CompTrack: Information Bottleneck-Guided Low-Rank Dynamic Token Compression for Point Cloud Tracking

2025-11-19

AI Summary

To address the dual challenges of spatial redundancy (background noise interference) and informational redundancy (inefficient foreground representation) in LiDAR point cloud-based 3D single-object tracking, this paper proposes an end-to-end dynamic token compression framework. Methodologically, grounded in the information bottleneck principle, the framework jointly leverages information-entropy-driven spatial foreground prediction, online singular value decomposition (SVD), and low-rank feature approximation to achieve adaptive background suppression and structured foreground refinement. It further generates lightweight proxy tokens to construct highly compact and discriminative representations. Evaluated on KITTI, nuScenes, and Waymo Open Dataset, the method achieves state-of-the-art (SOTA) performance while maintaining real-time inference speed—90 FPS on a single RTX 3090 GPU—demonstrating a compelling trade-off between accuracy and computational efficiency.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Learning on the Edge & Model Compression
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

3D single object tracking (SOT) in LiDAR point clouds is a critical task in computer vision and autonomous driving. Despite great success having been achieved, the inherent sparsity of point clouds introduces a dual-redundancy challenge that limits existing trackers: (1) vast spatial redundancy from background noise impairs accuracy, and (2) informational redundancy within the foreground hinders efficiency. To tackle these issues, we propose CompTrack, a novel end-to-end framework that systematically eliminates both forms of redundancy in point clouds. First, CompTrack incorporates a Spatial Foreground Predictor (SFP) module to filter out irrelevant background noise based on information entropy, addressing spatial redundancy. Subsequently, its core is an Information Bottleneck-guided Dynamic Token Compression (IB-DTC) module that eliminates the informational redundancy within the foreground. Theoretically grounded in low-rank approximation, this module leverages an online SVD analysis to adaptively compress the redundant foreground into a compact and highly informative set of proxy tokens. Extensive experiments on KITTI, nuScenes and Waymo datasets demonstrate that CompTrack achieves top-performing tracking performance with superior efficiency, running at a real-time 90 FPS on a single RTX 3090 GPU.

Problem

• Eliminates spatial redundancy from background noise in point clouds
• Reduces informational redundancy within foreground objects for efficiency
• Compresses dynamic tokens using low-rank approximation for real-time tracking

Innovation

• Spatial Foreground Predictor filters background noise
• Information Bottleneck-guided Dynamic Token Compression
• Online SVD adaptively compresses redundant foreground tokens