List your company

Video-to-Text Retrieval on DiDeMo (test)

67.5R@1VidLA
13.52427.53741.5555.563Mar 9, 2021Jan 13, 2022Nov 19, 2022Sep 25, 2023Jul 31, 2024Jun 6, 2025Apr 12, 2026

Evaluation results

MethodLinks
VidLABackbone=CLIP-ViT-B/16, dual-softmax=true2024.0367.591.7——87.582.2
VidLABackbone=CLIP-ViT-B/32, dual-softmax=true2024.0362.291.1——85.179.5
VidLABackbone=CLIP-ViT-B/162024.0359.690.2——84.478
UMTBackbone=CLIP-ViT-B/16, two-stage method with candidate re-ranking=true2024.0359.590.5——84.978.3
InternVideoBackbone=ViT-L, dual-softmax=true2024.0359.1—————
InternVideo2 6Bretrieval_stage=Dual, model_scale=6B, zero_shot=true2026.0157.1———79.9—
VidLABackbone=CLIP-ViT-B/322024.035588.9——82.875.6
VIRTUE v1 7Bretrieval_stage=Dual, model_scale=7B, zero_shot=true2026.0152.5———79—
DiffusionRetBackbone=CLIP-ViT-B/32, dual-softmax=true2024.0350.382.9——75.169.4
DRLBackbone=CLIP-ViT-B/162024.0349.983.3——75.469.5
NeighborRetr2025.0348.483.72974.6—
PAU2023.0948.185.729.874.2—
X-CLIPBackbone=ViT-B/162022.0747.8——10.576.8—
RAPType=Adapter, Frozen visual encoder=true2024.0547.783.2—9.574.4—
CLIP4Clip-MeanPBackbone=ViT-B/162022.0747.2——10.574—
VIRTUE v2 7Bretrieval_stage=Dual, model_scale=7B, zero_shot=true2026.0147———75.8—
BaselineCondition=without PAU2023.0946.984.228.474.4—
HBI2023.0346.282.728.773—
Diffusion2025.0346.282.2210.774.3—
HBI2025.0346.282.728.773—
UCOFIA2023.094681.5212.171.9—
EMCL-Net2025.0345.782.7210.974.3—
CAMOEBackbone=CLIP-ViT-B/32, dual-softmax=true2024.0345.5———71.2—
DRLBackbone=CLIP-ViT-B/322024.0345.482.1——72.666.7
CLIP4Clip-seqTransfBackbone=ViT-B/162022.0744.7——10.674—
VoPF+PParams (M)=0.4 (0.328%)2022.1144.781.129.971.2—
VoPF+CParams (M)=14.1 (11.785%)2022.1144.481.829.571.8—
RAPType=Adapter, Frozen visual encoder=false2024.054480.1—10.569.2—
VoPFParams (M)=0.1 (0.103%)2022.1143.581.429.870.9—
CLIP4Clip-meanPAggregation Strategy=mean pooling2023.0943.480.2217.569.9—
X-CLIPBackbone=ViT-B/322022.0743.1——10.972.2—
X-CLIP2023.0943.1——10.972.2—
X-Clip2025.0343.1——10.977.2—
CLIP4Clip-seqTransfAggregation Strategy=sequential Transformer2023.0942.780.6211.670.9—
CLIP4Clip-meanPTraining Dataset=WIT + DiDeMo, End-to-End=true2021.0442.580.2211.670.6—
CLIP4Clip-MeanPBackbone=ViT-B/322022.0742.5——11.670.6—
CLIP4Clip2023.0942.580.2211.670.6—
CLIP4ClipBackbone=CLIP-ViT-B/322024.0342.580.2——70.664.4
CLIP4Clip-seqLSTMTraining Dataset=WIT + DiDeMo, End-to-End=true2021.0442.479.2211.869.2—
CLIP4ClipType=Fine-tune, Frozen visual encoder=false2024.0542.279.3—11.870.3—
VIRTUE-Embed 7Bretrieval_stage=Single, model_category=MLLM-based, model_scale=7B, zero_shot=true2026.0142.2———68.5—
CLIP4Clip-seqTransfTraining Dataset=WIT + DiDeMo, End-to-End=true2021.0441.479.1212.468.2—
CLIP4Clip-seqTransfBackbone=ViT-B/322022.0741.4——12.468.2—
CLIP4Clip2023.0341.479.1212.468.2—
VoPPParams (M)=0.5 (0.441%)2022.1140.678.6211.668.3—
VoPPType=Prompt2024.0540.678.6—11.668.3—
FullParams (M)=119.8 (100%)2022.1140.278.7211.968.4—
SSFType=Adapter2024.054077.4—13.267.1—
LORAType=Adapter2024.0539.777.3—13.966.8—
VoPCParams (M)=14.3 (11.898%)2022.1139.176.7313.865.3—
VoPCType=Prompt2024.0539.176.7—13.865.3—
CaRe 7Bretrieval_stage=Single, model_category=MLLM-based, model_scale=7B, zero_shot=true2026.0139.1———66—
VoPParams (M)=0.1 (0.103%)2022.1138.176.5213.565.7—
PartialParams (M)=7.7 (6.410%)2022.1136.974.521764.2—
BiasParams (M)=0.1 (0.104%)2022.1136.875.8215.165.7—
AdapterATTNParams (M)=2.0 (1.655%)2022.1136.374.8215.464.4—
ST-AdapterType=Adapter2024.0535.972—20.161—
AdapterFFNParams (M)=2.0 (1.655%)2022.1135.675.6314.864.3—
LanguageBindretrieval_stage=Single, model_category=CLIP-based, zero_shot=true2026.0135.6———62.8—
ProjParams (M)=0.7 (0.547%)2022.1134.572.6318.860.9—
VPTType=Prompt2024.0533.169.9—22.759.8—
LaTzero-shot=false2022.0732.772.73—61.1—
FullTriplets (Ratio)=-, Time (mins)=-2026.0432.7———58.6—
CoOpType=Prompt2024.0532.368.2—23.457—
HoPATriplets (Ratio)=200 (2.38%), Time (mins)=45.72026.0431.8———57.2—
HoPATriplets (Ratio)=100 (1.19%), Time (mins)=41.92026.0431.5———56.6—
HoPATriplets (Ratio)=500 (5.96%), Time (mins)=53.62026.0431.3———57.3—
LoRSTriplets (Ratio)=500 (5.96%), Time (mins)=122.32026.0430.3———56.6—
RepBlendTriplets (Ratio)=500 (5.96%), Time (mins)=122.82026.0430———56.1—
RandomTriplets (Ratio)=500 (5.96%), Time (mins)=-2026.0429.9———54.9—
MTT-VLTriplets (Ratio)=500 (5.96%), Time (mins)=120.22026.0429.5———56.3—
LoRSTriplets (Ratio)=200 (2.38%), Time (mins)=96.32026.0429.2———54.8—
TESLATriplets (Ratio)=500 (5.96%), Time (mins)=118.92026.0429.1———55.9—
RepBlendTriplets (Ratio)=200 (2.38%), Time (mins)=97.42026.0429———55—
Long-CLIP L/14retrieval_stage=Single, model_category=CLIP-based, model_scale=L/14, zero_shot=true2026.0128.5———54.1—
RandomTriplets (Ratio)=200 (2.38%), Time (mins)=-2026.0428.3———51.4—
LoRSTriplets (Ratio)=100 (1.19%), Time (mins)=79.42026.0428———52.7—
MTT-VLTriplets (Ratio)=200 (2.38%), Time (mins)=95.02026.0427.9———53.2—
TESLATriplets (Ratio)=200 (2.38%), Time (mins)=97.42026.0427.8———53.5—
RepBlendTriplets (Ratio)=100 (1.19%), Time (mins)=78.22026.0427.7———52.9—
RandomTriplets (Ratio)=100 (1.19%), Time (mins)=-2026.0427.4———50.3—
MTT-VLTriplets (Ratio)=100 (1.19%), Time (mins)=77.32026.0426.6———51.5—
TESLATriplets (Ratio)=100 (1.19%), Time (mins)=78.12026.0426.6———51.5—
ATP2022.0626.1———50.5—
LaTzero-shot=true2022.0722.556.87—45.2—
CLIP4Clip-tightTransfTraining Dataset=WIT + DiDeMo, End-to-End=true2021.0421.564.8522.451.1—
TT-CE+Training Dataset=DiDeMo, End-to-End=false2021.0421.161.16.3—47.3—
TT-CE+2022.0721.1———47.3—
TT-CE2023.0321.161.1—6.347.3—
TT-CE2023.0921.161.16.3—47.3—
ClipBERT2022.0620.4———48—
CLIPinput_format=single-frame2022.0620.2———42.5—
CLIP4ClipType=Fine-tune, Frozen visual encoder=true2024.0520.255—43.144.2—
MMLoss=MM2021.031752.31042.6——
VideoCLIPzero-shot=true2022.0616.6———46.9—
MMLoss=MM [19]2021.0315.6—8.242.4——
CETraining Dataset=DiDeMo, End-to-End=false2021.0415.6—8.242.440.9—
CE2022.0715.6——42.440.9—
CE2023.0315.6—8.242.440.9—
CE2023.0915.6—8.242.440.9—