Video-to-Text Retrieval on LSMDC
46.7R@1InternVideo2s2-6B
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| InternVideo2s2-6BFinetuning=true2024.03 | 46.7 | — | — | — | — | — | — | |
| UMT-LFinetuning=true2024.03 | 41.4 | — | — | — | — | — | — | |
| InternVideoBackbone=ViT-L/142022.12 | 34.9 | — | — | — | — | — | — | |
| ViCLIPFinetuning=true2024.03 | 32.5 | — | — | — | — | — | — | |
| CLIP-ViPReranking (Matching Loss)=true2023.10 | 29.4 | 50.6 | 59 | — | — | — | — | |
| X-CLIPBackbone=ViT-B/162022.07 | 26.9 | 46.2 | — | — | 41.9 | — | — | |
| X-CLIP2022.12 | 26.9 | — | — | — | — | — | — | |
| EMCL-NetPre-trained weights=CLIP (ViT-B/32), Inverted softmax=true2022.11 | 26.7 | 44.7 | 54.4 | 8 | — | — | — | |
| InternVideoBackbone=UniformerV2, Video patch dropping=true2023.10 | 24.7 | 44 | 53.1 | 8 | — | — | — | |
| VIOLETv2Reranking (Matching Loss)=true2023.10 | 24 | 43.5 | 54.1 | — | — | — | — | |
| CLIP4Clip2022.12 | 23.8 | — | — | — | — | — | — | |
| CLIP4ClipFinetuning=true2024.03 | 23.8 | — | — | — | — | — | — | |
| CLIP4Clip-seqTransfBackbone=ViT-B/162022.07 | 23.2 | 42.4 | — | — | 47.4 | — | — | |
| DREAMYear=-2026.06 | 23.1 | 40.9 | 51.3 | 10 | 51 | — | — | |
| DiffusionRet2023.03 | 23 | 43.5 | 51.5 | 9 | 40.2 | 118 | — | |
| DiffusionRetPost-processing=QB-Norm2023.03 | 22.8 | 43.2 | 51.6 | 9 | 40 | 117.6 | — | |
| DiscoVLABackbone=CLIP (ViT-B/32), # Params (M)=0.562025.06 | 22.8 | 41.9 | 51.2 | — | 46.6 | 115.9 | — | |
| X-Pool2022.03 | 22.7 | 42.6 | 51.2 | 10 | 47.4 | — | — | |
| CLIP4Clip-MeanPBackbone=ViT-B/162022.07 | 22.6 | 50.5 | — | — | 50.3 | — | — | |
| CLIPFinetuning=true2024.03 | 22.6 | — | — | — | — | — | — | |
| CLIP4ClipReranking (Matching Loss)=false2023.10 | 22.6 | 41 | 49.1 | 11 | — | — | — | |
| X-CLIPBackbone=ViT-B/322022.07 | 22.5 | 42.2 | — | — | 50.7 | — | — | |
| InternVideoBackbone=UniformerV2, Video patch dropping=false2023.10 | 22.5 | 43.7 | 54.3 | 8 | — | — | — | |
| UMPYear=20242026.06 | 22.5 | 40.7 | 50.8 | — | 51.7 | — | — | |
| VoPF+CParams (M)=14.12022.11 | 22.3 | 40.3 | 50.7 | 10 | 51.1 | — | — | |
| VoPF+CBackbone=CLIP (ViT-B/32), # Params (M)=14.102025.06 | 22.3 | 40.3 | 50.7 | — | 51.1 | 113.3 | — | |
| EMCL-NetPre-trained weights=CLIP (ViT-B/32), Inverted softmax=false2022.11 | 22.2 | 40.6 | 49.2 | 12 | — | — | — | |
| EMCL-Net2023.03 | 22.2 | 40.6 | 49.2 | 12 | — | 112 | — | |
| InternVideoBackbone=ViT, Video patch dropping=true2023.10 | 22 | 42.3 | 52.5 | 9 | — | — | — | |
| FullParams (M)=119.82022.11 | 21.9 | 40 | 48.2 | 12 | 50.7 | — | — | |
| BridgeFormerReranking (Matching Loss)=false2023.10 | 21.8 | 41.1 | 50.6 | 10 | — | — | — | |
| VoPF+PParams (M)=0.42022.11 | 21.5 | 40.6 | 50.7 | 10 | 50.8 | — | — | |
| VoPF+PBackbone=CLIP (ViT-B/32), # Params (M)=0.42025.06 | 21.5 | 40.6 | 50.7 | — | 50.8 | 112.8 | — | |
| Align and TellYear=20232026.06 | 21.3 | 40.2 | 50.1 | 11 | — | — | — | |
| VoPFParams (M)=0.12022.11 | 21.2 | 39.4 | 49.2 | 11 | 52.3 | — | — | |
| CLIP4ClipAggregation=seqTransf2022.03 | 20.8 | 39 | 48.6 | 12 | 54.2 | — | — | |
| CLIP4Clip-seqTransfBackbone=ViT-B/322022.07 | 20.8 | 39 | — | — | 54.2 | — | — | |
| CLIP4ClipPre-trained weights=CLIP (ViT-B/32), Inverted softmax=false2022.11 | 20.8 | 39 | 48.6 | 12 | — | — | — | |
| CLIP4Clip2023.03 | 20.8 | 39 | 48.6 | 12 | 54.2 | 108.4 | — | |
| CLIP4ClipAggregation=mean-pooling2022.03 | 20.6 | 39.4 | 47.5 | 13 | 56.7 | — | — | |
| CLIP4Clip-MeanPBackbone=ViT-B/322022.07 | 20.6 | 39.4 | — | — | 56.7 | — | — | |
| Full fine-tuningBackbone=CLIP (ViT-B/32), # Params (M)=123.542025.06 | 20.6 | 39.4 | 47.5 | — | 56.7 | 107.5 | — | |
| CLIP4ClipYear=20222026.06 | 20.6 | 39.4 | 47.5 | 13 | 56.7 | — | — | |
| VoPCParams (M)=14.32022.11 | 20.3 | 38.7 | 48.5 | 11 | 56.9 | — | — | |
| AdapterFFNParams (M)=2.02022.11 | 19.8 | 38.4 | 47 | 12 | 57.8 | — | — | |
| AdapterATTNParams (M)=2.02022.11 | 19.7 | 37.6 | 46.3 | 13 | 55.4 | — | — | |
| VoPPParams (M)=0.52022.11 | 19.7 | 38.9 | 48.1 | 12 | 55.4 | — | — | |
| VoPParams (M)=0.12022.11 | 18.5 | 36.1 | 45.3 | 14 | 59.5 | — | — | |
| HAT-VTRBackbone=CLIP4Clip2026.02 | 18.02 | 32.93 | 40.44 | — | — | — | — | |
| BiasParams (M)=0.12022.11 | 18 | 36 | 44.9 | 15 | 62.2 | — | — | |
| HAT-VTRBackbone=Xpool2026.02 | 17.72 | 33.33 | 41.84 | — | — | — | — | |
| TeachText-CE+2022.03 | 17.5 | 36 | 45 | 14.3 | — | — | — | |
| TT-CE+2022.07 | 17.5 | 36 | — | — | — | — | — | |
| TT-CE2023.03 | 17.5 | 36 | 45 | 14.3 | — | 98.5 | — | |
| InternVideo (ViT Backbone)Zero-shot=true, Backbone=ViT, WiSE-FT=true2023.10 | 17.5 | 29.9 | 38 | 25 | — | — | — | |
| HAT-VTRBackbone=CLIP4Clip2026.02 | 17.42 | 33.13 | 40.04 | — | — | — | — | |
| HAT-VTRBackbone=Xpool2026.02 | 17.32 | 33.53 | 42.14 | — | — | — | — | |
| ProjParams (M)=0.72022.11 | 17.1 | 32.6 | 39.9 | 21 | 76.4 | — | — | |
| TentBackbone=Xpool2026.02 | 17.02 | 33.33 | 41.24 | — | — | — | — | |
| SARBackbone=Xpool2026.02 | 16.92 | 33.33 | 41.14 | — | — | — | — | |
| EATABackbone=Xpool2026.02 | 16.92 | 33.53 | 40.84 | — | — | — | — | |
| XpoolBackbone=Xpool2026.02 | 16.82 | 33.33 | 41.24 | — | — | — | — | |
| READBackbone=Xpool2026.02 | 16.82 | 32.93 | 40.94 | — | — | — | — | |
| TCRBackbone=Xpool2026.02 | 16.82 | 33.53 | 41.04 | — | — | — | — | |
| InternVideoZero-shot=true, WiSE-FT=true2023.10 | 16 | 30.2 | 36.7 | 28 | — | — | — | |
| PartialParams (M)=7.72022.11 | 15.9 | 33.2 | 41.5 | 18 | 72.3 | — | — | |
| EATABackbone=CLIP4Clip2026.02 | 15.82 | 29.83 | 38.74 | — | — | — | — | |
| BridgeFormerZero-shot=true2023.10 | 15.5 | 30.7 | 38.7 | 22 | — | — | — | |
| TCRBackbone=Xpool2026.02 | 15.32 | 29.43 | 36.74 | — | — | — | — | |
| SARBackbone=CLIP4Clip2026.02 | 15.32 | 29.23 | 37.44 | — | — | — | — | |
| EATABackbone=Xpool2026.02 | 15.22 | 29.03 | 36.54 | — | — | — | — | |
| TentBackbone=CLIP4Clip2026.02 | 15.12 | 28.83 | 37.24 | — | — | — | — | |
| CLIP4ClipBackbone=CLIP4Clip2026.02 | 14.91 | 28.43 | 36.84 | — | — | — | — | |
| TCRBackbone=CLIP4Clip2026.02 | 14.81 | 27.93 | 37.44 | — | — | — | — | |
| TentBackbone=Xpool2026.02 | 14.81 | 28.73 | 36.94 | — | — | — | — | |
| SARBackbone=CLIP4Clip2026.02 | 14.71 | 27.03 | 34.83 | — | — | — | — | |
| SARBackbone=Xpool2026.02 | 14.61 | 28.73 | 36.84 | — | — | — | — | |
| READBackbone=CLIP4Clip2026.02 | 14.51 | 28.43 | 36.44 | — | — | — | — | |
| TentBackbone=CLIP4Clip2026.02 | 14.41 | 27.13 | 35.04 | — | — | — | — | |
| EATABackbone=CLIP4Clip2026.02 | 14.41 | 26.73 | 35.24 | — | — | — | — | |
| TCRBackbone=CLIP4Clip2026.02 | 14.41 | 30.23 | 38.04 | — | — | — | — | |
| XpoolBackbone=Xpool2026.02 | 14.31 | 28.43 | 36.74 | — | — | — | — | |
| READBackbone=Xpool2026.02 | 14.21 | 27.93 | 36.44 | — | — | — | — | |
| T2VLADPre-trained weights=BERT-Base, Inverted softmax=false2022.11 | 14.2 | 33.5 | 41.7 | 17 | — | — | — | |
| CLIP4ClipBackbone=CLIP4Clip2026.02 | 13.41 | 27.13 | 34.83 | — | — | — | — | |
| READBackbone=CLIP4Clip2026.02 | 13.41 | 26.23 | 34.63 | — | — | — | — | |
| InternVideoZero-shot=true2022.12 | 13.2 | — | — | — | — | — | — | |
| JSFusion2022.03 | 12.3 | 28.6 | 38.9 | 20 | — | — | — | |
| JSFusion2022.07 | 12.3 | 28.6 | — | — | — | — | — | |
| MMTPre-trained weights=BERT-Base, Inverted softmax=false2022.11 | 12.1 | 29.3 | 37.9 | 22.5 | — | — | — | |
| CLIPZero-shot=true2022.12 | 10.5 | — | — | — | — | — | — | |
| FrozenZero-shot=true2023.10 | 9.3 | 22 | 30.1 | 51 | — | — | — | |
| FROZENReranking (Matching Loss)=false2023.10 | 9.3 | 22 | 30.1 | 51 | — | — | — | |
| Straight-CLIP2022.03 | 6.8 | 16.4 | 22.1 | 73 | — | — | — | |
| CLIP-straight2022.07 | 6.8 | 16.4 | — | — | — | — | — | |
| NoiseEstZero-shot=true2023.10 | 4.2 | 11.6 | 17.1 | 119 | — | — | — | |
| CLIPZero-shot=true2024.03 | — | — | — | — | — | — | 11.9 | |
| ESATYear=20232026.06 | — | 31.9 | 42.2 | 18 | 71.61 | — | — | |
| InternVideo-LZero-shot=true2024.03 | — | — | — | — | — | — | 13.2 | |
| InternVideo2_s2-1BZero-shot=true2024.03 | — | — | — | — | — | — | 27.3 |