Research map / Point clouds / Point cloud classification
Going Further with Point Pair Features
Stefan Hinterstoisser, Vincent Lepetit, Naresh Rajkumar, Kurt Konolige
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 accounting for sensor noise (feature spreading and vote flagging) (ablation) | Drost-PPF · Occlusion dataset (8 objects) | share of matching-score gain: 43.1% | Improved |
| add smart sampling of point pairs (two voting balls, voxel grid) (ablation) | Drost-PPF · Occlusion dataset (8 objects) | share of matching-score gain: 41.3% | Improved |
| add pre-processing of 3D model and scene (keep pairs with large normal angle) (ablation) | Drost-PPF · Occlusion dataset (8 objects) | share of matching-score gain: 15.6% | Improved |
| use full method versus Drost-PPF runtime | Drost-PPF · 640x480 depth maps | processing time: about 6 times faster | Improved |
| use full method versus Krull et al. despite depth only | Krull et al. · Occlusion dataset (8 objects), average | Matching score (%): 73.6% -> 76.3% | Improved |
| use full method versus Drost-PPF on Linemod objects, for Duck | Drost-PPF · Linemod dataset (13 objects), Duck | Matching score (%): 46.0% -> 98.2% | Improved |
| replace the flag array over model point pairs with array indexed by quantized PPFs (ablation) | direct flagging method · Linemod dataset (13 objects) | runtime: slow-down of factor 3 for 1000 model points | Improved |
| use full method versus Linemod template method on Glue object | Linemod · Linemod dataset (13 objects), Glue | Matching score (%): 91.8% -> 75.4% | Hurt |
Closest papers by meaning
- SPOT: Selective Point Cloud Voting for Better Proposal in Point Cloud Object Detection (ECCV 2020 · similarity 0.86)
- Not All Points Are Equal: Learning Highly Efficient Point-Based Detectors for 3D LiDAR Point Clouds (CVPR 2022 · similarity 0.86)
- Point2Seq: Detecting 3D Objects As Sequences (CVPR 2022 · similarity 0.86)
- Back-Tracing Representative Points for Voting-Based 3D Object Detection in Point Clouds (CVPR 2021 · similarity 0.85)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · similarity 0.84)
- AS-Det: Active Sampling for Adaptive 3D Object Detection in Point Clouds (AAAI 2025 · similarity 0.84)
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.
