Research map / Point clouds / Segmentation and detection
A Realism Metric for Generated LiDAR Point Clouds
Larissa T. Triess, Christoph B. Rist, David Peter, J. Marius Zöllner
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 gradient-reversal adversaries (lambda=0.3) to the classifier (ablation) | realism metric · metric calibration | classifier accuracy vs adversary accuracy: 93 classifier vs 50 adversary (chance level) | Improved |
| raise adversarial factor lambda too high (ablation) | realism metric · metric calibration | classifier accuracy | Hurt |
| use one adversary per category instead of a single common adversary (ablation) | realism metric · feature embedding t-SNE | feature clustering | Improved |
| add adversary to classifier (ablation) | realism metric · unseen PandaSet | embedding | Improved |
| per-category adversaries instead of Real-only adversary (ablation) | realism metric · Misc category | feature mixing | Improved |
| up-sample with L1-CNN instead of nearest neighbor | KITTI up-sampling · KITTI val segmentation | mean IoU: 41.5 -> 49.1 | Improved |
| up-sample with GAN instead of L1-CNN | KITTI up-sampling · KITTI val segmentation | mean IoU: 49.1 -> 47.2 | Hurt |
| train on original high-resolution KITTI instead of up-sampled | DarkNet21 · KITTI val segmentation | mean IoU: 49.1 -> 51.4 | Improved |
| use L2-CNN or GAN generated data for dynamic classes | DarkNet21 · KITTI val segmentation | person IoU: 10.9 -> 16.5 | Mixed |
| rank up-sampling by realism score rather than reconstruction error | realism metric · KITTI 4x up-sampling | ranking | Improved |
Closest papers by meaning
- NeRF-LiDAR: Generating Realistic LiDAR Point Clouds with Neural Radiance Fields (AAAI 2024 · similarity 0.88)
- LiDAR-Net: A Real-scanned 3D Point Cloud Dataset for Indoor Scenes (CVPR 2024 · similarity 0.88)
- Towards Realistic Scene Generation with LiDAR Diffusion Models (CVPR 2024 · similarity 0.88)
- GS-LiDAR: Generating Realistic LiDAR Point Clouds with Panoramic Gaussian Splatting (ICLR 2025 · similarity 0.87)
- LiSu: A Dataset and Method for LiDAR Surface Normal Estimation (CVPR 2025 · similarity 0.86)
- TULIP: Transformer for Upsampling of LiDAR Point Clouds (CVPR 2024 · 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.
