Temporal Video Grounding on ActivityNet TimeLens (test)
72.3Recall@0.3Gemini-2.5-Pro
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Gemini-2.5-Pro2026.02 | 72.3 | 64.2 | 47.1 | — | |
| Gemini-2.5-Proresult_source=TimeLens paper [37]2026.05 | 72.3 | 64.2 | 47.1 | 58.1 | |
| TimeLens-8Bframework=read-then-regenerate2026.05 | 68.9 | 58.2 | 41.2 | 53.3 | |
| TimeLens-8Bframework=original2026.05 | 68.7 | 58 | 40.8 | 53.1 | |
| Gemini-2.5-Flash2026.02 | 66.8 | 57.5 | 41.3 | — | |
| Gemini-2.5-Flashresult_source=TimeLens paper [37]2026.05 | 66.8 | 57.5 | 41.3 | 52.5 | |
| Qwen3-VL-8Bframework=read-then-regenerate2026.05 | 64.2 | 52.9 | 35.4 | 48.2 | |
| Gemini-2.0-Flash2026.02 | 62.9 | 54 | 37.7 | — | |
| Gemini-2.0-Flashresult_source=TimeLens paper [37]2026.05 | 62.9 | 54 | 37.7 | 49.3 | |
| Qwen3-VL-8Bframework=original2026.05 | 62.2 | 51.6 | 34.7 | 46.9 | |
| Video-OPDbackbone=Qwen3-VL-8B2026.02 | 60.5 | 45.6 | 35.8 | — | |
| MiMo-VL-7Bframework=read-then-regenerate2026.05 | 59.6 | 47.6 | 30.4 | 44.1 | |
| OP-RKDbackbone=Qwen3-VL-8B2026.02 | 59.1 | 44.9 | 30.6 | — | |
| OP-FKDbackbone=Qwen3-VL-8B2026.02 | 58.6 | 44.2 | 30.5 | — | |
| GRPObackbone=Qwen3-VL-8B2026.02 | 58.6 | 42.7 | 32.1 | — | |
| MiMo-VL-7Bframework=original2026.05 | 58.3 | 45.8 | 29.2 | 42.6 | |
| GPT-52026.02 | 57.4 | 44.9 | 30.4 | — | |
| GPT-5result_source=TimeLens paper [37]2026.05 | 57.4 | 44.9 | 30.4 | 42.9 | |
| GPT-4o2026.02 | 55.2 | 41.4 | 25.8 | — | |
| GPT-4oresult_source=TimeLens paper [37]2026.05 | 55.2 | 41.4 | 25.8 | 40.4 | |
| MiMo-VL-7B2026.02 | 49.3 | 38.7 | 22.4 | — | |
| Qwen2.5-VL-7Breproduced=true2026.02 | 47.2 | 32.5 | 20.2 | — | |
| TVG-R1-7Breproduced=true2026.02 | 46.7 | 31 | 18.6 | — | |
| Qwen2.5-VL-7B2026.05 | 45.6 | 32 | 17 | 32.5 | |
| Time-R1-7B2026.02 | 44.8 | 31 | 19 | — | |
| Time-R1-7Bresult_source=TimeLens paper [37]2026.05 | 44.8 | 31 | 19 | 33.1 | |
| Qwen3-VL-8B-Instructreproduced=true2026.02 | 41.2 | 30.7 | 20 | — | |
| VideoChat-R1.5-7Breproduced=true2026.02 | 40.6 | 25.3 | 16.4 | — | |
| Grounded-VideoLLMresult_source=TimeLens paper [37]2026.05 | 39.2 | 29.6 | 19.5 | 30 | |
| VideoChat-Flash-7B2026.02 | 35.5 | 21.8 | 10.5 | — | |
| VideoChat-Flash-7Bresult_source=TimeLens paper [37]2026.05 | 35.5 | 21.8 | 10.5 | 24.8 | |
| VideoChat-R1-7B2026.02 | 35 | 23.9 | 11.3 | — | |
| VideoChat-R1-7Bresult_source=TimeLens paper [37]2026.05 | 35 | 23.9 | 11.3 | 25 | |
| TimeSuiteresult_source=TimeLens paper [37]2026.05 | 27.1 | 17.5 | 8.6 | 19.8 |