Research map / Point clouds / Point cloud classification
Point-Query Quadtree for Crowd Counting, Localization, and More
Chengxin Liu, Hao Lu, Zhiguo Cao, Tongliang Liu
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 point-query quadtree instead of sparse queries only (ablation) | PET · UCF-QNRF | MAE: 90.70 -> 79.53 | Improved |
| use point-query quadtree instead of sparse queries only (ablation) | PET · ShanghaiTech PartA | MAE: 53.59 -> 49.34 | Improved |
| use point-query quadtree instead of dense queries only (ablation) | PET · UCF-QNRF | MAE: 98.14 -> 79.53 | Improved |
| use progressive attention rather than non-progressive rectangle window (ablation) | PET · ShanghaiTech PartA | MAE: 52.00 -> 49.34 | Improved |
| use rectangle windows in both encoder and decoder instead of square (ablation) | PET · ShanghaiTech PartA | MAE: 52.48 -> 49.34 | Improved |
| use progressive rectangle window attention instead of global attention (ablation) | PET · ShanghaiTech PartA | MAE: 52.58 -> 49.34 | Improved |
| use progressive rectangle window attention instead of deformable attention (ablation) | PET · ShanghaiTech PartA | MAE: 51.0 -> 49.34 | Improved |
| retrain PET with refined point annotations | PET · ShanghaiTech PartA | MAE: 49.34 -> 48.52 | Improved |
| apply PET to partial annotation learning with 10% annotated region versus Xu et al. | PET · ShanghaiTech PartA | MAE: 72.79 -> 60.36 | Improved |
| apply PET to partial annotation learning on PartB | PET · ShanghaiTech PartB | MSE: MAE 12.03 -> 10.75; MSE 18.70 -> 19.40 | Mixed |
| PET versus prior methods on UCF-QNRF MSE | PET · UCF-QNRF | MSE: 144.32 vs 141.3 (CLTR) | No clear effect |
Closest papers by meaning
- Rethinking Counting and Localization in Crowds: A Purely Point-Based Framework (ICCV 2021 · similarity 0.89)
- Local Spacing-Aware Hungarian Matching for Stable Point-Supervised Crowd Counting (ECCV 2026 · similarity 0.89)
- Point-to-Region Loss for Semi-Supervised Point-Based Crowd Counting (CVPR 2025 · similarity 0.89)
- Crowd Counting and Density Estimation by Trellis Encoder-Decoder Networks (CVPR 2019 · similarity 0.87)
- Learning to Count via Unbalanced Optimal Transport (AAAI 2021 · similarity 0.87)
- Recurrent Attentive Zooming for Joint Crowd Counting and Precise Localization (CVPR 2019 · 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.
