Research map / Detection and segmentation / Tracking
Delving into Dynamic Scene Cue-Consistency for Robust 3D Multi-Object Tracking
Haonan Zhang, Xinyao Wang, Boxi Wu, Tu Zheng, Wang Yunhua, Zheng Yang
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 temporal encoder (ablation) | DSC-Track · nuScenes val | AMOTA (%): 67.5 -> 73.2 | Improved |
| add temporal encoder (ablation) | DSC-Track · nuScenes val | IDS: 1015 -> 298 | Improved |
| add cue-consistent attention (ablation) | DSC-Track · nuScenes val | AMOTA (%): 70.5 -> 73.2 | Improved |
| add cue-consistent attention (ablation) | DSC-Track · nuScenes val | IDS: 950 -> 298 | Improved |
| add geometric encoder (ablation) | DSC-Track · nuScenes val | AMOTA (%): 71.8 -> 73.2 | Improved |
| add PPF-based relative geometry (ablation) | DSC-Track · nuScenes val | AMOTA (%): 72.1 -> 73.2 | Improved |
| add contextual features (ablation) | DSC-Track · nuScenes val | AMOTA (%): 72.0 -> 73.2 | Improved |
| use cue-consistent attention instead of max pooling (ablation) | DSC-Track · nuScenes val | AMOTA (%): 72.3 -> 73.2 | Improved |
| use cue-consistent attention instead of vanilla cross-attention (ablation) | DSC-Track · nuScenes val | AMOTA (%): 72.8 -> 73.2 | Improved |
| use cue-consistent attention instead of vanilla cross-attention (ablation) | DSC-Track · nuScenes val | IDS: 390 -> 298 | Improved |
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.
- 3DMOTFormer — 3DMOTFormer: Graph Transformer for Online 3D Multi-Object Tracking (ICCV 2023) · better on 5 of 5 shared results
- TrajectoryFormer — TrajectoryFormer: 3D Object Tracking Transformer with Predictive Trajectory Hypotheses (ICCV 2023) · better on 3 of 3 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
- SpOT: Spatiotemporal Modeling for 3D Object Tracking (ECCV 2022 · similarity 0.89)
- Time3D: End-to-End Joint Monocular 3D Object Detection and Tracking for Autonomous Driving (CVPR 2022 · similarity 0.88)
- RegTrack: Simplicity Beneath Complexity in Robust Multi-Modal 3D Multi-Object Tracking (TPAMI 2026 · similarity 0.88)
- TrajectoryFormer: 3D Object Tracking Transformer with Predictive Trajectory Hypotheses (ICCV 2023 · similarity 0.88)
- TAPIP3D: Tracking Any Point in Persistent 3D Geometry (NeurIPS 2025 · similarity 0.87)
- Tracktention: Leveraging Point Tracking to Attend Videos Faster and Better (CVPR 2025 · 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.
