Performance Evaluation of Random Set Based Pedestrian Tracking Algorithms

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. Head and body detections are unreliable in the sense that the probability of detection is low and false detections are non-uniformly distributed. The tracking error is measured using the recently proposed OSPA metric for tracks (OSPA-T), adopted as the only known mathematically rigorous metric for measuring the distance between two sets of tracks. The paper presents the correct proof of the triangle inequality for the OSPA-T. A comparative analysis is presented under various conditions.