Research map / Point clouds / Point cloud classification
GLT-T: Global-Local Transformer Voting for 3D Single Object Tracking in Point Clouds
Jiahao Nie, Zhiwei He, Yuxiang Yang, Mingyu Gao, Jing Zhang
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 |
|---|---|---|---|
| Replace VoteNet voting with the GLT module (ablation) | GLT-T voting scheme on KITTI Car · KITTI Car | Success: 61.7 -> 66.4 | Improved |
| Replace VoteNet voting with the GLT module (ablation) | GLT-T voting scheme on KITTI Car · KITTI Car | Precision: 76.1 -> 80.7 | Improved |
| Add the importance-weighted training strategy to GLT (ablation) | GLT-T on KITTI Car · KITTI Car | Success: 66.4 -> 68.2 | Improved |
| Add the importance-weighted training strategy to GLT (ablation) | GLT-T on KITTI Car · KITTI Car | Precision: 80.7 -> 82.1 | Improved |
| Add the global transformer block (ablation) | GLT-T voting module on KITTI Car · KITTI Car | Success: 61.7 -> 64.8 | Improved |
| Add the global transformer block (ablation) | GLT-T voting module on KITTI Car · KITTI Car | Precision: 76.1 -> 79.0 | Improved |
| Add the local transformer block after the global transformer (ablation) | GLT-T voting module on KITTI Car · KITTI Car | Success: 64.8 -> 66.4 | Improved |
| Add the local transformer block after the global transformer (ablation) | GLT-T voting module on KITTI Car · KITTI Car | Precision: 79.0 -> 80.7 | Improved |
| Use sparse sampling size m=16 instead of m=4 (ablation) | GLT-T global transformer · KITTI Car | Success: 61.3 -> 68.2 | Improved |
| Increase sparse sampling size from m=16 to m=32 (ablation) | GLT-T global transformer · KITTI Car | Success: 68.2 -> 67.6 | Hurt |
| Increase sparse sampling size from m=16 to m=32 (ablation) | GLT-T global transformer · KITTI Car | FLOPs / G: 3.56 -> 6.79 | Hurt |
| Set local KNN sampling size to n=8 rather than n=16 (ablation) | GLT-T local transformer · KITTI Car | Success: 68.2 -> 66.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PTTR — PTTR: Relational 3D Point Cloud Object Tracking With Transformer (CVPR 2022) · better on 1 of 1 shared result
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
- Exploring Point-BEV Fusion for 3D Point Cloud Object Tracking With Transformer (TPAMI 2024 · similarity 0.89)
- Revisiting Siamese-Based 3D Single Object Tracking With a Versatile Transformer (TPAMI 2025 · similarity 0.88)
- PTTR: Relational 3D Point Cloud Object Tracking With Transformer (CVPR 2022 · similarity 0.88)
- Point Transformer V3: Simpler Faster Stronger (CVPR 2024 · similarity 0.87)
- M3SOT: Multi-Frame, Multi-Field, Multi-Space 3D Single Object Tracking (AAAI 2024 · similarity 0.87)
- Towards Category Unification of 3D Single Object Tracking on Point Clouds (ICLR 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.
