Object Recognition on YTVIS
28.4Top-1 AccuracyDSSA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DSSAstate_representation=local (l_k^t), evaluation_protocol=Two-layer MLP trained on frozen representations2026.06 | 28.4 | 66.6 | 52.1 | 9,269 | |
| RandSF.Qtsimevaluation_protocol=Two-layer MLP trained on frozen representations2026.06 | 26.1 | 60.9 | 54.5 | 7,579 | |
| DSSAstate_representation=identity (e_k^t), evaluation_protocol=Two-layer MLP trained on frozen representations2026.06 | 24.2 | 63.2 | 50.7 | 9,292 | |
| SlotContrastevaluation_protocol=Two-layer MLP trained on frozen representations2026.06 | 19.9 | 49.1 | 53.5 | 9,259 |