List your company
ResearchFormerly WizWand

Interaction Recognition on ImageNet-1k 1 object

89Interaction RecognitionMaskSigLIP
27.6443.5759.575.43May 7, 2026
Updated 6mo ago

Evaluation results

MethodLinks
MaskSigLIPModel (Size)=SigLIP-2 (ViT-L/16)2026.0589
METAGAMEModel (Size)=SigLIP-2 (ViT-B/32), Base Method=MaskSigLIP2026.0589
METAGAMEModel (Size)=SigLIP-2 (ViT-L/16), Base Method=MaskSigLIP2026.0588
FIxLIPModel (Size)=SigLIP-2 (ViT-B/32)2026.0587
METAGAMEModel (Size)=CLIP (ViT-B/16), Explanation Basis=MaskCLIP2026.0586
METAGAMEModel (Size)=MetaCLIP-2 (ViT-H/14), Explanation Basis=Attention2026.0585
METAGAMEModel (Size)=CLIP (ViT-B/16), Explanation Basis=Attention2026.0584
AttentionModel (Size)=MetaCLIP-2 (ViT-H/14)2026.0584
FIxLIPModel (Size)=SigLIP-2 (ViT-L/16)2026.0580
METAGAMEModel (Size)=MetaCLIP-2 (ViT-H/14), Explanation Basis=Grad-ECLIP2026.0579
AttentionModel (Size)=CLIP (ViT-B/16)2026.0577
METAGAMEModel (Size)=CLIP (ViT-B/16), Explanation Basis=Grad-ECLIP2026.0576
FIxLIPModel (Size)=CLIP (ViT-B/16)2026.0576
Grad-ECLIPModel (Size)=MetaCLIP-2 (ViT-H/14)2026.0576
METAGAMEModel (Size)=SigLIP-2 (ViT-L/16), Base Method=Grad-ESigLIP2026.0575
Grad-ECLIPModel (Size)=CLIP (ViT-B/16)2026.0571
METAGAMEModel (Size)=SigLIP-2 (ViT-B/32), Base Method=Grad-ESigLIP2026.0571
FIxLIPModel (Size)=MetaCLIP-2 (ViT-H/14)2026.0570
Grad-ESigLIPModel (Size)=SigLIP-2 (ViT-B/32)2026.0568
METAGAMEModel (Size)=SigLIP-2 (ViT-L/16), Base Method=Attention2026.0560
METAGAMEModel (Size)=SigLIP-2 (ViT-B/32), Base Method=Attention2026.0557
Grad-ESigLIPModel (Size)=SigLIP-2 (ViT-L/16)2026.0554
MaskSigLIPModel (Size)=SigLIP-2 (ViT-B/32)2026.0553
AttentionModel (Size)=SigLIP-2 (ViT-B/32)2026.0550
AttentionModel (Size)=SigLIP-2 (ViT-L/16)2026.0547
MaskCLIPModel (Size)=CLIP (ViT-B/16)2026.0530