Long-form Video Understanding on LVU
76.47Relation Attribute AccuracySVT SCALE
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| SVT SCALEEvaluation Protocol=Linear probe2023.04 | 76.47 | 42.27 | 74.02 | 49.09 | 58.97 | 62.76 | 39.23 | — | — | — | |
| Movies2ScenesEvaluation Protocol=Linear probe2023.04 | 71.2 | 42.2 | 68.2 | 70.9 | 55.9 | 53.7 | 57.8 | — | — | — | |
| Movies2Scene2025.03 | 71.2 | 42.2 | 68.2 | — | — | — | — | — | — | 60.5 | |
| HierarQLLM finetuning=true2025.03 | 69.4 | 49.3 | 85.1 | — | — | — | — | — | — | 67.9 | |
| HierarQLLM finetuning=false2025.03 | 67.9 | 48.7 | 83.8 | — | — | — | — | — | — | 66.8 | |
| S52024.04 | 67.1 | 42.1 | 73.5 | 67.3 | 65.4 | 51.3 | 48 | — | — | 59.2 | |
| S52024.11 | 67.1 | 42.1 | 73.5 | 67.3 | 65.4 | 51.3 | 48 | — | — | 59.2 | |
| S52025.03 | 67.1 | 42.1 | 73.5 | — | — | — | — | — | — | 60.9 | |
| AdaCM²2024.11 | 63.1 | 40.2 | 86.2 | 75.4 | 68 | 77 | 62.5 | — | — | 67.5 | |
| VideoMambae2e=true, Backbone=VM-Ti, frames=322024.03 | 62.5 | 40.43 | 70.37 | 67.29 | 65.24 | 52.98 | 48.23 | 26 | 2.9 | — | |
| VideoMamba2025.03 | 62.5 | 40.4 | 70.4 | — | — | — | — | — | — | 57.8 | |
| LF-VILA2025.03 | 61.5 | 41.3 | 68 | — | — | — | — | — | — | 56.9 | |
| LSSD (B.3)Representations=Video-only, Evaluation Protocol=Linear probe2023.04 | 60.98 | 46.03 | 63.41 | 69.16 | 60.88 | 48.81 | 52.48 | — | — | — | |
| LSSD (B.4)Representations=Video-only, Evaluation Protocol=Linear probe2023.04 | 60.98 | 47.62 | 65.85 | 67.29 | 61.73 | 41.67 | 55.32 | — | — | — | |
| TranS4mer2025.03 | 59.5 | 39.2 | 70.9 | — | — | — | — | — | — | 56.5 | |
| STCAEvaluation Protocol=Linear probe2023.04 | 59.25 | 41.62 | 69.15 | 66.7 | 56.62 | 52.93 | 53.3 | — | — | — | |
| LSSD (B.5)Representations=Video-only, Evaluation Protocol=Linear probe2023.04 | 58.54 | 46.03 | 64.63 | 69.16 | 64.29 | 41.07 | 52.48 | — | — | — | |
| MA-LMM2024.04 | 58.2 | 44.8 | 80.3 | 74.6 | 61 | 70.4 | 51.9 | — | — | 63 | |
| MA-LMM2024.11 | 58.2 | 44.8 | 80.3 | 74.6 | 61 | 70.4 | 51.9 | — | — | 63 | |
| MA-LMM2025.03 | 58.2 | 44.8 | 80.3 | — | — | — | — | — | — | 61.1 | |
| ViS4mere2e=false, Backbone=ViT-L2024.03 | 57.14 | 40.79 | 67.44 | 62.61 | 54.71 | 48.8 | 44.75 | 26 | 3.63 | — | |
| ViS4merEvaluation Protocol=Linear probe2023.04 | 57.14 | 40.79 | 67.44 | 62.61 | 54.71 | 48.8 | 44.75 | — | — | — | |
| VIS4mer2024.04 | 57.1 | 40.8 | 67.4 | 62.6 | 54.7 | 48.8 | 44.8 | — | — | 53.7 | |
| VIS4mer2024.11 | 57.1 | 40.8 | 67.4 | 62.6 | 54.7 | 48.8 | 44.8 | — | — | 53.7 | |
| VIS4mer2025.03 | 57.1 | 40.8 | 67.4 | — | — | — | — | — | — | 55.1 | |
| Obj_T4mer2024.04 | 54.8 | 33.2 | 52.9 | 47.7 | 52.7 | 36.3 | 37.8 | — | — | 45 | |
| Obj-T4mer2024.11 | 54.8 | 33.2 | 52.9 | 47.7 | 52.7 | 36.3 | 37.8 | — | — | 45.1 | |
| Obj_T4mer2025.03 | 54.8 | 33.2 | 52.9 | — | — | — | — | — | — | 47 | |
| Object TransformerEvaluation Protocol=Linear probe2023.04 | 54.7 | 40.3 | 60 | 58.9 | 56.1 | 35.1 | 40.6 | — | — | — | |
| Object Trans.e2e=false, Backbone=ResNet2024.03 | 53.1 | 39.4 | 56.9 | 51.2 | 54.6 | 34.5 | 39.1 | 23 | 3.55 | — | |
| VideoBERTe2e=false, Backbone=S3D2024.03 | 52.8 | 37.9 | 54.9 | 47.3 | 51.9 | 38.5 | 36.1 | 32 | 4.46 | — | |
| VideoBERT2024.04 | 52.8 | 37.9 | 54.9 | 47.3 | 51.9 | 38.5 | 36.1 | — | — | 45.6 | |
| VideoBERT2024.11 | 52.8 | 37.9 | 54.9 | 47.3 | 51.9 | 38.5 | 36.1 | — | — | 45.6 | |
| VideoBERT2025.03 | 52.8 | 37.9 | 54.9 | — | — | — | — | — | — | 48.5 | |
| LST2024.04 | 52.5 | 37.3 | 62.8 | 56.1 | 52.7 | 42.3 | 39.2 | — | — | 49 | |
| LST2024.11 | 52.5 | 37.3 | 62.8 | 56.1 | 52.7 | 42.3 | 39.2 | — | — | 49 | |
| LSTe2e=false, Backbone=ViT-L2024.03 | 52.38 | 37.31 | 62.79 | 56.07 | 52.7 | 42.26 | 39.16 | 31 | 3.83 | — | |
| Performere2e=false, Backbone=ViT-L2024.03 | 50 | 38.8 | 60.46 | 58.87 | 49.45 | 48.21 | 41.25 | 31 | 3.93 | — | |
| Orthoformere2e=false, Backbone=ViT-L2024.03 | 50 | 39.3 | 66.27 | 55.14 | 55.79 | 47.02 | 43.35 | 29 | 3.86 | — | |
| Performer2024.04 | 50 | 38.8 | 60.5 | 58.9 | 49.5 | 48.2 | 41.3 | — | — | 49.6 | |
| Orthoformer2024.04 | 50 | 38.3 | 66.3 | 55.1 | 55.8 | 47 | 43.4 | — | — | 50.8 | |
| Performer2024.11 | 50 | 38.8 | 60.5 | 58.9 | 49.5 | 48.2 | 41.3 | — | — | 49.6 | |
| Orthoformer2024.11 | 50 | 38.3 | 66.3 | 55.1 | 55.8 | 47 | 43.4 | — | — | 50.8 | |
| Orthoformer2025.03 | 50 | 38.3 | 66.3 | — | — | — | — | — | — | 51.5 |