3D Object Detection on ARKitScenes 1.0 (test)
14.6Cab APVGGT-Det
Evaluation results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| VGGT-Detsetting=Sensor-Geometry-Free (SG-Free)2026.03 | 14.6 | 58.8 | 15.2 | 48.2 | 26.3 | 9.6 | 56.4 | 55.1 | 64.9 | 30.9 | 57.7 | 9.1 | 0 | 0 | 10.9 | 12.5 | 5.9 | 28 | |
| NeRF-Detsetting=Sensor-Geometry-Free (SG-Free), training=retrained with multi-view camera poses predicted by VGGT2026.03 | 13.3 | 51.7 | 8.8 | 28.1 | 8.2 | 4.5 | 35.7 | 42.4 | 71.7 | 14.8 | 19.4 | 6.6 | 0.1 | 0 | 1.3 | 0 | 0.2 | 18.1 | |
| MVSDetsetting=Sensor-Geometry-Free (SG-Free), training=retrained with multi-view camera poses predicted by VGGT2026.03 | 10.9 | 53 | 9 | 34.3 | 7.7 | 7.4 | 40.1 | 43.8 | 69.8 | 16.3 | 25.5 | 5.7 | 0.5 | 0 | 1.2 | 0 | 3.7 | 19.4 | |
| ImVoxelNetsetting=Sensor-Geometry-Free (SG-Free), training=retrained with multi-view camera poses predicted by VGGT2026.03 | 6.3 | 42.4 | 4.4 | 26.7 | 3.4 | 5.9 | 23.5 | 20.9 | 47.1 | 9.1 | 15 | 4.4 | 0 | 0 | 0.1 | 0 | 0.8 | 12.4 |