Image Classification on APTOS
85.27AccuracySiCoVa
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SiCoVaBackbone=ResNet50 + Jigsaw SSL + CAM refinement2026.05 | 85.27 | 71.5 | 75.29 | 69.12 | 0.91 | |
| TripletBackbone=ResNet502026.05 | 84.17 | 62.53 | 72.26 | 62.94 | 0.87 | |
| TripletBackbone=ResNet50 + Jigsaw SSL + CAM refinement2026.05 | 83.9 | 66.74 | 69.55 | 65.35 | 0.84 | |
| SiCoVaBackbone=ResNet502026.05 | 83.77 | 67.56 | 71.6 | 65.17 | 0.9 | |
| SupervisedBackbone=ViT-base (16x16)2026.05 | 83 | 65 | 72 | 62 | 0.89 | |
| SupervisedBackbone=ResNet502026.05 | 80.27 | 61.77 | 64.89 | 60.07 | 0.85 | |
| SiCoVaBackbone=ViT-base (16x16)2026.05 | 79.95 | 57.79 | 69.02 | 54.92 | 0.83 | |
| SiCoVaBackbone=ViT-base (16x16) + Jigsaw SSL2026.05 | 79.67 | 57.39 | 68.72 | 54.5 | 0.82 | |
| SiCoVaBackbone=ResNet50 + Jigsaw SSL2026.05 | 79 | 57 | 59 | 56 | 0.85 | |
| ClementPBackbone=ViT-base (16x16)2026.05 | 69 | 51 | 54 | 54 | 0.88 | |
| ClementPBackbone=ResNet502026.05 | 51 | 28 | 35 | 38 | 0.66 |