Efficient and accurate object detection with simultaneous classification and tracking

TL;DR

This paper introduces a detection framework that combines classification and tracking in point cloud streams, improving efficiency and accuracy for moving object detection in robotics.

Contribution

It presents a novel simultaneous classification and tracking framework with a fusion model for key observations, reducing redundant processing and enhancing detection performance.

Findings

Outperforms traditional tracking-by-detection in accuracy.

Reduces computational effort and energy consumption.

Effective for moving object detection in point cloud data.

Abstract

Interacting with the environment, such as object detection and tracking, is a crucial ability of mobile robots. Besides high accuracy, efficiency in terms of processing effort and energy consumption are also desirable. To satisfy both requirements, we propose a detection framework based on simultaneous classification and tracking in the point stream. In this framework, a tracker performs data association in sequences of the point cloud, guiding the detector to avoid redundant processing (i.e. classifying already-known objects). For objects whose classification is not sufficiently certain, a fusion model is designed to fuse selected key observations that provide different perspectives across the tracking span. Therefore, performance (accuracy and efficiency of detection) can be enhanced. This method is particularly suitable for detecting and tracking moving objects, a process that would require expensive computations if solved using conventional procedures. Experiments were conducted on the benchmark dataset, and the results showed that the proposed method outperforms original tracking-by-detection approaches in both efficiency and accuracy.