Multi-label Image Classification on MS-COCO (test)
83.7mAPTResNet-L
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| TResNet-LPre-training Dataset=ImageNet-21K-P2021.04 | 83.7 | — | — | — | — | — | |
| TResNet-MPre-training Dataset=ImageNet-21K-P2021.04 | 82.2 | — | — | — | — | — | |
| TResNet-LPre-training Dataset=ImageNet-1K2021.04 | 81.1 | — | — | — | — | — | |
| ResNet50Pre-training Dataset=ImageNet-21K-P2021.04 | 80.5 | — | — | — | — | — | |
| TResNet-MPre-training Dataset=ImageNet-1K2021.04 | 79.5 | — | — | — | — | — | |
| OFA595Pre-training Dataset=ImageNet-21K-P2021.04 | 77.7 | — | — | — | — | — | |
| ResNet50Pre-training Dataset=ImageNet-1K2021.04 | 76.7 | — | — | — | — | — | |
| MobileNetV3Pre-training Dataset=ImageNet-21K-P2021.04 | 74.9 | — | — | — | — | — | |
| OFA595Pre-training Dataset=ImageNet-1K2021.04 | 74.9 | — | — | — | — | — | |
| S-Clsknowledge distillation=true2018.09 | 74.6 | 69.2 | 74 | 66.8 | 72.7 | — | |
| MobileNetV3Pre-training Dataset=ImageNet-1K2021.04 | 73 | — | — | — | — | — | |
| FitNets2018.09 | 72.5 | 65.2 | 70.9 | 62.3 | 68.3 | — | |
| RMAMresolution=512, cropping=10crop2018.09 | 72.2 | — | — | 66.5 | 71.3 | — | |
| Attention transfer2018.09 | 71.4 | 64.6 | 69.8 | 61.6 | 67.8 | — | |
| Distillation2018.09 | 71.3 | 64.7 | 69.3 | 61.5 | 67.6 | — | |
| S-Clsknowledge distillation=false2018.09 | 70.9 | 63.6 | 67 | 60.7 | 66.7 | — | |
| RLSD2018.09 | 68.2 | — | — | 62 | 66.5 | — | |
| MIML-FCN-BB2018.09 | 66.2 | — | — | — | — | — | |
| MCG-CNN-LSTM2018.09 | 64.4 | — | — | 58.1 | 61.3 | — | |
| ADDSBackbone=ViT-L-336, Input Resolution=448x4482022.08 | 59.18 | — | — | — | — | 77.34 | |
| ADDSBackbone=ViT-L-336, Input Resolution=336x3362022.08 | 54.52 | — | — | — | — | 51.52 | |
| ML-DecoderBackbone=ViT-L-336 (f), Input Resolution=336x3362022.08 | 43.84 | — | — | — | — | 35.08 | |
| ML-DecoderBackbone=ViT-L-336 (uf), Input Resolution=336x3362022.08 | 43.75 | — | — | — | — | 17.09 | |
| ML-DecoderBackbone=TresNet-L, Input Resolution=448x4482022.08 | 30.69 | — | — | — | — | 16.69 | |
| CNN-LSEP2018.09 | — | 62.9 | 68.3 | — | — | — | |
| CNN-RNN2018.09 | — | — | — | 60.4 | 67.8 | — | |
| CNN-SREL-RNN2018.09 | — | 63.4 | 72.5 | — | — | — | |
| RARLFresolution=512, cropping=10crop2018.09 | — | — | — | 65.6 | 70.5 | — |