Affordance Grounding on AGD20k v1 (Unseen)
1.043KLDTokAG
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TokAGEvaluation Protocol=zero-shot, Model Scale=32B2026.07 | 1.043 | 0.455 | 1.549 | |
| TokAGEvaluation Protocol=zero-shot, Model Scale=2B2026.07 | 1.06 | 0.455 | 1.514 | |
| OOALSupervision=One-shot2026.03 | 1.07 | 0.461 | 0.015 | |
| PanoAffordanceNetSupervision=One-shot2026.03 | 1.185 | 0.475 | 0.0142 | |
| Moon et al.Evaluation Protocol=weakly-supervised2026.07 | 1.243 | 0.405 | 1.368 | |
| LoopTransSupervision=Weakly2026.03 | 1.247 | 0.403 | 0.0132 | |
| LoopTransEvaluation Protocol=weakly-supervised2026.07 | 1.247 | 0.403 | 1.315 | |
| WSMASupervision=Weakly2026.03 | 1.335 | 0.382 | 0.0122 | |
| WSMAEvaluation Protocol=weakly-supervised2026.07 | 1.335 | 0.382 | 1.22 | |
| INTRAEvaluation Protocol=weakly-supervised2026.07 | 1.365 | 0.375 | 1.209 | |
| OS-AGDOSupervision=One-shot2026.03 | 1.398 | 0.382 | 0.0117 | |
| LOCATESupervision=Weakly2026.03 | 1.405 | 0.372 | 0.0116 | |
| LOCATEEvaluation Protocol=weakly-supervised2026.07 | 1.405 | 0.372 | 1.157 | |
| Espresso-2DMode=Zero-shot2025.06 | 1.571 | 0.376 | 1.016 | |
| Cross-View-AG+Evaluation Protocol=weakly-supervised2026.07 | 1.765 | 0.279 | 0.882 | |
| OursMethod Type=affordance grounding2022.03 | 1.787 | 0.285 | 0.829 | |
| Cross-View-AGEvaluation Protocol=weakly-supervised2026.07 | 1.787 | 0.285 | 0.829 | |
| LISA-7BMode=Zero-shot2025.06 | 1.83 | 0.256 | 0.765 | |
| M2SA-7BMode=Zero-shot2025.06 | 1.925 | 0.227 | 0.657 | |
| AffordanceNetMode=Zero-shot2025.06 | 1.932 | 0.226 | 0.7 | |
| Molmo+SAM2Mode=Zero-shot2025.06 | 1.953 | 0.226 | 0.718 | |
| DeepGazeIIMethod Type=saliency detection2022.03 | 1.99 | 0.256 | 0.597 | |
| HotspotsMethod Type=affordance grounding2022.03 | 1.994 | 0.237 | 0.577 | |
| TS-CAMMethod Type=weakly supervised object localization2022.03 | 2.104 | 0.201 | 0.151 | |
| EILMethod Type=weakly supervised object localization2022.03 | 2.167 | 0.227 | 0.33 | |
| EgoGazeMethod Type=saliency detection2022.03 | 4.285 | 0.211 | 0.35 | |
| MlnetMethod Type=saliency detection2022.03 | 5.012 | 0.263 | 0.595 | |
| SPAMethod Type=weakly supervised object localization2022.03 | 7.425 | 0.169 | 0.262 |