Research map / Point clouds / Point cloud classification
CMT: Context-Matching-Guided Transformer for 3D Tracking in Point Clouds
Zhiyang Guo, Yunyao Mao, Wengang Zhou, Min Wang, Houqiang Li
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 two-stage context matching (ablation) | BAT · nuScenes Truck | Success: 45.3 -> 46.8 | Improved |
| add transformer fusion in place of BAFF (ablation) | BAT · nuScenes Truck | Success: 45.3 -> 48.7 | Improved |
| add orientation encoder (ablation) | BAT + Transformer · nuScenes Truck | Success: 48.7 -> 49.9 | Improved |
| add context matching (ablation) | BAT + Transformer + OE · nuScenes Truck | Success: 49.9 -> 52.5 | Improved |
| add orientational contextual feature eO (ablation) | CMT · nuScenes Truck | Success: 51.6 -> 52.5 | Improved |
| use P as PE and B as feature instead of both as features without PE (ablation) | CMT · nuScenes Truck | Success: 50.2 -> 51.7 | Improved |
| use concatenated P and B as PE instead of P-only PE (ablation) | CMT · nuScenes Truck | Success: 50.9 -> 52.5 | Improved |
| use concatenated P and B as PE instead of B-only PE (ablation) | CMT · nuScenes Truck | Success: 51.5 -> 52.5 | Improved |
| increase k1 from 4 to 16 (ablation) | CMT · nuScenes Truck | FPS: 41.7 -> 32 | Hurt |
| use CMT instead of BAT | CMT · nuScenes (average) | Success: 41.8 -> 48.5 | Improved |
Closest papers by meaning
- 3D Siamese Transformer Network for Single Object Tracking on Point Clouds (ECCV 2022 · similarity 0.92)
- Revisiting Siamese-Based 3D Single Object Tracking With a Versatile Transformer (TPAMI 2025 · similarity 0.91)
- CXTrack: Improving 3D Point Cloud Tracking With Contextual Information (CVPR 2023 · similarity 0.90)
- Towards Category Unification of 3D Single Object Tracking on Point Clouds (ICLR 2024 · similarity 0.89)
- M3SOT: Multi-Frame, Multi-Field, Multi-Space 3D Single Object Tracking (AAAI 2024 · similarity 0.89)
- Point Transformer V3: Simpler Faster Stronger (CVPR 2024 · 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.
