Demo2Vec

Reported on 5 benchmarks across 1 task

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision5 results

Video-to-image Affordance GroundingonOPRA (28x28)
AUC-J
0.85
best: 0.89 (Afformer)
Video-to-image Affordance GroundingonOPRA (28x28)
KLD
1.2
best: 1.47 (Hotspot)
Video-to-image Affordance GroundingonOPRA (28x28)
SIM
0.48
best: 0.53 (Afformer)
Video-to-image Affordance GroundingonOPRA
KLD
2.34
Video-to-image Affordance GroundingonOPRA
Top-1 Action Accuracy
40.79
best: 52.27 (Afformer (ViTDet-B encoder))