List your company

Long-form Video Understanding on LVU

76.47Relation Attribute AccuracySVT SCALE
48.941256.088163.23570.3819Apr 12, 2023Aug 6, 2023Dec 1, 2023Mar 26, 2024Jul 21, 2024Nov 14, 2024Mar 11, 2025

Evaluation results

MethodLinks
SVT SCALEEvaluation Protocol=Linear probe2023.0476.4742.2774.0249.0958.9762.7639.23———
Movies2ScenesEvaluation Protocol=Linear probe2023.0471.242.268.270.955.953.757.8———
Movies2Scene2025.0371.242.268.2——————60.5
HierarQLLM finetuning=true2025.0369.449.385.1——————67.9
HierarQLLM finetuning=false2025.0367.948.783.8——————66.8
S52024.0467.142.173.567.365.451.348——59.2
S52024.1167.142.173.567.365.451.348——59.2
S52025.0367.142.173.5——————60.9
AdaCM²2024.1163.140.286.275.4687762.5——67.5
VideoMambae2e=true, Backbone=VM-Ti, frames=322024.0362.540.4370.3767.2965.2452.9848.23262.9—
VideoMamba2025.0362.540.470.4——————57.8
LF-VILA2025.0361.541.368——————56.9
LSSD (B.3)Representations=Video-only, Evaluation Protocol=Linear probe2023.0460.9846.0363.4169.1660.8848.8152.48———
LSSD (B.4)Representations=Video-only, Evaluation Protocol=Linear probe2023.0460.9847.6265.8567.2961.7341.6755.32———
TranS4mer2025.0359.539.270.9——————56.5
STCAEvaluation Protocol=Linear probe2023.0459.2541.6269.1566.756.6252.9353.3———
LSSD (B.5)Representations=Video-only, Evaluation Protocol=Linear probe2023.0458.5446.0364.6369.1664.2941.0752.48———
MA-LMM2024.0458.244.880.374.66170.451.9——63
MA-LMM2024.1158.244.880.374.66170.451.9——63
MA-LMM2025.0358.244.880.3——————61.1
ViS4mere2e=false, Backbone=ViT-L2024.0357.1440.7967.4462.6154.7148.844.75263.63—
ViS4merEvaluation Protocol=Linear probe2023.0457.1440.7967.4462.6154.7148.844.75———
VIS4mer2024.0457.140.867.462.654.748.844.8——53.7
VIS4mer2024.1157.140.867.462.654.748.844.8——53.7
VIS4mer2025.0357.140.867.4——————55.1
Obj_T4mer2024.0454.833.252.947.752.736.337.8——45
Obj-T4mer2024.1154.833.252.947.752.736.337.8——45.1
Obj_T4mer2025.0354.833.252.9——————47
Object TransformerEvaluation Protocol=Linear probe2023.0454.740.36058.956.135.140.6———
Object Trans.e2e=false, Backbone=ResNet2024.0353.139.456.951.254.634.539.1233.55—
VideoBERTe2e=false, Backbone=S3D2024.0352.837.954.947.351.938.536.1324.46—
VideoBERT2024.0452.837.954.947.351.938.536.1——45.6
VideoBERT2024.1152.837.954.947.351.938.536.1——45.6
VideoBERT2025.0352.837.954.9——————48.5
LST2024.0452.537.362.856.152.742.339.2——49
LST2024.1152.537.362.856.152.742.339.2——49
LSTe2e=false, Backbone=ViT-L2024.0352.3837.3162.7956.0752.742.2639.16313.83—
Performere2e=false, Backbone=ViT-L2024.035038.860.4658.8749.4548.2141.25313.93—
Orthoformere2e=false, Backbone=ViT-L2024.035039.366.2755.1455.7947.0243.35293.86—
Performer2024.045038.860.558.949.548.241.3——49.6
Orthoformer2024.045038.366.355.155.84743.4——50.8
Performer2024.115038.860.558.949.548.241.3——49.6
Orthoformer2024.115038.366.355.155.84743.4——50.8
Orthoformer2025.035038.366.3——————51.5