Research map / Point clouds / Point cloud classification
CASSPR: Cross Attention Single Scan Place Recognition
Yan Xia, Mariia Gladkova, Rui Wang, Qianyun Li, Uwe Stilla, João F Henriques, Daniel Cremers
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 hierarchical cross-attention transformer (point + voxel branches) to MinkLoc3D-S (ablation) | MinkLoc3D-S · TUM City Campus | AR@1%: 85.7 -> 94.3 | Improved |
| add lightweight self-attention units (ablation) | MinkLoc3D-S · TUM City Campus | AR@1: 69.1 -> 84.9 | Improved |
| combine HCAT and LSA (ablation) | CASSPR · TUM City Campus | AR@1: 77.7 -> 85.6 (vs HCAT only) | Improved |
| use voxel branch as query and point branch as key/value in the second-stage fusion (ablation) | HCAT · TUM City Campus | AR@1%: 89.2 (switched) -> 97.1 | Improved |
| increase number of LSA units from 0 to 6 (ablation) | CASSPR_LSA · Oxford RobotCar | AR@1: 93.0 -> 94.7 | Improved |
| add 1 to 4 LSA units (partial) (ablation) | CASSPR_LSA · Oxford RobotCar | AR@1: 93.0 -> 92.9 (3 and 4 units) | No clear effect |
| increase horizontal quantization angle from 6 to 25 degrees (sparser points) (ablation) | CASSPR · USyd subset | AR@1: decrease of about 3% for CASSPR vs about 6% for MinkLoc3D-S | Hurt |
| reduce maximum LiDAR range to 20m (ablation) | CASSPR · USyd | AR@1: 85.0-87.8 -> 77.7 | Hurt |
| use CASSPR attention instead of PCAN/SOE-Net attention | place recognition attention · TUM City Campus | inference time (ms): 66.8 -> 29.7 | Improved |
| add point branch, HCAT and LSA to a pure voxel network | MinkLoc3D-S · TUM City Campus | inference time (ms): 6.0 -> 29.7 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SVT-Net — SVT-Net: Super Light-Weight Sparse Voxel Transformer for Large Scale Place Recognition (AAAI 2022) · better on 2 of 2 shared results
- PCAN — PCAN: 3D Attention Map Learning Using Contextual Information for Point Cloud Based Retrieval (CVPR 2019) · better on 2 of 2 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
- L2RSI: Cross-view LiDAR-based Place Recognition for Large-scale Urban Scenes via Remote Sensing Imagery (NeurIPS 2025 · similarity 0.88)
- BEVPlace: Learning LiDAR-based Place Recognition using Bird's Eye View Images (ICCV 2023 · similarity 0.88)
- ForestLPR: LiDAR Place Recognition in Forests Attentioning Multiple BEV Density Images (CVPR 2025 · similarity 0.87)
- HOTFormerLoc: Hierarchical Octree Transformer for Versatile Lidar Place Recognition Across Ground and Aerial Views (CVPR 2025 · similarity 0.86)
- DeepPointMap: Advancing LiDAR SLAM with Unified Neural Descriptors (AAAI 2024 · similarity 0.86)
- LiDAR-Net: A Real-scanned 3D Point Cloud Dataset for Indoor Scenes (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.
