Research map / Point clouds / Registration and matching
Differentiable Registration of Images and LiDAR Point Clouds with VoxelPoint-to-Pixel Matching
Junsheng Zhou, Baorui Ma, Wenyuan Zhang, Yi Fang, Yu-Shen Liu, Zhizhong Han
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 the voxel branch to the triplet network (ablation) | VP2P-Match · KITTI | Acc.: 73.53 -> 83.04 | Improved |
| add the point branch to the triplet network (ablation) | VP2P-Match · KITTI | Acc.: 80.80 -> 83.04 | Improved |
| use adaptive-weighted loss instead of contrastive loss (ablation) | VP2P-Match · KITTI | Acc.: 77.01 -> 83.04 | Improved |
| add end-to-end differentiable probabilistic PnP supervision (ablation) | VP2P-Match · KITTI | Acc.: 82.18 -> 83.04 | Improved |
| increase image resolution from 40x128 to 160x512 (ablation) | VP2P-Match · KITTI | Acc.: 82.50 -> 83.04 | Improved |
| increase point count from 20480 to 40960 (ablation) | VP2P-Match · KITTI | Acc.: 82.75 -> 83.04 | Improved |
| increase point count from 40960 to 61440 (ablation) | VP2P-Match · KITTI | Acc.: 83.04 -> 83.70 | Improved |
| VP2P-Match vs CorrI2P (voxel+point vs pixel matching, A-W loss, diff PnP) | VP2P-Match · nuScenes | Acc.: 49.00 -> 88.33 | Improved |
| use EPnP at inference instead of GN-based PnP | VP2P-Match · Pose inference time | seconds per frame: 2.38 -> 0.19 | Improved |
Closest papers by meaning
- FreeReg: Image-to-Point Cloud Registration Leveraging Pretrained Diffusion Models and Monocular Depth Estimators (ICLR 2024 · similarity 0.90)
- DeepI2P: Image-to-Point Cloud Registration via Deep Classification (CVPR 2021 · similarity 0.90)
- Diff2I2P: Differentiable Image-to-Point Cloud Registration with Diffusion Prior (ICCV 2025 · similarity 0.89)
- SPEAL: Skeletal Prior Embedded Attention Learning for Cross-Source Point Cloud Registration (AAAI 2024 · similarity 0.87)
- Cross-PCR: A Robust Cross-Source Point Cloud Registration Framework (AAAI 2025 · similarity 0.87)
- CrossOver: 3D Scene Cross-Modal Alignment (CVPR 2025 · similarity 0.87)
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.
