Fine-grained Visual Categorization on Stanford Cars (test)
96.3AccuracyRelational Proxy
Evaluation results
| Method | Links | |
|---|---|---|
| Relational ProxyBackbone=ResNet-50, Local views (k)=7, Averaged over runs=52022.10 | 96.3 | |
| Proposed (CAP)Transfer/Joint Learning Strategy=false2021.01 | 95.7 | |
| CAP2022.10 | 95.7 | |
| CSQA-NetBackbone=ResNet-502024.03 | 95.6 | |
| CSQA-NetBackbone=Swin-B2024.03 | 95.6 | |
| CALBackbone=ResNet1012022.05 | 95.5 | |
| SIA-NetBackbone=ResNet-502024.03 | 95.5 | |
| P2P-NetBackbone=ResNet-502024.03 | 95.4 | |
| PMRCBackbone=ResNet-502024.03 | 95.4 | |
| API-NetBackbone=DenseNet1612022.05 | 95.3 | |
| Baseline + DCALBackbone=R50-ViT-Base2022.05 | 95.3 | |
| PARTType=Joint, Backbone=ResNet1012022.12 | 95.3 | |
| API-NetBackbone=DenseNet-1612024.03 | 95.3 | |
| DCALBackbone=R50-ViT-B2024.03 | 95.3 | |
| PMGBackbone=ResNet502022.05 | 95.1 | |
| API-NetType=Feature Based, Backbone=ResNet1012022.12 | 95.1 | |
| PMGType=Feature Based, Backbone=ResNet502022.12 | 95.1 | |
| PARTType=Joint, Backbone=ResNet502022.12 | 95.1 | |
| MMAL2022.10 | 95 | |
| DPNBackbone=ResNet-502024.03 | 95 | |
| ViT-NeTBackbone=Swin-B2024.03 | 95 | |
| SPSBackbone=ResNet502022.05 | 94.9 | |
| CMNBackbone=ResNet-502024.03 | 94.9 | |
| ACC-ViTBackbone=ViT-B-162024.03 | 94.9 | |
| AutoAugTransfer/Joint Learning Strategy=true2021.01 | 94.8 | |
| TransFGBackbone=ViT-Base2022.05 | 94.8 | |
| TransFG2022.10 | 94.8 | |
| API-NetType=Feature Based, Backbone=ResNet502022.12 | 94.8 | |
| TransFGBackbone=ViT-B-162024.03 | 94.8 | |
| S3NBackbone=ResNet502022.05 | 94.7 | |
| GPipeTransfer/Joint Learning Strategy=false2021.01 | 94.6 | |
| Cross-XBackbone=ResNet502022.05 | 94.6 | |
| DCLTransfer/Joint Learning Strategy=false2021.01 | 94.5 | |
| DCLBackbone=ResNet502022.05 | 94.5 | |
| CINBackbone=ResNet502022.05 | 94.5 | |
| LIOBackbone=ResNet502022.05 | 94.5 | |
| DBTNet2022.10 | 94.5 | |
| CINType=Feature Based, Backbone=ResNet1012022.12 | 94.5 | |
| Cross-XType=Feature Based, Backbone=ResNet502022.12 | 94.5 | |
| DCLType=Feature Based, Backbone=ResNet502022.12 | 94.5 | |
| DCLBackbone=ResNet-502024.03 | 94.5 | |
| MCLossTransfer/Joint Learning Strategy=false2021.01 | 94.4 | |
| SPSBackbone=ResNet-502024.03 | 94.4 | |
| BARMTransfer/Joint Learning Strategy=false2021.01 | 94.3 | |
| SwinTransBackbone=Swin-B2024.03 | 94.2 | |
| CINType=Feature Based, Backbone=ResNet502022.12 | 94.1 | |
| FDLBackbone=DenseNet1612022.05 | 94 | |
| BaselineBackbone=R50-ViT-Base2022.05 | 94 | |
| NTS-NetBackbone=ResNet502022.05 | 93.9 | |
| MGE-CNNBackbone=ResNet502022.05 | 93.9 | |
| NTSNetType=Part Based, Backbone=ResNet502022.12 | 93.9 | |
| AKENBackbone=VGG-192024.03 | 93.9 | |
| MaxEnt2022.10 | 93.85 | |
| TASNBackbone=ResNet502022.05 | 93.8 | |
| Baseline + DCALBackbone=DeiT-Base2022.05 | 93.8 | |
| DFL-CNNType=Feature Based, Backbone=ResNet502022.12 | 93.8 | |
| MpT-TransBackbone=ViT-B-162024.03 | 93.8 | |
| HBPType=Feature Based, Backbone=VGG-162022.12 | 93.7 | |
| TASNType=Feature Based, Backbone=ResNet502022.12 | 93.7 | |
| Baseline + DCALBackbone=ViT-Base2022.05 | 93.4 | |
| iSQRT-COVBackbone=ResNet-101, Dim.=32K2017.12 | 93.3 | |
| iSQRT-COV-Net with ResNet-101Backbone=ResNet-1012019.04 | 93.3 | |
| P-CNNBackbone=VGG-192024.03 | 93.3 | |
| MAMCBackbone=ResNet1012022.05 | 93 | |
| MaxEntType=Feature Based, Backbone=DenseNet-1612022.12 | 93 | |
| MAMCType=Feature Based, Backbone=ResNet1012022.12 | 93 | |
| PCBackbone=DenseNet1612022.05 | 92.9 | |
| BaselineBackbone=DeiT-Base2022.05 | 92.9 | |
| PCType=Feature Based, Backbone=DenseNet-1612022.12 | 92.9 | |
| iSQRT-COVBackbone=ResNet-50, Dim.=32K2017.12 | 92.8 | |
| iSQRT-COV-Net (32K)Backbone=ResNet-50, Feature Dimension=32K2019.04 | 92.8 | |
| MA-CNNBackbone=VGG192022.05 | 92.8 | |
| PA-CNNType=Part Based, Backbone=VGG-19, Additional Bounding Box or Segmentation Annotation=true2022.12 | 92.8 | |
| MA-CNNType=Part Based, Backbone=VGG-192022.12 | 92.8 | |
| GrassmannType=Feature Based, Backbone=VGG-162022.12 | 92.8 | |
| iSQRT-COVBackbone=VGG-D, Dim.=32K2017.12 | 92.5 | |
| G2DeNetBackbone=VGG-D, Dim.=263K2017.12 | 92.5 | |
| G2DeNetBackbone=VGG-D162019.04 | 92.5 | |
| iSQRT-COV-NetBackbone=VGG-D162019.04 | 92.5 | |
| RA-CNNBackbone=VGG192022.05 | 92.5 | |
| BaselineBackbone=ViT-Base2022.05 | 92.5 | |
| KPBackbone=VGG-D, Dim.=13K2017.12 | 92.4 | |
| KPBackbone=VGG-D162019.04 | 92.4 | |
| Baseline + DCALBackbone=DeiT-Small2022.05 | 92.3 | |
| MPN-COV-NetBackbone=VGG-D162019.04 | 92.2 | |
| Improved B-CNNBackbone=VGG-D, Dim.=262K2017.12 | 92 | |
| Improved B-CNNBackbone=VGG-D162019.04 | 92 | |
| iSQRT-COVBackbone=ResNet-50, Dim.=8K2017.12 | 91.7 | |
| HIHCABackbone=VGG-D, Dim.=9K2017.12 | 91.7 | |
| HIHCABackbone=VGG-D162019.04 | 91.7 | |
| iSQRT-COV-Net (8K)Backbone=ResNet-50, Feature Dimension=8K2019.04 | 91.7 | |
| BilinearType=Feature Based, Backbone=VGG-162022.12 | 91.3 | |
| FFVT2022.10 | 91.25 | |
| CBPBackbone=VGG-D, Dim.=8K2017.12 | 91.2 | |
| CBPBackbone=VGG-D162019.04 | 91.2 | |
| KPBackbone=ResNet-50, Dim.=14K2017.12 | 91.1 | |
| KPBackbone=ResNet-502019.04 | 91.1 | |
| LRBPBackbone=VGG-D, Dim.=10K2017.12 | 90.9 | |
| LRBPBackbone=VGG-D162019.04 | 90.9 | |
| BaselineBackbone=DeiT-Small2022.05 | 90.7 |