Video Grounding on MAD (test)
17.3Recall@1 (IoU=0.1)ReVisionLLM-I
Evaluation results
| Method | Links | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ReVisionLLM-IVariant=Processes more frames for higher accuracy, Reference Name=Ours†2024.11 | 17.3 | 31.4 | — | — | — | 12.7 | 23.5 | — | — | — | 6.7 | 13.1 | — | — | — | 17.5 | |
| ReVisionLLMReference Name=Ours2024.11 | 15 | 25.1 | — | — | — | 11 | 18.8 | — | — | — | 5.8 | 10.5 | — | — | — | 14.4 | |
| DeCafNetinput features=same as all prior methods2025.05 | 13.25 | 27.73 | — | — | — | 10.96 | 23.68 | — | — | — | 7.06 | 16.13 | — | — | — | 16.47 | |
| MSTG2026.01 | 12.76 | 26.43 | 34.08 | 54.84 | — | 10.94 | 22.6 | 29.41 | 48.26 | — | 6.92 | 15.43 | 20.7 | 37.77 | — | — | |
| RGNetinput features=same as all prior methods2025.05 | 12.43 | 25.12 | — | — | — | 9.48 | 18.72 | — | — | — | 5.61 | 10.86 | — | — | — | 13.7 | |
| RGNet2024.11 | 12.4 | 25.1 | — | — | — | 9.5 | 18.7 | — | — | — | 5.6 | 10.9 | — | — | — | 13.7 | |
| SOONet2024.11 | 11.3 | 23.2 | — | — | — | 9 | 19.6 | — | — | — | 5.3 | 13.1 | — | — | — | 13.6 | |
| SOONet2024.04 | 11.26 | 23.21 | 30.36 | 50.32 | — | 9 | 19.64 | 26 | 44.78 | — | 5.32 | 13.14 | 17.84 | 32.59 | — | — | |
| SOONetinput features=same as all prior methods2025.05 | 11.26 | 23.21 | — | — | — | 9 | 19.64 | — | — | — | 5.32 | 13.14 | — | — | — | 13.59 | |
| SOONet2026.01 | 11.26 | 23.21 | 30.36 | 50.32 | — | 9 | 19.64 | 26 | 44.78 | — | 5.32 | 13.14 | 17.84 | 32.59 | — | — | |
| SnAG2024.11 | 10.3 | 24.4 | — | — | — | 8.5 | 20.6 | — | — | — | 5.5 | 13.7 | — | — | — | 13.8 | |
| SnAGTraining mode=video-centric2024.04 | 10.28 | 24.42 | 32.23 | 52.28 | — | 8.46 | 20.6 | 27.5 | 46.68 | — | 5.55 | 13.75 | 19 | 35.24 | — | — | |
| SnAGinput features=same as all prior methods2025.05 | 10.28 | 24.42 | — | — | — | 8.46 | 20.6 | — | — | — | 5.55 | 13.75 | — | — | — | 13.84 | |
| SnAG2026.01 | 10.28 | 24.42 | 32.23 | 52.28 | — | 8.46 | 20.6 | 27.5 | 46.68 | — | 5.55 | 13.75 | 19 | 35.24 | — | — | |
| SnAGTraining mode=query-centric2024.04 | 10.18 | 24.3 | 32.05 | 52.05 | — | 8.37 | 20.35 | 27.25 | 46.53 | — | 5.44 | 13.51 | 18.75 | 34.83 | — | — | |
| Zero-shot CLIPGuidance Model=true, query-dependent=true, audiovisual features=true, zero-shot=true2023.02 | 9.3 | 18.96 | 24.3 | 39.79 | 47.35 | 4.65 | 13.06 | 17.73 | 32.23 | 39.58 | 2.16 | 7.4 | 11.09 | 23.21 | 29.68 | — | |
| M-Guide2024.11 | 9.3 | 18.9 | — | — | — | 4.6 | 13.1 | — | — | — | 2.2 | 7.4 | — | — | — | 9.3 | |
| M-Guideinput features=same as all prior methods2025.05 | 9.3 | 18.96 | — | — | — | 4.65 | 13.06 | — | — | — | 2.16 | 7.4 | — | — | — | 9.26 | |
| CONE2024.04 | 8.9 | 20.51 | 27.2 | 43.36 | — | 6.87 | 16.11 | 21.53 | 34.73 | — | 4.1 | 9.59 | 12.82 | 20.56 | — | — | |
| CONE2024.11 | 8.9 | 20.5 | — | — | — | 6.9 | 16.1 | — | — | — | 4.1 | 9.6 | — | — | — | 11 | |
| CONEinput features=same as all prior methods2025.05 | 8.9 | 20.51 | — | — | — | 6.87 | 16.11 | — | — | — | 4.1 | 9.59 | — | — | — | 11.01 | |
| CONE2026.01 | 8.9 | 20.51 | 27.2 | 43.36 | — | 6.87 | 16.11 | 21.53 | 34.73 | — | 4.1 | 9.59 | 12.82 | 20.56 | — | — | |
| CLIP2024.11 | 6.6 | 15.1 | — | — | — | 3.1 | 9.9 | — | — | — | 1.5 | 5.4 | — | — | — | 6.9 | |
| Zero-shot CLIPzero-shot=true2023.02 | 6.57 | 15.05 | 20.26 | 37.92 | 47.73 | 3.13 | 9.85 | 14.13 | 28.71 | 36.98 | 1.39 | 5.44 | 8.38 | 18.8 | 24.99 | — | |
| CLIP2024.04 | 6.57 | 15.05 | 20.26 | 37.92 | — | 3.13 | 9.85 | 14.13 | 28.71 | — | 1.39 | 5.44 | 8.38 | 18.8 | — | — | |
| VLG-NetGuidance Model=true, query-dependent=true, audiovisual features=true2023.02 | 5.6 | 16.07 | 23.64 | 45.35 | 55.59 | 4.28 | 13.14 | 19.86 | 39.77 | 49.38 | 2.48 | 8.78 | 13.72 | 30.22 | 39.12 | — | |
| Moment-DETRGuidance Model=true, query-dependent=true, audiovisual features=true2023.02 | 5.07 | 16.3 | 24.79 | 50.06 | 61.79 | 3.82 | 12.6 | 19.43 | 40.52 | 50.35 | 2.39 | 7.9 | 12.06 | 24.87 | 30.81 | — | |
| VLG-Net2024.04 | 3.64 | 11.66 | 17.39 | 39.78 | — | 2.76 | 9.31 | 14.56 | 34.27 | — | 1.65 | 5.99 | 9.77 | 24.93 | — | — | |
| VLG-Net2026.01 | 3.64 | 11.66 | 17.39 | 39.78 | — | 2.76 | 9.31 | 14.56 | 34.27 | — | 1.65 | 5.99 | 9.77 | 24.93 | — | — | |
| M-DETR2024.11 | 3.6 | 13 | — | — | — | 2.8 | 9.9 | — | — | — | 1.7 | 5.6 | — | — | — | 6.1 | |
| VLG-Net2023.02 | 3.5 | 11.74 | 18.32 | 38.41 | 49.65 | 2.63 | 9.49 | 15.2 | 33.68 | 43.95 | 1.61 | 6.23 | 10.18 | 25.33 | 34.18 | — | |
| VTimeLLM*Ranking method=CONE, Note=Trained by this paper on current datasets2024.11 | 1.4 | 3.1 | — | — | — | 1.3 | 2.5 | — | — | — | 0.6 | 1.1 | — | — | — | 1.7 | |
| Moment-DETR2023.02 | 0.31 | 1.52 | 2.79 | 11.08 | 19.65 | 0.24 | 1.14 | 2.06 | 7.97 | 14.29 | 0.16 | 0.68 | 1.2 | 4.71 | 8.46 | — | |
| Moment-DETR2024.04 | 0.31 | 1.52 | 2.79 | 11.08 | — | 0.24 | 1.14 | 2.06 | 7.97 | — | 0.16 | 0.28 | 1.2 | 4.71 | — | — | |
| Moment-DETR2026.01 | 0.31 | 1.52 | 2.79 | 11.08 | — | 0.24 | 1.14 | 2.06 | 7.97 | — | 0.16 | 0.28 | 1.2 | 4.71 | — | — |