Research map / 3D vision / Face and shape
"Segment, Lift and Fit: Automatic 3D Shape Labeling from 2D Prompts"
Jianhao Li, Tianyu Sun, Zhongdao Wang, Enze Xie, Bailan Feng, Hongbo Zhang, Ze Yuan, Ke Xu, Jiaheng Liu, Ping Luo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add point cloud alignment term to mask alignment term (ablation) | SLF · KITTI val Car | AP_3D Easy: 6.9 -> (valid results) | Improved |
| add mask alignment term on top of point cloud alignment (ablation) | SLF · KITTI val Car | AP_3D | Improved |
| add ground alignment term (ablation) | SLF · KITTI val Car | AP_3D | Improved |
| increase PCA shape dimension from 3 to 5 (ablation) | SLF · KITTI val Car | AP_3D | Improved |
| increase PCA dimension from 5 to 10 (ablation) | SLF · KITTI val Car | AP_3D | No clear effect |
| use more CAD models for PCA shape prior (5, 30, to 79) (ablation) | SLF · KITTI val Car | AP_3D | Improved |
| increase SAM prompt points from 3 to 8 (ablation) | SLF_point · KITTI val Car | AP_3D | No clear effect |
| use 1 point prompt instead of 3 (ablation) | SLF_point · KITTI val Car Easy | AP_3D | Hurt |
| replace human ground-truth masks with SAM masks from box prompts (ablation) | SLF · KITTI val Car | AP_3D / AP_BEV: Mod. AP_3D 87.8 -> 88.2; Hard 83.8 -> 83.9; Easy 87.6 -> 87.9 | Mixed |
| train detectors on SLF pseudo labels instead of FGR pseudo labels | OpenPCDet detectors · KITTI val Car | AP_3D Moderate/Hard: PointPillar Mod. 89.4 -> 92.1 | Improved |
| use SLF instead of supervised MTrans for cross-dataset labeling | auto-labeler · nuScenes val | mAP: 45.9 -> 57.2 | Improved |
| downsample LiDAR from 64 to fewer beams (ablation) | SLF · KITTI val Car | AP_3D | Hurt |
Closest papers by meaning
- OpenBox: Annotate Any Bounding Boxes in 3D (NeurIPS 2025 · similarity 0.86)
- UniPAD: A Universal Pre-training Paradigm for Autonomous Driving (CVPR 2024 · similarity 0.86)
- Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance (AAAI 2025 · similarity 0.86)
- MWSIS: Multimodal Weakly Supervised Instance Segmentation with 2D Box Annotations for Autonomous Driving (AAAI 2024 · similarity 0.86)
- LabelAny3D: Label Any Object 3D in the Wild (NeurIPS 2025 · similarity 0.86)
- Any3DIS: Class-Agnostic 3D Instance Segmentation by 2D Mask Tracking (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
