Visual Servoing on Hollywood poster Perturbed
95.4Convergence RateART-VS: DINOv3-6x6
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ART-VS: DINOv3-6x6Method Type=Training-free: patch-level ViT, Backbone Architecture=DINOv3, Resolution / Tiling Configuration=6x62026.06 | 95.4 | 1.01 | 0.88 | 1,767 | 11.9 | |
| ART-VS: AM-Radio-6x6Method Type=Training-free: patch-level ViT, Backbone Architecture=AM-Radio, Resolution / Tiling Configuration=6x62026.06 | 91 | 0.81 | 0.69 | 2,779 | 6.8 | |
| ART-VS: DINOv2-7x7Method Type=Training-free: patch-level ViT, Backbone Architecture=DINOv2, Resolution / Tiling Configuration=7x72026.06 | 87.2 | 0.88 | 0.78 | 2,174 | 10.5 | |
| ViT-VS: DINOv3-1440Method Type=Training-free: patch-level ViT, Backbone Architecture=DINOv3, Resolution / Tiling Configuration=14402026.06 | 81 | 1.65 | 1.43 | 2,431 | 0.9 | |
| ViT-VS: DINOv2-518Method Type=Training-free: patch-level ViT, Backbone Architecture=DINOv2, Resolution / Tiling Configuration=5182026.06 | 77.8 | 2.6 | 2.22 | 1,270 | 6.1 | |
| ViT-VS: DINOv2-308Method Type=Training-free: patch-level ViT, Backbone Architecture=DINOv2, Resolution / Tiling Configuration=3082026.06 | 76.6 | 2.15 | 1.83 | 1,230 | 14.4 | |
| DMLVSMethod Type=Scene-specific finetuning2026.06 | 76 | 1.93 | 1.92 | — | — | |
| AKAZE-IBVSMethod Type=Training-free: pixel-level descriptors2026.06 | 64.8 | 0.09 | 0.08 | — | 1.7 | |
| ORB-IBVSMethod Type=Training-free: pixel-level descriptors2026.06 | 48 | 0.46 | 0.39 | — | 20.8 | |
| SIFT-IBVSMethod Type=Training-free: pixel-level descriptors2026.06 | 38.8 | 0.29 | 0.25 | — | 2.8 | |
| PBVS-CNNMethod Type=Scene-specific finetuning, Backbone Architecture=CNN2026.06 | 36.8 | 3.22 | 2.37 | — | — |