Multi-label Recognition on NUS-WIDE (mAP)
52.78mAPBCP
Evaluation Results
| Method | Links | |
|---|---|---|
| BCPBackbone=ViT-B/16, BP-free=true2026.06 | 52.78 | |
| BCPBackbone=ViT-B/32, BP-FREE=true2026.06 | 52.71 | |
| BCE-LSSupervision=Fully Supervised, Label=Full, Backbone=ResNet-1012026.06 | 51.7 | |
| BCPBackbone=RN-101, BP-FREE=true2026.06 | 50.84 | |
| BCESupervision=Fully Supervised, Label=Full, Backbone=ResNet-1012026.06 | 50.7 | |
| BCPBackbone=RN-50, BP-free=true2026.06 | 50.34 | |
| G2NetPLSupervision=Weakly Supervised, Label=One, Backbone=ResNet-1012026.06 | 48.5 | |
| SCABackbone=ViT-B/16, BP-free=true2026.06 | 48.08 | |
| SSGBackbone=ViT-B/16, BP-free=true2026.06 | 47.93 | |
| LL-RSupervision=Weakly Supervised, Label=One, Backbone=ResNet-1012026.06 | 47.4 | |
| STSBackbone=ViT-B/16, BP-free=false2026.06 | 47.23 | |
| TDABackbone=ViT-B/16, BP-free=true2026.06 | 46.72 | |
| BEMBackbone=ViT-B/16, BP-free=false2026.06 | 46.55 | |
| DMNBackbone=ViT-B/16, BP-free=true2026.06 | 46.27 | |
| TPTBackbone=ViT-B/16, BP-free=false2026.06 | 46.15 | |
| DiffTPTBackbone=ViT-B/16, BP-free=false2026.06 | 46.13 | |
| DARTBackbone=ViT-B/16, BP-free=false2026.06 | 45.91 | |
| CLIPBackbone=ViT-B/16, BP-free=true2026.06 | 45.65 | |
| TailorCLIPSupervision=Unsupervised, Label=None, Backbone=ResNet-1012026.06 | 45.6 | |
| TailorCLIPSupervision=Unsupervised, Label=None, Backbone=ViT-L/142026.06 | 44.6 | |
| CCDSupervision=Unsupervised, Label=None, Backbone=ResNet-1012026.06 | 44.5 | |
| TDABackbone=ViT-B/32, BP-FREE=true2026.06 | 44.13 | |
| BEMBackbone=ViT-B/32, BP-FREE=false2026.06 | 44.12 | |
| CDULSupervision=Unsupervised, Label=None, Backbone=ResNet-1012026.06 | 44 | |
| TPTBackbone=ViT-B/32, BP-FREE=false2026.06 | 43.63 | |
| BEMBackbone=RN-101, BP-FREE=false2026.06 | 43.62 | |
| DualCoOpSupervision=Unsupervised, Label=None, Backbone=ResNet-1012026.06 | 43.6 | |
| DIFFTPTBackbone=ViT-B/32, BP-FREE=false2026.06 | 43.42 | |
| DMNBackbone=ViT-B/32, BP-FREE=true2026.06 | 43.41 | |
| RLCFBackbone=ViT-B/32, BP-FREE=false2026.06 | 43.29 | |
| ROLESupervision=Weakly Supervised, Label=One, Backbone=ResNet-1012026.06 | 43.2 | |
| RLCFBackbone=ViT-B/16, BP-free=false2026.06 | 43.18 | |
| ADAPTBackbone=ViT-B/16, BP-free=true2026.06 | 43.18 | |
| TDABackbone=RN-101, BP-FREE=true2026.06 | 43.13 | |
| TPTBackbone=RN-101, BP-FREE=false2026.06 | 43.1 | |
| DIFFTPTBackbone=RN-101, BP-FREE=false2026.06 | 42.93 | |
| CLIPBackbone=ViT-B/32, BP-FREE=true2026.06 | 42.9 | |
| DMNBackbone=RN-101, BP-FREE=true2026.06 | 42.71 | |
| BEMBackbone=RN-50, BP-free=false2026.06 | 42.53 | |
| TDABackbone=RN-50, BP-free=true2026.06 | 42.34 | |
| TPTBackbone=RN-50, BP-free=false2026.06 | 41.97 | |
| CLIPBackbone=RN-101, BP-FREE=true2026.06 | 41.93 | |
| CLIPBackbone=RN-50, BP-free=true2026.06 | 41.53 | |
| DiffTPTBackbone=RN-50, BP-free=false2026.06 | 41.33 | |
| DMNBackbone=RN-50, BP-free=true2026.06 | 41.32 | |
| Partial BCESupervision=Weakly Supervised, Label=10%, Backbone=ResNet-1012026.06 | 39.4 | |
| DOTABackbone=RN-50, BP-free=true2026.06 | 38.88 | |
| DOTABackbone=ViT-B/32, BP-FREE=true2026.06 | 33.6 | |
| DOTABackbone=ViT-B/16, BP-free=true2026.06 | 32.74 | |
| RLCFBackbone=RN-101, BP-FREE=false2026.06 | 31.77 | |
| RLCFBackbone=RN-50, BP-free=false2026.06 | 29.83 | |
| DOTABackbone=RN-101, BP-FREE=true2026.06 | 19.73 | |
| BCE-LSSupervision level=Fully supervised, Annotation=Fully labeled, Frozen=false2026.05 | 0.517 | |
| AEVLPSupervision level=Weakly supervised, Annotation=Single positive labeled, Frozen=false2026.05 | 0.507 | |
| PIAASupervision level=Training free, Annotation=Annotation free, Frozen=true2026.05 | 0.506 | |
| VLPLSupervision level=Weakly supervised, Annotation=Single positive labeled, Frozen=false2026.05 | 0.496 | |
| GR LossSupervision level=Weakly supervised, Annotation=Single positive labeled, Frozen=false2026.05 | 0.491 | |
| G2 NetPLSupervision level=Weakly supervised, Annotation=Single positive labeled, Frozen=false2026.05 | 0.485 | |
| SPARCSupervision level=Training free, Annotation=Annotation free, Frozen=true2026.05 | 0.475 | |
| LL-RSupervision level=Weakly supervised, Annotation=Single positive labeled, Frozen=false2026.05 | 0.474 | |
| CCDSupervision level=Unsupervised, Annotation=Annotation free, Frozen=false2026.05 | 0.445 | |
| CLIPSupervision level=Training free, Annotation=Annotation free, Frozen=true2026.05 | 0.444 | |
| CDULSupervision level=Unsupervised, Annotation=Annotation free, Frozen=false2026.05 | 0.44 | |
| ROLESupervision level=Unsupervised, Annotation=Annotation free, Frozen=false2026.05 | 0.432 | |
| NaiveANSupervision level=Unsupervised, Annotation=Annotation free, Frozen=false2026.05 | 0.408 | |
| TagCLIPSupervision level=Training free, Annotation=Annotation free, Frozen=true2026.05 | 0.387 |