Research map / Point clouds / Point cloud classification
PICCOLO: Point Cloud-Centric Omnidirectional Localization
Junho Kim, Changwoon Choi, Hojun Jang, Young Min Kim
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 |
|---|---|---|---|
| use sampling loss instead of photometric loss (ablation) | PICCOLO · Stanford2D-3D-S (all areas) | translation error (m): 1.41 -> 0.03 | Improved |
| use sampling loss instead of photometric loss (ablation) | PICCOLO · Stanford2D-3D-S (all areas) | rotation error (deg): 42.29 -> 0.66 | Improved |
| add known gravity direction (ablation) | PICCOLO · Stanford2D-3D-S (all areas) | translation / rotation error: 0.03 -> 0.01 m; 0.66 -> 0.34 deg | Improved |
| use sampling loss with gravity direction instead of photometric (ablation) | PICCOLO · Stanford2D-3D-S (all areas) | translation error (m): 0.93 -> 0.01 | Improved |
| flip query images (ablation) | PICCOLO · Stanford2D-3D-S (all areas) | translation / rotation error: 0.03 -> 0.03 m; 0.66 -> 0.69 deg | No clear effect |
| randomly rotate the point cloud (ablation) | PICCOLO · Stanford2D-3D-S (all areas) | translation / rotation error: 0.03 -> 0.23 m; 0.66 -> 2.21 deg | Hurt |
| add scene changes (handheld) | PICCOLO · OmniScenes | translation error (m): 0.02 -> 0.77 | Hurt |
| add scene changes (handheld) | PICCOLO · OmniScenes | accuracy: 0.71 -> 0.43 | Hurt |
| add scene changes (robot) | PICCOLO · OmniScenes | accuracy: 0.77 -> 0.55 | Hurt |
| add color histogram filtering to loss-only initialization (ablation) | PICCOLO · Stanford2D-3D-S (Area 3 rooms) | recall | Improved |
| increase Nt and Nr initial poses (ablation) | PICCOLO · Stanford2D-3D-S (Area 2 auditoriums) | localization error | Improved |
| use semantic labels instead of RGB input | PICCOLO · Stanford2D-3D-S (all areas) | translation / rotation error: 0.03 -> 0.01 m; 0.66 -> 0.28 deg | Improved |
Later papers that compared with this method
1 later accepted paper reports PICCOLO as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- CPO: Change Robust Panorama to Point Cloud Localization (ECCV 2022 · better on 6 of 7 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- CPO: Change Robust Panorama to Point Cloud Localization (ECCV 2022 · similarity 0.88)
- 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries (CVPR 2024 · similarity 0.85)
- BUFFER-X: Towards Zero-Shot Point Cloud Registration in Diverse Scenes (ICCV 2025 · similarity 0.85)
- ODGS-SLAM: Omnidirectional Gaussian Splatting SLAM (CVPR 2026 · similarity 0.85)
- Combining Absolute and Semi-Generalized Relative Poses for Visual Localization (IJCV 2026 · similarity 0.85)
- Reloc3r: Large-Scale Training of Relative Camera Pose Regression for Generalizable, Fast, and Accurate Visual Localization (CVPR 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.
