| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Relative Pose Estimation | PhotoTourism (test) | mAP@5°67.58 | 30 | |
| Novel View Synthesis | Phototourism (additional scenes) | PSNR22.95 | 15 | |
| Novel View Synthesis | Phototourism Trevi Fountain R/2 downsampled (test) | PSNR24.45 | 11 | |
| Novel View Synthesis | Phototourism Sacre Coeur R/2 downsampled (test) | PSNR24.99 | 11 | |
| Novel View Synthesis | Phototourism Brandenburg Gate R/2 downsampled (test) | PSNR29.65 | 11 | |
| Camera Pose Estimation | Phototourism (additional scenes) | Rotation Error2.644 | 10 | |
| Relative pose estimation | PhotoTourism SPSG 4K minimal samples (test) | mAA (e)0.919 | 9 | |
| Relative Pose Estimation | PhotoTourism RootSIFT (test) | mAA (Error)82.7 | 9 | |
| Camera pose estimation | Phototourism Large noise | ATE0.2664 | 8 | |
| Camera pose estimation | Phototourism Medium noise | ATE0.2645 | 8 | |
| Camera pose estimation | Phototourism Small noise | ATE0.2641 | 8 | |
| Novel View Synthesis | PhotoTourism Taj Mahal NeRF-W | PSNR25.15 | 8 | |
| Video Generation | Phototourism (test) | Consistency1 | 5 | |
| Novel View Synthesis | PhotoTourism NeRF-W (Mean) | PSNR24.06 | 5 | |
| Novel View Synthesis | PhotoTourism NeRF-W (Brandenburg Gate) | PSNR26.5 | 5 | |
| Novel View Synthesis | Phototourism variable appearance (test) | PSNR27 | 5 | |
| Camera Pose Estimation | PhotoTourism (Mean) | Rotation Error (deg)1.053 | 4 | |
| Camera Pose Estimation | PhotoTourism Sacre Coeur | Rotation Error (deg)1.452 | 4 | |
| Camera Pose Estimation | PhotoTourism Taj Mahal | Rotation Error (deg)0.549 | 4 | |
| Camera Pose Estimation | PhotoTourism Trevi Fountain | Rotation Error (deg)1.59 | 4 | |
| Camera Pose Estimation | PhotoTourism Brandenburg Gate | Rotation Error (deg)0.621 | 4 |