Temporal Grounding on ActivityNet Captions
58.32Recall@1 (IoU=0.5)MKTN
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MKTNType=Multi-Pair (MP)2024.12 | 58.32 | — | 35.28 | — | 86.2 | 71.49 | — | |
| EaTRType=VLP, Zero-shot=false2025.12 | 58.2 | — | 37.6 | — | — | — | — | |
| Time-R1Type=RL, Zero-shot=false, fine-tuned on dataset=true2025.12 | 55.6 | 73.3 | 34 | — | — | — | — | |
| TempR1Type=RL, Zero-shot=false, fine-tuned on dataset=true2025.12 | 55.4 | 72.5 | 34.1 | — | — | — | 51.9 | |
| SSRNType=VLP, Zero-shot=false2025.12 | 54.5 | — | 33.2 | — | — | — | — | |
| SCADABackbone=C3D2026.04 | 52.69 | — | 34.73 | — | 82.6 | 66.67 | 46.75 | |
| G2LBackbone=C3D2026.04 | 51.68 | — | 33.35 | — | 81.32 | 67.6 | — | |
| MMNFeature extractor=C3D2021.09 | 48.59 | 65.05 | 29.26 | 87.25 | 79.5 | 64.76 | — | |
| MMNSupervision=Full supervision2023.08 | 48.59 | 65.05 | 29.26 | 87.25 | 79.5 | 64.76 | — | |
| MMNBackbone=C3D2026.04 | 48.59 | — | 29.26 | — | 79.5 | 64.76 | — | |
| MMNType=Fully-supervised (FS)2024.12 | 48.59 | — | 29.26 | — | 79.5 | 64.76 | — | |
| SCADABackbone=I3D2026.04 | 48.49 | — | 31.36 | — | 81.08 | 66.33 | 45.69 | |
| CBLNFeature extractor=C3D2021.09 | 48.12 | 66.34 | 27.6 | 88.91 | 79.32 | 63.41 | — | |
| VDIType=Fully-supervised (FS)2024.12 | 48.09 | — | 28.76 | — | 79.69 | 64.88 | — | |
| MATSupervision=Full supervision2023.08 | 48.02 | — | 31.78 | — | 78.02 | 63.18 | — | |
| SSCSFeature extractor=C3D2021.09 | 46.67 | 61.35 | 27.56 | 86.89 | 78.37 | 63.78 | — | |
| SSCSSupervision=Full supervision2023.08 | 46.67 | 61.35 | 27.56 | 86.89 | 78.37 | 63.78 | — | |
| MS-2D-TANBackbone=C3D2026.04 | 46.16 | — | 29.21 | — | 78.8 | 60.85 | — | |
| MS-2D-TANBackbone=I3D2026.04 | 45.5 | — | 28.28 | — | 79.36 | 61.7 | — | |
| DRNFeature extractor=C3D2021.09 | 45.45 | — | 24.39 | — | 77.97 | 50.3 | — | |
| CPNFeature extractor=C3D2021.09 | 45.1 | 62.81 | 28.1 | — | — | — | — | |
| CPNBackbone=C3D2026.04 | 45.1 | — | 28.1 | — | — | — | 45.7 | |
| FVMRFeature extractor=C3D2021.09 | 45 | 60.63 | 26.85 | 86.11 | 77.42 | 61.04 | — | |
| CMINFeature extractor=C3D2021.09 | 44.62 | 63.61 | 24.48 | 82.39 | 69.66 | 52.96 | — | |
| 2D-TANFeature extractor=C3D2021.09 | 44.51 | 59.46 | 26.54 | 85.53 | 77.13 | 61.96 | — | |
| 2D-TANSupervision=Full supervision2023.08 | 44.51 | 59.46 | 26.54 | 85.53 | 77.13 | 61.96 | — | |
| 2D-TANType=Fully-supervised (FS)2024.12 | 44.51 | — | 26.54 | — | 77.13 | 61.96 | — | |
| IVG-DCLFeature extractor=C3D2021.09 | 43.84 | 63.22 | 27.1 | — | — | — | — | |
| UniVTGType=VLP, Zero-shot=false2025.12 | 43.4 | 56.1 | 24.3 | — | — | — | — | |
| VSLNetBackbone=I3D2026.04 | 43.22 | — | 26.16 | — | — | — | 43.19 | |
| MRTNetType=Fully-supervised (FS)2024.12 | 42.02 | — | 24.25 | — | — | — | — | |
| SlotVTGSource dataset=QVHighlights, LLM size=7B2026.03 | 42 | 61.7 | 25.3 | — | — | — | 44.1 | |
| LGIFeature extractor=C3D2021.09 | 41.51 | 58.52 | 23.07 | — | — | — | — | |
| LGISupervision=Full supervision2023.08 | 41.51 | 58.52 | 23.07 | — | — | — | — | |
| Chrono-QwenSource dataset=QVHighlights, LLM size=7B2026.03 | 41.4 | 60.7 | 24.8 | — | — | — | 43.4 | |
| TAR-TVGType=RL, Zero-shot=false2025.12 | 39.8 | 61.5 | 19.8 | — | — | — | 41.1 | |
| PS-VTGSupervision=Glance supervision2023.08 | 39.59 | 59.71 | 21.98 | — | — | — | — | |
| PMIFeature extractor=C3D2021.09 | 38.28 | 59.69 | 17.83 | — | — | — | — | |
| TRACEType=SFT, Zero-shot=true2025.12 | 37.7 | — | 24 | — | — | — | 39 | |
| TRACEBackbone=LLM-based2026.04 | 37.7 | — | 24 | — | — | — | 39 | |
| SCDMFeature extractor=C3D2021.09 | 36.75 | 54.8 | 19.86 | 77.29 | 64.99 | 41.53 | — | |
| D3GSupervision=Glance supervision2023.08 | 36.68 | 58.25 | 18.54 | 87.84 | 74.21 | 52.47 | — | |
| PFUSupervision=Glance supervision2023.08 | 36.35 | 59.63 | 16.61 | — | — | — | — | |
| SlotVTGSource dataset=QVHighlights, LLM size=3B2026.03 | 35.7 | 56.1 | 21.1 | — | — | — | 40 | |
| VIGASupervision=Glance supervision2023.08 | 35.39 | 59.78 | 16.25 | 72.19 | 53.19 | 32.69 | — | |
| Chrono-QwenSource dataset=QVHighlights, LLM size=3B2026.03 | 35.3 | 55.2 | 20.8 | — | — | — | 39.2 | |
| HawkEyeType=SFT, Zero-shot=false, fine-tuned on dataset=true2025.12 | 34.7 | 55.9 | 17.9 | — | — | — | 39.1 | |
| VideoChat-R1Type=RL, Zero-shot=false2025.12 | 33.4 | — | 17.7 | — | — | — | 36.6 | |
| CNMSupervision=Weak supervision2023.08 | 33.33 | 55.68 | — | — | — | — | — | |
| QSPNFeature extractor=C3D2021.09 | 33.26 | 52.13 | 13.43 | 77.72 | 62.39 | 40.78 | — | |
| SlotVTGSource dataset=Charades-STA, LLM size=7B2026.03 | 33.2 | 52 | 16.7 | — | — | — | 35.5 | |
| MMDistType=Weakly-supervised (WS)2024.12 | 32.98 | — | — | — | — | — | — | |
| CRMSupervision=Weak supervision2023.08 | 32.19 | 55.26 | — | — | — | — | — | |
| CPLType=Weakly-supervised (WS)2024.12 | 31.37 | — | — | — | 43.13 | — | — | |
| CPLSupervision=Weak supervision2023.08 | 31.24 | 53.67 | — | 63.05 | 43.14 | — | — | |
| VCAType=Weakly-supervised (WS)2024.12 | 31 | — | — | — | 53.83 | — | — | |
| VideoMindSource dataset=Zero-Shot, LLM size=7B2026.03 | 30.3 | 48.4 | 15.7 | — | — | — | 33.3 | |
| WSTANType=Weakly-supervised (WS)2024.12 | 30.01 | — | — | — | 63.42 | — | — | |
| ICVCType=Weakly-supervised (WS)2024.12 | 29.52 | — | — | — | 66.61 | — | — | |
| HawkEyeType=SFT, Zero-shot=true2025.12 | 29.3 | 49.1 | 10.7 | — | — | — | 32.7 | |
| HawkEyeSource dataset=Zero-Shot, LLM size=7B2026.03 | 29.3 | 49.1 | 10.7 | — | — | — | 32.7 | |
| HawkEyeBackbone=LLM-based2026.04 | 29.3 | — | 10.7 | — | — | — | 32.7 | |
| Chrono-QwenSource dataset=Charades-STA, LLM size=7B2026.03 | 29.2 | 46.5 | 14.6 | — | — | — | 32.6 | |
| CTRLFeature extractor=C3D2021.09 | 29.01 | 47.43 | 10.34 | 75.32 | 59.17 | 37.54 | — | |
| CTRLSupervision=Full supervision2023.08 | 29.01 | 47.43 | 10.34 | 75.32 | 59.17 | 37.54 | — | |
| TempR1Type=RL, Zero-shot=false2025.12 | 28.9 | 47.6 | 14 | — | — | — | 32.2 | |
| SlotVTGSource dataset=Charades-STA, LLM size=3B2026.03 | 28.7 | 47.7 | 14.4 | — | — | — | 32.2 | |
| VTimeLLMType=SFT, Zero-shot=true2025.12 | 27.8 | 44 | 14.3 | — | — | — | 30.4 | |
| VideoMindSource dataset=Zero-Shot, LLM size=2B2026.03 | 26.5 | 44 | 12.6 | — | — | — | 30.1 | |
| LCNetSupervision=Weak supervision2023.08 | 26.33 | 48.49 | — | 82.51 | 62.66 | — | — | |
| Chrono-QwenSource dataset=Charades-STA, LLM size=3B2026.03 | 26.3 | 44.4 | 13.1 | — | — | — | 30.1 | |
| MomentorBackbone=LLM-based2026.04 | 23 | — | 12.4 | — | — | — | 29.3 | |
| UniTimeSource dataset=Zero-Shot, LLM size=7B2026.03 | 22.8 | — | 14.1 | — | — | — | 27.3 | |
| ChatVTGType=SFT, Zero-shot=true2025.12 | 22.5 | 40.7 | 9.4 | — | — | — | 27.2 | |
| MUSEGType=RL, Zero-shot=false2025.12 | 22.4 | 38.3 | 10.5 | — | — | — | 26.5 | |
| Chrono-BLIPSource dataset=Charades-STA, LLM size=4B2026.03 | 22.4 | 41.8 | 9.7 | — | — | — | 27.7 | |
| Chrono-BLIPSource dataset=QVHighlights, LLM size=4B2026.03 | 22.4 | 41.8 | 9.7 | — | — | — | 27.7 | |
| CG-DETRSource dataset=QVHighlights, LLM size=-2026.03 | 21.5 | 37.7 | 10.4 | — | — | — | 26 | |
| MCNFeature extractor=C3D2021.09 | 21.36 | 39.35 | 6.43 | 68.12 | 53.23 | 29.7 | — | |
| EaTRSource dataset=QVHighlights, LLM size=-2026.03 | 20.9 | 36.7 | 9.7 | — | — | — | 25.3 | |
| EaTRSource dataset=Charades-STA, LLM size=-2026.03 | 18.8 | 36.9 | 7.3 | — | — | — | 24.1 | |
| CG-DETRSource dataset=Charades-STA, LLM size=-2026.03 | 16.8 | 32.6 | 6.8 | — | — | — | 22.1 | |
| TimeSuiteSource dataset=Zero-Shot, LLM size=7B2026.03 | 16.6 | — | 9.3 | — | — | — | 22 | |
| VideoChat-R1.5Type=RL, Zero-shot=false2025.12 | 15.6 | 28.3 | 7.2 | — | — | — | 20.2 | |
| Qwen-2.5-VL-7BType=SFT, Zero-shot=true2025.12 | 13 | 22.5 | 6.3 | — | — | — | 17.2 |