by support size)
75.9mIoU (5^0)TLG
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| TLGBackbone=Resnet50, A. Type=I, Shots=5-shot2025.08 | 75.9 | 82.6 | 77.7 | 75.3 | 77.9 | — | |
| TLGBackbone=Resnet50, A. Type=I, Shots=1-shot2025.08 | 75.4 | 81.8 | 77.4 | 74 | 77.2 | — | |
| PGMA-NetBackbone=Resnet50, A. Type=P, Shots=5-shot2025.08 | 74 | 81.5 | 71.9 | 73.3 | 75.2 | — | |
| VRP-SAMBackbone=Resnet50, A. Type=P, Shots=1-shot2025.08 | 73.9 | 78.3 | 70.6 | 65 | 71.9 | — | |
| PGMA-NetBackbone=Resnet50, A. Type=P, Shots=1-shot2025.08 | 73.4 | 80.8 | 70.5 | 71.7 | 74.1 | — | |
| TLGBackbone=VGG16, A. Type=I, Shots=5-shot2025.08 | 73 | 82.4 | 78.1 | 73.7 | 76.8 | — | |
| TLGBackbone=VGG16, A. Type=I, Shots=1-shot2025.08 | 72.5 | 81.5 | 77.5 | 72.8 | 76.1 | — | |
| HSNetBackbone=ResNet-100, Shots=52022.11 | 71.8 | 74.4 | 67 | 68.3 | 70.4 | — | |
| HDMNetBackbone=Resnet50, A. Type=P, Shots=5-shot2025.08 | 71.3 | 76.2 | 71.3 | 68.5 | 71.8 | — | |
| HDMNetBackbone=Resnet50, A. Type=P, Shots=1-shot2025.08 | 71 | 75.4 | 68.9 | 62.1 | 69.4 | — | |
| HSNetBackbone=ResNet-50, Shots=52022.11 | 70.3 | 73.2 | 67.4 | 67.1 | 69.5 | — | |
| IPRNetBackbone=ResNet-50, Shots=52022.11 | 70.2 | 75.6 | 68.9 | 66.2 | 70.2 | — | |
| IPRNetBackbone=ResNet-100, Shots=52022.11 | 70 | 75.9 | 71.8 | 65.8 | 70.9 | — | |
| DRNetBackbone=Resnet50, A. Type=P, Shots=5-shot2025.08 | 69.2 | 73.9 | 65.4 | 65.3 | 68.5 | — | |
| AFANetBackbone=Resnet50, A. Type=I, Shots=5-shot2025.08 | 69 | 70.4 | 61.3 | 64 | 66.2 | — | |
| DRNetBackbone=VGG16, A. Type=P, Shots=5-shot2025.08 | 68.9 | 68.2 | 64.6 | 57.1 | 64.7 | — | |
| HDMNetBackbone=VGG16, A. Type=P, Shots=5-shot2025.08 | 68.1 | 73.1 | 71.8 | 64 | 69.3 | — | |
| IPRNetBackbone=ResNet-100, Shots=12022.11 | 67.8 | 74.6 | 65.7 | 62.2 | 67.5 | — | |
| HSNetBackbone=ResNet-100, Shots=12022.11 | 67.3 | 72.3 | 62 | 63.1 | 66.2 | — | |
| AFANetBackbone=VGG16, A. Type=I, Shots=5-shot2025.08 | 66.3 | 68.7 | 61.6 | 59.7 | 64.1 | — | |
| DRNetBackbone=Resnet50, A. Type=P, Shots=1-shot2025.08 | 66.1 | 68.8 | 61.3 | 58.2 | 63.6 | — | |
| MLCBackbone=ResNet-100, Shots=52022.11 | 65.8 | 74.9 | 71.4 | 63.1 | 68.8 | — | |
| AFANetBackbone=Resnet50, A. Type=I, Shots=1-shot2025.08 | 65.7 | 68.5 | 60.6 | 61.5 | 64 | — | |
| DRNetBackbone=VGG16, A. Type=P, Shots=1-shot2025.08 | 65.4 | 66.2 | 54.2 | 53.8 | 59.9 | — | |
| IPRNetBackbone=ResNet-50, Shots=12022.11 | 65.2 | 72.9 | 63.3 | 61.3 | 65.7 | — | |
| PFENet++Backbone=Resnet50, A. Type=P, Shots=5-shot2025.08 | 65.2 | 73.6 | 74.1 | 65.3 | 69.6 | — | |
| HDMNetBackbone=VGG16, A. Type=P, Shots=1-shot2025.08 | 64.8 | 71.4 | 67.7 | 56.4 | 65.1 | — | |
| ASGNetBackbone=ResNet-100, Shots=52022.11 | 64.6 | 71.3 | 64.2 | 57.3 | 64.4 | — | |
| HSNetBackbone=ResNet-50, Shots=12022.11 | 64.3 | 70.7 | 60.3 | 60.5 | 64 | — | |
| PFENet++Backbone=VGG16, A. Type=P, Shots=5-shot2025.08 | 64.3 | 72 | 70 | 62.7 | 67.3 | — | |
| AFANetBackbone=VGG16, A. Type=I, Shots=1-shot2025.08 | 64.1 | 65.9 | 59.2 | 56.9 | 61.5 | — | |
| ASGNetBackbone=ResNet-50, Shots=52022.11 | 63.7 | 70.6 | 64.2 | 57.4 | 63.9 | — | |
| IMR-HSNetBackbone=Resnet50, A. Type=I, Shots=5-shot2025.08 | 63.6 | 69.6 | 56.3 | 57.4 | 61.8 | — | |
| MLCBackbone=ResNet-50, Shots=52022.11 | 63.5 | 71.6 | 71.2 | 58.1 | 66.1 | — | |
| PFENetBackbone=ResNet-50, Shots=52022.11 | 63.1 | 70.7 | 55.8 | 57.9 | 61.9 | — | |
| PFENetBackbone=ResNet-100, Shots=52022.11 | 62.8 | 70.4 | 54.9 | 57.6 | 61.4 | — | |
| IMR-HSNetBackbone=Resnet50, A. Type=I, Shots=1-shot2025.08 | 62.6 | 69.1 | 56.1 | 56.7 | 61 | — | |
| PFENetBackbone=ResNet-50, Shots=12022.11 | 61.7 | 69.5 | 55.4 | 56.3 | 60.8 | — | |
| MLCBackbone=ResNet-100, Shots=12022.11 | 60.8 | 71.3 | 61.5 | 56.9 | 62.6 | — | |
| PFENet++Backbone=Resnet50, A. Type=P, Shots=1-shot2025.08 | 60.6 | 70.3 | 65.6 | 60.3 | 64.2 | — | |
| PFENetBackbone=ResNet-100, Shots=12022.11 | 60.5 | 69.4 | 54.4 | 55.9 | 60.1 | — | |
| IMR-HSNetBackbone=VGG16, A. Type=I, Shots=5-shot2025.08 | 60.5 | 65.3 | 55 | 51.7 | 58.1 | — | |
| PPNetBackbone=ResNet-100, Shots=52022.11 | 60.3 | 70 | 69.4 | 60.7 | 65.1 | — | |
| ASGNetBackbone=ResNet-100, Shots=12022.11 | 59.8 | 67.4 | 55.6 | 54.4 | 59.3 | — | |
| ASNetBackbone=ResNet50, N-way=1-way, K-shot=5-shot2022.03 | 59.2 | 63.5 | 41.2 | 58.7 | 55.7 | — | |
| MLCBackbone=ResNet-50, Shots=12022.11 | 59.2 | 71.2 | 65.6 | 52.5 | 62.1 | — | |
| PFENet++Backbone=VGG16, A. Type=P, Shots=1-shot2025.08 | 59.2 | 69.6 | 66.8 | 60.7 | 64.1 | — | |
| PPNetBackbone=ResNet-50, Shots=52022.11 | 58.9 | 68.3 | 66.8 | 58 | 63 | — | |
| ASGNetBackbone=ResNet-50, Shots=12022.11 | 58.8 | 67.9 | 56.8 | 53.7 | 59.3 | — | |
| IMR-HSNetBackbone=VGG16, A. Type=I, Shots=1-shot2025.08 | 58.2 | 63.9 | 52.9 | 51.2 | 56.5 | — | |
| Zhang et al.Backbone=Resnet50, A. Type=I, Shots=1-shot2025.08 | 56.9 | 62.5 | 60.3 | 49.9 | 57.4 | — | |
| HSNetBackbone=ResNet50, N-way=1-way, K-shot=5-shot2022.03 | 56.2 | 61.3 | 40.2 | 54.2 | 53 | — | |
| PANetBackbone=ResNet-50, Shots=52022.11 | 55.3 | 67.2 | 61.3 | 53.2 | 59.3 | — | |
| FWBBackbone=ResNet-100, Shots=52022.11 | 54.8 | 67.4 | 62.2 | 55.3 | 59.9 | — | |
| ASNetBackbone=ResNet50, N-way=2-way, K-shot=5-shot2022.03 | 53.4 | 60.4 | 35.9 | 50.6 | 50.1 | — | |
| PPNetBackbone=ResNet-100, Shots=12022.11 | 52.7 | 62.8 | 57.4 | 47.7 | 55.2 | — | |
| ASNetN-way=1-way, K-shot=1-shot, Supervision=Strong labels, Backbone=ResNet-50, Framework=iFSL2022.03 | 51.7 | 61.5 | 43.3 | 52.8 | 52.3 | — | |
| FWBBackbone=ResNet-100, Shots=12022.11 | 51.3 | 64.5 | 56.7 | 52.2 | 56.2 | — | |
| Siam et al.Backbone=Resnet50, A. Type=I, Shots=1-shot2025.08 | 49.5 | 65.5 | 50 | 49.2 | 53.5 | — | |
| HSNetN-way=1-way, K-shot=1-shot, Supervision=Strong labels, Backbone=ResNet-50, Framework=iFSL2022.03 | 49.1 | 59.7 | 41 | 49 | 49.7 | — | |
| PPNetBackbone=ResNet-50, Shots=12022.11 | 48.6 | 60.6 | 55.7 | 46.5 | 52.8 | — | |
| ASNetN-way=2-way, K-shot=1-shot, Supervision=Strong labels, Backbone=ResNet-50, Framework=iFSL2022.03 | 48.5 | 58.3 | 36.3 | 48.3 | 47.8 | — | |
| PANetBackbone=ResNet50, N-way=2-way, K-shot=5-shot2022.03 | 46.2 | 57.4 | 46.7 | 47.6 | 49.5 | — | |
| PANetBackbone=ResNet50, N-way=1-way, K-shot=5-shot2022.03 | 45.6 | 56.2 | 44.6 | 49.2 | 48.9 | — | |
| PANetBackbone=ResNet-50, Shots=12022.11 | 44 | 57.5 | 50.8 | 44 | 49.1 | — | |
| PFENetBackbone=ResNet50, N-way=1-way, K-shot=5-shot2022.03 | 42.8 | 56.3 | 36.2 | 47.3 | 45.7 | — | |
| HSNetBackbone=ResNet50, N-way=2-way, K-shot=5-shot2022.03 | 42.5 | 58.9 | 32 | 44.1 | 44.4 | — | |
| HSNetN-way=2-way, K-shot=1-shot, Supervision=Strong labels, Backbone=ResNet-50, Framework=iFSL2022.03 | 42.4 | 53.7 | 34 | 43.9 | 43.5 | — | |
| MIAPNetBackbone=Resnet50, A. Type=I, Shots=1-shot2025.08 | 41.7 | 51.3 | 42.2 | 41.8 | 44.2 | — | |
| PFENetN-way=1-way, K-shot=1-shot, Supervision=Strong labels, Backbone=ResNet-50, Framework=iFSL2022.03 | 38.3 | 54.7 | 35.1 | 43.8 | 43 | — | |
| Pix-MetaNetBackbone=VGG16, A. Type=I, Shots=1-shot2025.08 | 36.5 | 51.7 | 45.9 | 35.6 | 42.4 | — | |
| PFENetBackbone=ResNet50, N-way=2-way, K-shot=5-shot2022.03 | 35.9 | 50.5 | 33.3 | 35.4 | 38.8 | — | |
| PANetN-way=2-way, K-shot=1-shot, Supervision=Strong labels, Backbone=ResNet-50, Framework=iFSL2022.03 | 33.3 | 46 | 31.2 | 38.4 | 37.2 | — | |
| PANetN-way=1-way, K-shot=1-shot, Supervision=Strong labels, Backbone=ResNet-50, Framework=iFSL2022.03 | 32.8 | 45.8 | 31 | 35.1 | 36.2 | — | |
| PFENetN-way=2-way, K-shot=1-shot, Supervision=Strong labels, Backbone=ResNet-50, Framework=iFSL2022.03 | 31.1 | 47.3 | 30.8 | 32.2 | 35.3 | — | |
| ASNet_wN-way=2-way, K-shot=1-shot, Supervision=Weak labels (class tags), Backbone=ResNet-50, Framework=iFSL2022.03 | 11.4 | 20.8 | 12.5 | 15.9 | 15.1 | — | |
| ASNet_wN-way=1-way, K-shot=1-shot, Supervision=Weak labels (class tags), Backbone=ResNet-50, Framework=iFSL2022.03 | 10.8 | 20.2 | 13.1 | 16.1 | 15 | — | |
| BEITPre-training=IN-21k2022.09 | — | — | — | — | — | 0.38 | |
| BEITPre-training=Figures dataset2022.09 | — | — | — | — | — | 5.38 | |
| Copybaseline=true2022.09 | — | — | — | — | — | 12.92 | |
| MAEPre-training=IN-1k2022.09 | — | — | — | — | — | 1.92 | |
| MAEPre-training=Figures dataset2022.09 | — | — | — | — | — | 17.42 | |
| MAE-VQGANPre-training=IN-1k2022.09 | — | — | — | — | — | 2.22 | |
| MAE-VQGANPre-training=Figures dataset2022.09 | — | — | — | — | — | 27.83 | |
| VQGANPre-training=IN-1k2022.09 | — | — | — | — | — | 6.96 | |
| VQGANPre-training=Figures dataset2022.09 | — | — | — | — | — | 12.56 |