Generalized Few-Shot Semantic Segmentation on PASCAL-5i
78.5Base mIoUOpen-V (alpha=0.5, 5-shot)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Open-V (alpha=0.5, 5-shot)Trained?=False, OV?=True, alpha=0.5, shots=5-shot, backbone=frozen SAM3/CLIP2026.06 | 78.5 | 76.9 | — | 77.6 | |
| Open-V (alpha=0.5, 1-shot)Trained?=False, OV?=True, alpha=0.5, shots=1-shot, backbone=frozen SAM3/CLIP2026.06 | 78.4 | 77.5 | — | 77.9 | |
| Open-V (zero-shot, aligned)Trained?=False, OV?=True, alignment=aligned, evaluation_probe=split-0 200-episode, backbone=frozen SAM3/CLIP2026.06 | 76 | 76.9 | — | 76.5 | |
| BCMTrained?=True, OV?=False2026.06 | 75 | 47 | — | 57.8 | |
| Vis. PromptingTrained?=True, OV?=True2026.06 | 74.9 | 50.3 | — | 60.2 | |
| VPShots=5-shot, Backbone=ResNet-502026.03 | 74.86 | 50.34 | 62.6 | 60.2 | |
| Syn4SegShots=1-shot, Backbone=ResNet-502026.03 | 74.8 | 46.13 | 60.47 | 57.07 | |
| VPShots=1-shot, Backbone=ResNet-502026.03 | 74.58 | 34.99 | 54.79 | 47.63 | |
| Syn4SegShots=5-shot, Backbone=ResNet-502026.03 | 74.26 | 58.72 | 65.49 | 65.58 | |
| BAMShots=1-shot, Backbone=ResNet-502026.03 | 71.6 | 27.49 | 49.55 | 39.73 | |
| BAMShots=5-shot, Backbone=ResNet-502026.03 | 71.6 | 28.96 | 50.28 | 41.24 | |
| BCMShots=5-shot, Backbone=ResNet-502026.03 | 71.23 | 55.36 | 63.29 | 62.3 | |
| BCMShots=1-shot, Backbone=ResNet-502026.03 | 71.15 | 41.24 | 56.2 | 52.22 | |
| MiBShots=5-shot, Backbone=ResNet-502026.03 | 68.6 | 28.93 | 48.77 | 40.7 | |
| AlignDiffShots=5-shot, Backbone=ResNet-502026.03 | 68.43 | 47.41 | 57.92 | 56.01 | |
| AlignDiffShots=1-shot, Backbone=ResNet-502026.03 | 67.32 | 43.34 | 55.33 | 52.73 | |
| DIaMShots=1-shot, Backbone=ResNet-502026.03 | 66.79 | 27.36 | 47.08 | 38.82 | |
| CAPLShots=5-shot, Backbone=ResNet-502026.03 | 65.43 | 24.43 | 44.93 | 35.58 | |
| CAPLShots=1-shot, Backbone=ResNet-502026.03 | 64.8 | 17.46 | 41.13 | 27.51 | |
| DIaMShots=5-shot, Backbone=ResNet-502026.03 | 64.05 | 34.56 | 49.31 | 44.9 | |
| MiBShots=1-shot, Backbone=ResNet-502026.03 | 63.8 | 8.86 | 36.33 | 15.56 | |
| POPShots=1-shot, Backbone=ResNet-502026.03 | 46.68 | 19.94 | 33.32 | 27.94 | |
| Open-V (alpha=0, no align)Trained?=False, OV?=True, alpha=0, alignment=no align, backbone=frozen SAM3/CLIP2026.06 | 45.6 | 28.9 | — | 35.3 | |
| POPShots=5-shot, Backbone=ResNet-502026.03 | 41.5 | 36.26 | 38.8 | 38.7 | |
| PANetShots=5-shot, Backbone=ResNet-502026.03 | 32.95 | 15.25 | 24.1 | 20.85 | |
| PANetShots=1-shot, Backbone=ResNet-502026.03 | 31.88 | 11.25 | 21.57 | 16.63 | |
| SCLShots=5-shot, Backbone=ResNet-502026.03 | 9.11 | 1.83 | 5.47 | 3.05 | |
| CANetShots=5-shot, Backbone=ResNet-502026.03 | 9.05 | 1.52 | 5.29 | 2.6 | |
| SCLShots=1-shot, Backbone=ResNet-502026.03 | 8.88 | 2.44 | 5.66 | 3.83 | |
| PFENetShots=5-shot, Backbone=ResNet-502026.03 | 8.83 | 1.89 | 5.36 | 3.11 | |
| CANetShots=1-shot, Backbone=ResNet-502026.03 | 8.73 | 2.42 | 5.58 | 3.79 | |
| PFENetShots=1-shot, Backbone=ResNet-502026.03 | 8.32 | 2.67 | 5.5 | 4.04 |