Moment Retrieval on QVHighlights (test)
78.52R@1 (IoU=0.5)Tempo-R0
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Tempo-R0Base Model=Qwen2-VL, Size (B)=8.3, Trained Parameters (B)=8.3, Frames=null2025.12 | 78.52 | 65.23 | — | — | — | 54.5 | — | — | — | |
| SMARTBase Model=null, Size (B)=null, Trained Parameters (B)=null, Frames=802025.12 | 78.15 | 63.16 | — | 70.76 | 55.54 | — | — | — | — | |
| SMARTBackbone=CLIP+LLM2025.11 | 78.15 | 63.16 | — | 70.76 | 55.54 | — | — | — | — | |
| LLaVA-MRBackbone=BLIP-22024.11 | 76.59 | 61.48 | — | 69.41 | 54.4 | — | — | — | — | |
| LLaVA-MRBase Model=BLIP-2, Size (B)=3.94, Trained Parameters (B)=0.02, Frames=802025.12 | 76.59 | 61.48 | — | 69.41 | 54.4 | — | — | — | — | |
| LLaVA-MRBackbone=CLIP+LLM2025.11 | 76.59 | 61.48 | — | 69.41 | 54.4 | — | — | — | — | |
| CoSTLBackbone=Audio, Pretrain=False2026.05 | 75.43 | 54.24 | — | 71.48 | 48.45 | 46.68 | — | — | — | |
| Mr. BLIPBackbone=BLIP-22024.11 | 74.77 | 60.51 | — | 68.12 | 53.38 | — | — | — | — | |
| ChronoBase Model=BLIP-2, Size (B)=3.94, Trained Parameters (B)=0.02, Frames=602025.12 | 74.77 | 60.51 | — | 68.12 | 53.38 | 51.37 | — | — | — | |
| Mr.BLIPBackbone=BLIP-22025.11 | 74.77 | 60.51 | — | 68.12 | 53.38 | — | — | — | — | |
| FlashVTG + ClipTBPBackbone=InternVideo22026.04 | 71.94 | 57.87 | — | 73.03 | 56.22 | 53.47 | — | — | — | |
| FlashVTGBackbone=InternVideo22026.04 | 70.69 | 53.96 | — | 72.33 | 53.85 | 52 | — | — | — | |
| FlashVTGBackbone=SF+CLIP2025.11 | 70.69 | 53.96 | — | 72.33 | 53.85 | — | — | — | — | |
| VIDEOLIGHTS-BPre-training=true2024.12 | 70.36 | 55.25 | — | 69.53 | 49.17 | 47.94 | — | — | — | |
| VideoLights-B-ptBackbone=SF+CLIP+BLIP2025.11 | 70.36 | 55.25 | — | 69.53 | 49.17 | — | — | — | — | |
| CVA2026.03 | 70.05 | 55.32 | — | 69.49 | 48.45 | 47.49 | — | — | — | |
| FlashVTG + ClipTBPBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 68.52 | 53.61 | — | 68.85 | 52.24 | 50.02 | — | — | — | |
| UVCOM*Backbone=BLIP2 (CLIP L) + SlowFast R-50, Pretrain=False2026.05 | 68.52 | 51.94 | — | 65.4 | 44.06 | 43.78 | — | — | — | |
| VIDEOLIGHTSPre-training=true2024.12 | 68.48 | 52.53 | — | 67.31 | 46.76 | 45.01 | — | — | — | |
| UniVTG w/ PT + ClipTBPBackbone=CLIP ViT-B/32 and SlowFast R-50, Pre-training=true2026.04 | 68.45 | 53.94 | — | 66.47 | 49 | 46.55 | — | — | — | |
| VIDEOLIGHTS-BPre-training=false2024.12 | 68.29 | 52.79 | — | 67.58 | 47.3 | 46.53 | — | — | — | |
| R2-TuningBackbone=CLIP-B/16, PANN, Post Pre-train=No, #Params=2.7M2024.03 | 68.03 | 49.35 | — | 69.04 | 47.56 | 46.17 | — | — | — | |
| R2-tuningBackbone=CLIP B/32 + SlowFast R-50, Pretrain=False2026.05 | 68.03 | 49.35 | — | 69.04 | 47.56 | 46.17 | — | — | — | |
| R2-TuningBackbone=CLIP2025.11 | 68.03 | 49.35 | — | 69.04 | 47.56 | — | — | — | — | |
| R2-Tuning2025.12 | 68 | 49.4 | — | 69 | 47.6 | 46.2 | — | — | — | |
| Video Mamba Suite + ClipTBPBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 68 | 52.65 | — | 65.65 | 48.11 | 46.23 | — | — | — | |
| LLMEPET2025.12 | 66.7 | 49.9 | — | 65.8 | 43.9 | 44.1 | — | — | — | |
| Video Mamba SuiteBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 66.65 | 52.19 | — | 64.37 | 46.68 | 45.18 | — | — | — | |
| Object-Centric FrameworkFrame-level branch=true2025.12 | 66.2 | 52.3 | — | 70.1 | 50.8 | 49.2 | — | — | — | |
| UnLoc2025.12 | 66.1 | 46.7 | — | — | — | — | — | — | — | |
| R2-Tuning2024.12 | 66.08 | 48.9 | 44.16 | 68.09 | 47.65 | 46.1 | — | — | — | |
| FlashVTG2024.12 | 66.08 | 50 | 45.84 | 67.99 | 48.7 | 47.59 | — | — | — | |
| FlashVTGBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 66.08 | 50 | — | 67.99 | 48.7 | 47.59 | — | — | — | |
| CDTR2026.03 | 65.79 | 49.6 | — | 66.44 | 45.96 | 44.37 | — | — | — | |
| TR-DETR + Oursbaseline=TR-DETR2024.12 | 65.63 | 51.23 | 46.59 | 66.89 | 49.04 | 47.77 | — | — | — | |
| CG-DETR2024.12 | 65.43 | 48.38 | 44.1 | 64.51 | 42.77 | 42.86 | — | — | — | |
| CG-DETR2024.04 | 65.43 | 48.38 | — | 64.51 | 42.77 | 42.86 | — | — | — | |
| CG-DETRBackbone=SlowFast R-50, CLIP-B/32, PANN, Post Pre-train=No, #Params=9.0M2024.03 | 65.43 | 48.38 | — | 64.51 | 42.77 | 42.86 | — | — | — | |
| UniVTGBackbone=CLIP-B/32, Post Pre-train=4.2M Corpus, #Params=41.3M2024.03 | 65.43 | 50.06 | — | 64.06 | 45.02 | 43.63 | — | — | — | |
| CG-DETRBackbone=SF+CLIP2024.11 | 65.43 | 48.38 | — | 64.51 | 42.77 | — | — | — | — | |
| UniVTGPre-training=true2024.12 | 65.43 | 50.06 | — | 64.06 | 45.02 | 43.63 | — | — | — | |
| CG-DETR2026.03 | 65.43 | 48.38 | — | 64.51 | 42.77 | 42.86 | — | — | — | |
| CG-DETRBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 65.43 | 48.38 | — | 64.51 | 42.77 | 42.86 | — | — | — | |
| UniVTG w/ PTBackbone=CLIP ViT-B/32 and SlowFast R-50, Pre-training=true2026.04 | 65.43 | 50.06 | — | 64.06 | 45.02 | 43.63 | — | — | — | |
| CG-DETRBackbone=CLIP B/32 + SlowFast R-50, Pretrain=False2026.05 | 65.43 | 48.38 | — | 64.51 | 42.77 | 42.86 | — | — | — | |
| UniVTGBackbone=CLIP B/32 + SlowFast R-50, Pretrain=True2026.05 | 65.43 | 50.06 | — | 64.06 | 45.02 | 43.63 | — | — | — | |
| CG-DETRBackbone=SF+CLIP2025.11 | 65.43 | 48.38 | — | 64.51 | 42.77 | — | — | — | — | |
| CG-DETRFeatures=Slowfast + CLIP2023.11 | 65.4 | 48.4 | — | 64.5 | 42.8 | 42.9 | — | — | — | |
| UniVTG2025.12 | 65.4 | 50.1 | — | 64.1 | 45 | 43.6 | — | — | — | |
| CG-DETR2025.12 | 65.4 | 48.4 | — | 64.5 | 42.8 | 42.9 | — | — | — | |
| UVCOM + Oursbaseline=UVCOM2024.12 | 65.37 | 50.71 | 46.92 | 66.65 | 49.22 | 48.21 | — | — | — | |
| Object-Centric FrameworkFrame-level branch=false2025.12 | 65.1 | 50.6 | — | 68.4 | 47.8 | 46.4 | — | — | — | |
| TD-DETR + ClipTBPBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 64.84 | 51.94 | — | 65.22 | 47.91 | 46.98 | — | — | — | |
| TR-DETR2024.12 | 64.66 | 48.96 | 43.7 | 63.98 | 43.73 | 42.62 | — | — | — | |
| TR-DETRBackbone=SlowFast R-50, CLIP-B/32, PANN, Post Pre-train=No, #Params=12.0M2024.03 | 64.66 | 48.96 | — | 63.98 | 43.73 | 42.62 | — | — | — | |
| TR-DETR2024.12 | 64.66 | 48.96 | — | 63.98 | 43.73 | 42.62 | — | — | — | |
| TR-DETR2026.03 | 64.66 | 48.96 | — | 63.98 | 43.73 | 42.62 | — | — | — | |
| TR-DETRBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 64.66 | 48.96 | — | 63.98 | 43.73 | 42.62 | — | — | — | |
| UVCOMASR Captions Pretrain=true2023.11 | 64.53 | 48.31 | — | 64.78 | 43.65 | 43.8 | — | — | — | |
| UVCOMPre-training=true2024.12 | 64.53 | 48.31 | — | 64.78 | 43.65 | 43.8 | — | — | — | |
| TD-DETR2026.03 | 64.53 | 50.37 | — | 66.21 | 47.32 | 46.69 | — | — | — | |
| TD-DETRBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 64.53 | 50.37 | — | 66.21 | 47.32 | 46.69 | — | — | — | |
| LMRBackbone=SF+CLIP2025.11 | 64.4 | 47.21 | — | 64.65 | 43.16 | — | — | — | — | |
| QD-DETRASR Captions Pretrain=true2023.11 | 64.1 | 46.1 | — | 64.3 | 40.5 | 40.62 | — | — | — | |
| QD-DETRPre-training=true2024.12 | 64.1 | 46.1 | — | 64.3 | 40.5 | 40.62 | — | — | — | |
| UVCOMAudio Modality=true2023.11 | 63.81 | 48.7 | — | 64.47 | 44.01 | 43.27 | — | — | — | |
| QD-DETR + Oursbaseline=QD-DETR2024.12 | 63.62 | 48.77 | 45.02 | 65.5 | 47.78 | 46.61 | — | — | — | |
| UVCOM2025.12 | 63.6 | 47.5 | — | 63.4 | 42.7 | 43.2 | — | — | — | |
| UVCOM2024.12 | 63.55 | 47.47 | 43.85 | 63.37 | 42.67 | 43.18 | — | — | — | |
| UVCOM2023.11 | 63.55 | 47.47 | — | 63.37 | 42.67 | 43.18 | — | — | — | |
| UVCOMPre-training=false2024.12 | 63.55 | 47.47 | — | 63.37 | 42.67 | 43.18 | — | — | — | |
| UVCOM2026.03 | 63.55 | 47.47 | — | 63.37 | 42.67 | 43.18 | — | — | — | |
| UVCOMBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 63.55 | 47.47 | — | 63.37 | 42.67 | 43.18 | — | — | — | |
| UVCOMBackbone=CLIP B/32 + SlowFast R-50, Pretrain=False2026.05 | 63.55 | 47.47 | — | 63.37 | 42.67 | 43.18 | — | — | — | |
| UVCOMBackbone=SF+CLIP2025.11 | 63.55 | 47.47 | — | 63.37 | 42.67 | — | — | — | — | |
| VIDEOLIGHTSPre-training=false2024.12 | 63.36 | 48.7 | — | 63.81 | 42.87 | 43.38 | — | — | — | |
| QD-DETRSource Modality=V+A2023.03 | 63.06 | 45.1 | — | 63.04 | 40.1 | 40.19 | — | — | — | |
| QD-DETRAudio Modality=true2023.11 | 63.06 | 45.1 | — | 63.04 | 40.1 | 40.19 | — | — | — | |
| DiffusionVMRProposal spans=50, Sampling steps=52023.08 | 62.91 | 47.08 | — | 63.68 | 43.88 | 42.74 | — | — | — | |
| MESM2026.03 | 62.78 | 45.2 | — | 62.64 | 41.45 | 40.68 | — | — | — | |
| BAM-DETR2024.12 | 62.71 | 48.64 | — | 64.57 | 46.33 | 45.36 | — | — | — | |
| BAM-DETR2026.03 | 62.71 | 48.64 | — | 64.57 | 46.33 | 45.36 | — | — | — | |
| BAM-DETR2025.12 | 62.7 | 48.6 | — | 64.4 | 46.3 | 45.4 | — | — | — | |
| QD-DETRSource Modality=V2023.03 | 62.4 | 44.98 | — | 62.52 | 39.88 | 39.86 | — | — | — | |
| QD-DETR2023.05 | 62.4 | 44.9 | — | — | — | 39.8 | — | — | — | |
| QD-DETRProposal spans=50, Sampling steps=52023.08 | 62.4 | 44.98 | — | 62.52 | 39.88 | 39.86 | — | — | — | |
| QD-DETR2024.04 | 62.4 | 44.98 | — | 62.52 | 39.88 | 39.86 | — | — | — | |
| QD-DETRFeatures=Slowfast + CLIP2023.11 | 62.4 | 45 | — | 62.5 | 39.9 | 39.9 | — | — | — | |
| QD-DETRBackbone=SlowFast R-50, CLIP-B/32, PANN, Post Pre-train=No, #Params=7.6M2024.03 | 62.4 | 44.98 | — | 62.52 | 39.88 | 39.86 | — | — | — | |
| QD-DETRText Feat=CLIP, Video Feat=SF+C2023.06 | 62.4 | 44.98 | — | 62.52 | 39.88 | — | — | 39.86 | — | |
| QD-DETRBackbone=SF+CLIP2024.11 | 62.4 | 44.98 | — | 62.52 | 39.88 | — | — | — | — | |
| QD-DETRPre-training=false2024.12 | 62.4 | 44.98 | — | 63.17 | 42.05 | 41.44 | — | — | — | |
| QD-DETR2025.12 | 62.4 | 45 | — | 62.5 | 39.9 | 39.9 | — | — | — | |
| QD-DETR2026.03 | 62.4 | 44.98 | — | 62.52 | 39.88 | 39.86 | — | — | — | |
| QD-DETRBackbone=CLIP ViT-B/32 and SlowFast R-502026.04 | 62.4 | 44.98 | — | 62.52 | 39.88 | 39.86 | — | — | — | |
| QD-DETRBackbone=CLIP B/32 + SlowFast R-50, Pretrain=False2026.05 | 62.4 | 44.96 | — | 62.52 | 39.88 | 39.86 | — | — | — | |
| QD-DETRBackbone=SF+CLIP2025.11 | 62.4 | 44.98 | — | 62.52 | 39.88 | — | — | — | — | |
| TaskWeavetest from author provided checkpoint=true2024.12 | 61.87 | 46.24 | 42.45 | 63.75 | 43.63 | 43.33 | — | — | — | |
| EaTR2023.11 | 61.36 | 45.79 | — | 61.86 | 41.91 | 41.74 | — | — | — | |
| EaTR2024.12 | 61.36 | 45.79 | — | 61.86 | 41.91 | 41.74 | — | — | — |