Selective vote filtering improves 3D LiDAR object tracking speed and accuracy
SAVTrack: Selective Vote Aggregation for Reliability-Aware Point Cloud Tracking
Computer Vision and Pattern Recognition
Summary
Tracking objects in 3D using LiDAR data is hard when the data is sparse or incomplete, because some points on the object give bad guesses about its location. The authors propose a way to estimate how reliable each guessed position is, then ignore the less trustworthy ones before combining the guesses. This method, called SAVTrack, helps find the object more accurately and faster by focusing only on good information. It works well on popular driving datasets and is especially good when the object is hard to see clearly.
What this means in practice
- •For autonomous vehicle engineers: Improve 3D object tracking modules to be faster and more accurate when sensor data is sparse or noisy.
- •For robotics software developers: Enhance real-time tracking performance in robots operating in cluttered or partially observed environments through selective data aggregation.
Authors
Sifan Zhou, Linyue Tan, Qiwei Wang, Ziyu Zhao, Xiaobo Lu
Abstract
3D single object tracking (SOT) in LiDAR point clouds is essential for autonomous systems, but remains challenging under sparse and incomplete observations. In such cases, different target points provide highly uneven constraints on the object center, causing some point-to-center votes to be substantially less reliable than others. Existing point-based trackers typically aggregate these hypotheses without explicitly modeling their reliability, allowing inaccurate votes to contaminate proposal clustering and degrade localization accuracy. To address this issue, we propose \textbf{SAVTrack}, a motion-aware tracking framework with \textbf{Selective Vote Aggregation (SAV)}. SAVTrack estimates the reliability of each candidate vote from both local seed features and inter-frame motion context, and removes low-confidence hypotheses before proposal clustering. This pre-aggregation gating prevents unreliable hypotheses from affecting cluster formation while introducing only modest computational overhead. SAVTrack achieves competitive performance on KITTI and nuScenes, reaching 68.4/87.4 and 58.44/69.82 Success/Precision, respectively, while running at 82 FPS. It retains fewer than one-sixth of the candidate votes used by dense aggregation and remains particularly effective under sparse target observations.