Semantic Segmentation on COCO 20^i (cross-validation)
54.7Mean ScoreProtoFormer
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ProtoFormerBackbone=ResNet101, Shot=5-shot, Learnable parameters=0.6M2022.11 | 54.7 | 49.6 | 59.7 | 56.4 | 53 | 74.6 | |
| ProtoFormerBackbone=ResNet50, Shot=5-shot, Learnable parameters=0.6M2022.11 | 53.4 | 48.1 | 57.8 | 55 | 52.7 | 73.3 | |
| FECANetBackbone=ResNet50, Shots=5-shot2023.01 | 47.6 | 44.6 | 51.5 | 48.4 | 45.8 | 71.1 | |
| ProtoFormerBackbone=ResNet101, Shot=1-shot, Learnable parameters=0.6M2022.11 | 47 | 42.9 | 50.7 | 48.4 | 45.8 | 70 | |
| BAMBackbone=ResNet50, Shot=1-shot, Learnable parameters=26.7M2022.11 | 46.2 | 43.4 | 50.6 | 47.5 | 43.4 | — | |
| ProtoFormerBackbone=ResNet50, Shot=1-shot, Learnable parameters=0.6M2022.11 | 45.7 | 42.4 | 48.5 | 46.3 | 45.5 | 69.6 | |
| FECANetBackbone=ResNet50, Shots=1-shot2023.01 | 41.6 | 38.5 | 44.6 | 42.6 | 40.7 | 69.6 | |
| HSNetBackbone=ResNet101, Method=1-shot2022.01 | 41.2 | 37.2 | 44.1 | 42.4 | 41.3 | 69.1 | |
| HSNetBackbone=ResNet50, Shot=1-shot, Learnable parameters=2.5M2022.11 | 39.2 | 36.3 | 43.1 | 38.7 | 38.7 | 68.2 | |
| HSNetBackbone=ResNet50, Shots=1-shot2023.01 | 39.2 | 36.3 | 43.1 | 38.7 | 38.7 | 68.2 | |
| PFENetBackbone=ResNet101, Method=1-shot2022.01 | 38.5 | 36.8 | 41.8 | 38.7 | 36.7 | 63 | |
| FECANetBackbone=VGG16, Shots=1-shot2023.01 | 35.4 | 34.1 | 37.5 | 35.8 | 34.1 | 65.5 | |
| RePRIBackbone=ResNet50, Method=1-shot2022.01 | 34.1 | 32 | 38.7 | 32.7 | 33.1 | — | |
| RPMMBackbone=ResNet50, Method=1-shot2022.01 | 30.6 | 29.5 | 36.8 | 28.9 | 27 | — | |
| PMMBackbone=ResNet50, Method=1-shot2022.01 | 29.6 | 29.3 | 34.8 | 27.1 | 27.3 | — | |
| PPNetBackbone=ResNet50, Method=1-shot2022.01 | 29 | 28.1 | 30.8 | 29.5 | 27.7 | — | |
| PPNetBackbone=ResNet50, Shot=1-shot, Learnable parameters=31.5M2022.11 | 29 | 28.1 | 30.8 | 29.5 | 27.7 | — | |
| LSegBackbone=ViT-L/16, Method=zero-shot2022.01 | 27.2 | 28.1 | 27.5 | 30 | 23.2 | 59.9 | |
| DANBackbone=ResNet101, Method=1-shot2022.01 | 24.4 | — | — | — | — | 62.3 | |
| LSegBackbone=ResNet101, Method=zero-shot2022.01 | 23.4 | 22.1 | 25.1 | 24.9 | 21.5 | 57.9 | |
| FWBBackbone=ResNet101, Method=1-shot2022.01 | 21.2 | 17 | 18 | 21 | 28.9 | — | |
| ZS3NetBackbone=ResNet101, Method=zero-shot2022.01 | 21.1 | 18.8 | 20.1 | 24.8 | 20.5 | 55.1 | |
| FWBBackbone=VGG16, Shots=1-shot2023.01 | 20 | 18.4 | 16.7 | 19.6 | 25.4 | — |