Landmark Detection on Shenzhen (test)
2.58MRECDPM-align
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CDPM-alignAnnotation Budget=25-shot, lambda=52026.06 | 2.58 | 2.58 | 56 | 6.51 | |
| CDPM (NIH, lambda=0)Annotation Budget=25-shot2026.06 | 2.65 | 2.5 | 53.53 | 6.77 | |
| CDPM (NIH, lambda=0)Annotation Budget=10-shot2026.06 | 2.89 | 2.49 | 53.4 | 6.41 | |
| CDPM-alignAnnotation Budget=10-shot, lambda=52026.06 | 3.01 | 2.96 | 53.07 | 7.17 | |
| ResNet-101 (ImageNet)Annotation Budget=25-shot, Pre-training=ImageNet2026.06 | 3.09 | 3.16 | 45.67 | 6.91 | |
| Di Via et al.Annotation Budget=25-shot2026.06 | 3.1 | 2.71 | 52.93 | 8.12 | |
| Di Via et al.Annotation Budget=10-shot2026.06 | 3.44 | 3.01 | 52 | 7.92 | |
| ResNet-101 (SimCLR v2)Annotation Budget=25-shot, Pre-training=SimCLR v22026.06 | 3.58 | 3.74 | 44 | 8.86 | |
| ResNet-101 (DINO)Annotation Budget=25-shot, Pre-training=DINO2026.06 | 4.21 | 4.23 | 44.07 | 12.25 | |
| ResNet-101 (MoCo v3)Annotation Budget=25-shot, Pre-training=MoCo v32026.06 | 4.22 | 4.11 | 44 | 9.85 | |
| ResNet-101 (DINO)Annotation Budget=10-shot, Pre-training=DINO2026.06 | 4.92 | 5.13 | 43.87 | 15.15 | |
| CDPM ScratchAnnotation Budget=25-shot2026.06 | 5.02 | 4.75 | 53.2 | 18.06 | |
| ResNet-101 (ImageNet)Annotation Budget=10-shot, Pre-training=ImageNet2026.06 | 5.18 | 4.07 | 43.87 | 10.23 | |
| ResNet-101 (MoCo v3)Annotation Budget=10-shot, Pre-training=MoCo v32026.06 | 5.19 | 5.17 | 42.73 | 20.98 | |
| ResNet-101 (SimCLR v2)Annotation Budget=10-shot, Pre-training=SimCLR v22026.06 | 5.5 | 5.59 | 38.2 | 20.63 | |
| CDPM ScratchAnnotation Budget=10-shot2026.06 | 6.39 | 6.4 | 50.4 | 26.25 | |
| Zhu et al.Annotation Budget=25-shot2026.06 | 37.45 | 39.61 | 19.59 | 266.69 | |
| Zhu et al.Annotation Budget=10-shot2026.06 | 67.82 | 132.46 | 9.43 | 370.73 |