LERF
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| 3D Object Extraction | LeRF | Accuracy99.8 | 26 | |
| Open-Vocabulary 3D Scene Segmentation | LeRF-mask | Figurines mIoU90.8 | 17 | |
| Open-vocabulary 3D object selection | LERF | Ramen Score61.4 | 16 | |
| 3D Object Localization | LERF | Ramen Success Rate75.6 | 14 | |
| 3D Object Selection | LERF figurines scene | Peak VRAM8 | 14 | |
| 3D Semantic Segmentation | LERF (test) | mIoU62.1 | 13 | |
| 3D Scene Reconstruction | LERF average across four scenes | PSNR24.02 | 12 | |
| Open-vocabulary semantic segmentation | LeRF-OVS | mIoU64.4 | 12 | |
| 3D Semantic Segmentation | LERF | mIoU (Ramen)57.6 | 9 | |
| 3D Open-vocabulary Segmentation | LERF-style Dataset bed scene (test) | mIoU89.5 | 8 | |
| Open-vocabulary 3D Scene Understanding | LERF | Feature Distillation Time (h)1 | 7 | |
| 2D Semantic Segmentation | LERF Overall | mIoU60.02 | 6 | |
| 2D Localization | LERF (Overall) | mAcc84.57 | 6 | |
| Open Vocabulary 3D Semantic Segmentation | LERF (held-out test views) | Ramen mIoU52 | 5 | |
| Open-vocabulary 3D object selection | Lerf ovs (part) | mIoU44.1 | 4 | |
| Open-vocabulary 3D object retrieval | LERF | Ramen mIoU53.34 | 4 | |
| 3D Object Retrieval | LERF standard scene (~300 images) | Geometry Computation Time (mins)15 | 4 | |
| Open-vocabulary 2D object retrieval and localization | LERF | mIoU (Ramen Scene)63.4 | 4 | |
| Novel View Rendering | LERF Figurines | Speed (FPS)354.72 | 4 | |
| 3D reasoning segmentation | LERF | mIoU92.88 | 4 | |
| Semantic Segmentation | LERF-Masked | Figurines mIoU89 | 3 | |
| 2D Object Retrieval | LERF | mIoU56.84 | 3 | |
| 3D Object Retrieval | LERF | mIoU56.11 | 3 | |
| 3D object localization | LERF (test) | RAMEN Score0.732 | 3 | |
| Open-Vocabulary 3D Scene Segmentation | Lerf_ovs (whole scale) | mIoU51.78 | 2 |
Showing 25 of 26 rows