Action Detection on Charades
26mAP (%)T-MOR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| T-MORPre-train=PoseCap-1M (V+M+T), Backbone=UNIK, #Params=3.45M, Evaluation Protocol=Fine-tuning2026.06 | 26 | — | — | — | |
| MS-TCTBackbone=I3D, GFLOPs=6.62021.12 | 25.4 | — | — | — | |
| Coarse-FineBackbone=X3D2021.12 | 25.1 | — | — | — | |
| PDANBackbone=I3D, GFLOPs=3.22021.12 | 23.7 | — | — | — | |
| TGMBackbone=I3D, GFLOPs=1.22021.12 | 20.6 | — | — | — | |
| Super-eventBackbone=I3D, GFLOPs=0.82021.12 | 18.6 | — | — | — | |
| UNIK (Backbone)Pre-train=None, Backbone=UNIK, #Params=3.45M, Evaluation Protocol=Fine-tuning2026.06 | 18.6 | — | — | — | |
| MLADBackbone=I3D, GFLOPs=44.82021.12 | 18.4 | — | — | — | |
| T-MORPre-train=PoseCap-1M (V+M), Backbone=UNIK, #Params=3.45M, Evaluation Protocol=Fine-tuning2026.06 | 18.3 | — | — | — | |
| T-MORPre-train=PoseCap-1M (V+M+T), Backbone=UNIK, #Params=40.2K, Evaluation Protocol=Linear Probing2026.06 | 16.6 | — | — | — | |
| SD-TCNPre-train=PoseCap-1M (M), Backbone=UNIK, Evaluation Protocol=Fine-tuning2026.06 | 16.5 | — | — | — | |
| R-C3DBackbone=C3D2021.12 | 12.7 | — | — | — | |
| T-MORPre-train=PoseCap-1M (V+M), Backbone=UNIK, #Params=40.2K, Evaluation Protocol=Linear Probing2026.06 | 11.3 | — | — | — | |
| SD-TCNPre-train=Smarthome, Backbone=AGCN, Evaluation Protocol=Fine-tuning2026.06 | 9.8 | — | — | — | |
| TGMPre-train=Smarthome, Backbone=TCN, Evaluation Protocol=Fine-tuning2026.06 | 9 | — | — | — | |
| Bi-LSTMPre-train=Smarthome, Backbone=LSTM, Evaluation Protocol=Fine-tuning2026.06 | 8.2 | — | — | — | |
| Previous SoTAPre-train=PoseCap-1M (M), Backbone=UNIK, Evaluation Protocol=Linear Probing2026.06 | 7.7 | — | — | — | |
| UNIK (Backbone)Pre-train=None, Backbone=UNIK, #Params=40.2K, Evaluation Protocol=Linear Probing2026.06 | 6.1 | — | — | — | |
| Coarse-FineBackbone=X3D2024.03 | — | — | 6.1 | 25.1 | |
| CTRNBackbone=I3D2024.03 | — | — | — | 25.3 | |
| DualDETRBackbone=I3D2024.03 | — | — | 15.3 | 23.2 | |
| I3D + CTRNModality=RGB2021.10 | — | 25.3 | — | — | |
| I3D + CTRNModality=RGB+OF2021.10 | — | 27.8 | — | — | |
| I3D + MLADModality=RGB2021.10 | — | 18.4 | — | — | |
| I3D + MLADModality=RGB+OF2021.10 | — | 22.9 | — | — | |
| I3D + SupereventModality=RGB2021.10 | — | 18.6 | — | — | |
| I3D + TANModality=RGB+OF2021.10 | — | 17.6 | — | — | |
| I3D + TGMModality=RGB2021.10 | — | 20.6 | — | — | |
| I3D + TGMModality=RGB+OF2021.10 | — | 21.5 | — | — | |
| I3D + TGM + SupereventModality=RGB+OF2021.10 | — | 22.3 | — | — | |
| MLADBackbone=I3D2024.03 | — | — | — | 18.4 | |
| MS-TCTBackbone=I3D2024.03 | — | — | 7.9 | 25.4 | |
| PDANBackbone=I3D2024.03 | — | — | 8.5 | 23.7 | |
| PointTADBackbone=I3D2024.03 | — | — | 12.1 | 21 | |
| R-C3DModality=RGB2021.10 | — | 12.7 | — | — | |
| R-C3DBackbone=C3D2024.03 | — | — | — | 17.6 | |
| Super-eventBackbone=I3D2024.03 | — | — | — | 18.6 | |
| TGMBackbone=I3D2024.03 | — | — | — | 20.6 |