List your company

Segment-level Music-to-Video Retrieval on MusicVid-YT8M (test)

6Median RankMVPt (t = 11s)
-33.76234.62503771.38Jun 14, 2022

Evaluation results

MethodLinks
MVPt (t = 11s)Features=CLIP & DeepSim, Temporal Context=Transformers, Loss Function=InfoNCE, Segment duration (t)=11s2022.06629.3747.5265.3265.57
MVPtFeatures=CLIP & DeepSim, Temporal Context=Transformers, Loss Function=InfoNCE2022.061219.9834.8145.4142.37
Baseline + CLIP & DeepSim + Visual-time TransformersFeatures=CLIP & DeepSim, Temporal Context=Transformers (visual time)2022.062417.0726.8538.4338.54
Baseline + CLIP & DeepSim + Music-time TransformersFeatures=CLIP & DeepSim, Temporal Context=Transformers (music time)2022.062616.526.8637.1737.15
MVPt + X3D featuresFeatures=X3D, Temporal Context=Transformers, Loss Function=InfoNCE2022.06278.8319.8829.229.03
MVPt + TimeSformer featuresFeatures=TimeSformer, Temporal Context=Transformers, Loss Function=InfoNCE2022.06367.3617.5126.3426.09
MVPt (t = 4.5s)Features=CLIP & DeepSim, Temporal Context=Transformers, Loss Function=InfoNCE, Segment duration (t)=4.5s2022.065210.6216.624.5524.67
Baseline + CLIP & DeepSim featuresFeatures=CLIP & DeepSim2022.061074.718.1412.1410.91
Baseline2022.062770.742.794.834.33
Chance2022.061,0000.050.250.50.5