Performance Evaluation of Random Set Based Pedestrian Tracking Algorithms

TL;DR

This paper compares the error performance of three random set based multi-object trackers in pedestrian video tracking using a standard dataset and the OSPA metric, providing insights into their relative effectiveness.

Contribution

It offers a systematic evaluation of three random finite set based pedestrian tracking algorithms using a rigorous metric and a common dataset, highlighting their strengths and weaknesses.

Findings

Different algorithms show varying error rates under diverse conditions.

OSPA metric effectively quantifies tracking accuracy.

The evaluation provides a benchmark for future pedestrian tracking methods.

Abstract

The paper evaluates the error performance of three random finite set based multi-object trackers in the context of pedestrian video tracking. The evaluation is carried out using a publicly available video dataset of 4500 frames (town centre street) for which the ground truth is available. The input to all pedestrian tracking algorithms is an identical set of head and body detections, obtained using the Histogram of Oriented Gradients (HOG) detector. The tracking error is measured using the recently proposed OSPA metric for tracks, adopted as the only known mathematically rigorous metric for measuring the distance between two sets of tracks. A comparative analysis is presented under various conditions.