Video Question Answering on AGQA v2
57.8Object Relations AccIPRM
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| IPRMnumber of frames=322024.11 | 57.8 | 48 | 75.6 | 62.4 | 50.7 | 20 | 58.6 | — | 60.4 | — | — | |
| GFnumber of frames=322024.11 | 55 | 44.6 | 53.2 | 59.1 | 52.8 | 14.2 | 56.1 | 62.3 | 55.1 | — | — | |
| GSMTBackbone=CLIP2026.01 | 53.94 | 44.19 | 69.12 | 61.45 | 57.32 | 21.2 | — | — | 56.16 | 69.84 | 72.53 | |
| GSMTBackbone=AIO2026.01 | 53.67 | 43.44 | 53.84 | 56.26 | 49.83 | 21.73 | — | — | 52.61 | 56.1 | 56.61 | |
| MISTnumber of frames=322024.11 | 51.7 | 42.1 | 67.2 | 60.3 | 54.6 | 19.7 | 56.1 | 54.2 | 54.4 | — | — | |
| MISTBackbone=CLIP2026.01 | 51.68 | 42.05 | 67.24 | 60.33 | 54.62 | 19.69 | — | — | 54.39 | 67.18 | 68.99 | |
| MISTBackbone=AIO2026.01 | 51.43 | 41.34 | 53.14 | 53.49 | 47.48 | 20.18 | — | — | 50.96 | 54.67 | 55.37 | |
| ATP2026.01 | 50.15 | 39.78 | 48.25 | 51.79 | 49.59 | 18.96 | — | — | 49.79 | 49.76 | 46.25 | |
| AIO2026.01 | 48.34 | 37.53 | 49.61 | 50.81 | 45.36 | 18.97 | — | — | 48.59 | 48.99 | 49.66 | |
| Temp[ATP]number of frames=322024.11 | 48.2 | 39.8 | 48.3 | 51.8 | 49.6 | 19 | 50.6 | 58.3 | 49.8 | — | — | |
| All-in-Onenumber of frames=322024.11 | 46.2 | 37.5 | 49.6 | 50.8 | 45.4 | 19 | — | — | 48.6 | — | — | |
| HCRNnumber of frames=322024.11 | 40.3 | 33.6 | 49.7 | 50 | 43.8 | 5.5 | 36.3 | 48 | 42.1 | — | — |