Fine-grained 5-way classification on Stanford Cars
92.621-shot AccuracyARF-SFR-Net-Snapshot
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ARF-SFR-Net-SnapshotBackbone=ResNet-122026.04 | 92.62 | 98.15 | |
| ARF-SFR-NetBackbone=ResNet-122026.04 | 91.24 | 97.62 | |
| BDFRNetBackbone=ResNet-122026.04 | 90.28 | 97.26 | |
| SUITEDBackbone=ResNet-122026.04 | 89.9 | 96.53 | |
| FRN+TDM+BiFI-TDMBackbone=ResNet-122026.04 | 89.15 | 97.31 | |
| BSFANetBackbone=ResNet-122026.04 | 88.93 | 95.2 | |
| SRNetBackbone=ResNet-122026.04 | 88.02 | 96.23 | |
| FRN+TDMBackbone=ResNet-122026.04 | 86.91 | 96.11 | |
| FRNBackbone=ResNet-122026.04 | 86.9 | 95.69 | |
| TDM+CSCAMBackbone=ResNet-122026.04 | 86.86 | 95.63 | |
| C2-NetBackbone=ResNet-122026.04 | 86.48 | 94.07 | |
| FRN+CSCAMBackbone=ResNet-122026.04 | 86.24 | 95.55 | |
| ATR-NetBackbone=ResNet-122026.04 | 85.6 | 95.24 | |
| MFSI-NetBackbone=ResNet-122026.04 | 85.6 | 95.26 | |
| ARF-SFR-NetBackbone=Conv-42026.04 | 85.24 | 95.59 | |
| ARF-SFR-Net-SnapshotBackbone=Conv-42026.04 | 84.9 | 95.51 | |
| TST_MFLBackbone=ResNet-122026.04 | 82.82 | 93.61 | |
| DeepBDCBackbone=ResNet-122026.04 | 82.28 | 93.51 | |
| SUITEDBackbone=Conv-42026.04 | 82.21 | 92.39 | |
| C2-NetBackbone=Conv-42026.04 | 79.52 | 91.15 | |
| HelixFormerBackbone=ResNet-122026.04 | 79.4 | 92.26 | |
| BDFRNetBackbone=Conv-42026.04 | 75.33 | 90.91 | |
| TDM+CSCAMBackbone=Conv-42026.04 | 73.27 | 87.81 | |
| FRN+TDMBackbone=Conv-42026.04 | 72.26 | 89.55 | |
| PaCLBackbone=Conv-42026.04 | 72.21 | 88.02 | |
| FRN+CSCAMBackbone=Conv-42026.04 | 71.44 | 86.44 | |
| DANBackbone=Conv-42026.04 | 70.21 | 85.55 | |
| FRNBackbone=Conv-42026.04 | 67.48 | 87.97 | |
| ATR-NetBackbone=Conv-42026.04 | 66.07 | 82.39 | |
| DeepEMDBackbone=Conv-42026.04 | 61.63 | 72.95 | |
| DN4-DAEmbed.=Conv-64F, k=12019.03 | 61.51 | 89.6 | |
| LRPABNBackbone=Conv-42026.04 | 60.28 | 73.29 | |
| DN4Embed.=Conv-64F, k=12019.03 | 59.84 | 88.65 | |
| GNN+Embed.=Conv-64F2019.03 | 55.85 | 71.25 | |
| DSNBackbone=Conv-42026.04 | 53.45 | 65.19 | |
| BinoHeMBackbone=Conv-42026.04 | 51.53 | 70.62 | |
| Prototypical Nets+Embed.=Conv-64F2019.03 | 40.9 | 52.93 | |
| Matching Nets FCE+Embed.=Conv-64F2019.03 | 34.8 | 44.7 | |
| NBNN (Deep local features)Embed.=Conv-64F2019.03 | 28.18 | 38.27 | |
| k-NN (Deep global features)Embed.=Conv-64F2019.03 | 23.5 | 34.45 |