Image Classification on Stanford Cars (Accuracy, SID@5, Class-Independence, Contrastiveness)
94AccuracyDINO-QPM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DINO-QPMLocal. Features=✓2026.04 | 94 | 91.7 | 93.7 | 100 | |
| DINO-QPM CompactLocal. Features=✓, Model Variant=Compact2026.04 | 94 | — | 93.6 | 100 | |
| DINO-QSENNLocal. Features=✓2026.04 | 93.3 | 92.6 | 94 | 94.9 | |
| Dense F_frozLocal. Features=✓2026.04 | 92.9 | 93.1 | 98.7 | 82.8 | |
| DINO-SLDDLocal. Features=✓2026.04 | 92.9 | 90.9 | 93.9 | 94.9 | |
| Resnet50 BaselineLocal. Features=✓2026.04 | 92.5 | 51.5 | 97.9 | 75.1 | |
| Resnet50 QPMLocal. Features=✓2026.04 | 92.1 | 88.2 | 97.8 | 97.1 | |
| DINOv2 f_froz_CLS Linear ProbeLocal. Features=✗, Evaluation Protocol=Linear Probe2026.04 | 91.7 | 51.5 | 99.1 | 60.9 |