Multi-label Classification on NUS-WIDE
70.1mAPQ2L
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Q2LBackbone=CvT-w24, Pre-trained=ImageNet-22k2021.07 | 70.1 | 67.6 | 76.3 | |
| ML-Decoder + AAMBackbone=TResNet-L, Input resolution=448x448, GFLOPS=36.162022.09 | 68.3 | — | — | |
| GAT re-weightingBackbone=TResNet-L, Input resolution=448x448, GFLOPS=35.22022.09 | 68.1 | — | — | |
| ML-Decoder + AAMBackbone=EfficientNet-V2-s, Input resolution=448x448, GFLOPS=12.282022.09 | 67.6 | — | — | |
| ML-DecoderBackbone=EfficientNet-V2-s, Input resolution=448x448, GFLOPS=12.28, Training strategy=our training strategy2022.09 | 67.07 | — | — | |
| GAT re-weightingBackbone=EfficientNet-V2-s, Input resolution=448x448, GFLOPS=10.832022.09 | 66.85 | — | — | |
| Q2LBackbone=TresNetL2021.07 | 66.3 | 64 | 75 | |
| MlTr-1Backbone=MlTr-1, Pre-trained=ImageNet-22k2021.07 | 66.3 | 65 | 75.8 | |
| Q2LBackbone=TResNet-L, Input resolution=448x448, GFLOPS=60.42022.09 | 66.3 | — | — | |
| ML-GCNBackbone=EfficientNet-V2-s, Input resolution=448x448, GFLOPS=10.83, Training strategy=our training strategy2022.09 | 66.3 | — | — | |
| Q2LBackbone=EfficientNet-V2-s, Input resolution=448x448, GFLOPS=16.25, Training strategy=our training strategy2022.09 | 65.79 | — | — | |
| ASLBackbone=TresNetL2021.07 | 65.2 | 63.6 | 75 | |
| ASLBackbone=TResNet-L, Input resolution=448x448, GFLOPS=43.52022.09 | 65.2 | — | — | |
| ASLBackbone=EfficientNet-V2-s, Input resolution=448x448, GFLOPS=10.83, Training strategy=our training strategy2022.09 | 65.2 | — | — | |
| Q2LBackbone=ResNet1012021.07 | 65 | 63.1 | 75 | |
| Focal lossBackbone=TresNetL2021.07 | 64 | 62.9 | 74.7 | |
| BaselineBackbone=TresNetL2021.07 | 63.1 | 61.7 | 74.6 | |
| ICMEBackbone=ResNet1012021.07 | 62.8 | 60.7 | 74.1 | |
| SRNBackbone=ResNet1012021.07 | 62 | 58.5 | 73.4 | |
| MS-CMABackbone=ResNet1012021.07 | 61.4 | 60.5 | 73.8 | |
| GATNBackbone=ResNeXt-101, Input resolution=448x448, GFLOPS=362022.09 | 59.8 | — | — | |
| FedMPTFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 57.17 | 56.94 | 71.83 | |
| Fed-MaPLeFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 53.13 | 53.28 | 68.65 | |
| BCPFT-FREE=true, EKE-FREE=true, OTTA=true, Backbone=ViT-B/162026.06 | 52.78 | — | — | |
| FedMVPFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 52.73 | 52.61 | 70.67 | |
| FedMPTMask=10%, Venue=Ours2026.05 | 51.72 | 45.16 | 72.88 | |
| FedPGPFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 51.1 | 53.07 | 67.72 | |
| Fed-RAMFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 50.52 | 42.69 | 66.79 | |
| BCPFT-FREE=true, EKE-FREE=true, OTTA=true, Backbone=ResNet-502026.06 | 50.34 | — | — | |
| Fed-DualCoOpFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 48.65 | 50.39 | 67.52 | |
| CoMCFT-FREE=false, EKE-FREE=false, OTTA=false, Backbone=ResNet-502026.06 | 48.2 | — | — | |
| Fed-PosCoOpFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 47.44 | 46.03 | 64.47 | |
| SPARCFT-FREE=true, EKE-FREE=false, OTTA=false, Backbone=ViT-B/162026.06 | 47.3 | — | — | |
| FedAWAFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 44.4 | 47.78 | 64.75 | |
| Fed-TCPFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 42.9 | 44.77 | 64.3 | |
| FedTPGFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 42.14 | 40.36 | 65.3 | |
| Fed-DualCoOpMask=10%, Venue=NeurIPS’222026.05 | 40.13 | 35.57 | 61.16 | |
| Fed-SCPNetFramework=Federated Zero-Shot Learning, Zero-shot=true2026.05 | 38.35 | 41.62 | 61.61 | |
| Fed-SCPNetMask=10%, Venue=CVPR’232026.05 | 35.44 | 33.22 | 56.39 | |
| FedMPTMask=Avg, Venue=Ours2026.05 | 32.51 | 32.78 | 47.86 |