Assessing Multiple-Target Tracking Performance Of GNN Association Algorithm
Pavel Kulmon, Petra Stukovska · 2018
This paper deals with the performance assessment of multiple-target tracking algorithm. The goal is, provided two different tracking algorithms and data, to evaluate theirperformance and to decide which one of them is more suitable for real-time tracking applications. Since performance evaluation with multiple targets is a complex problem due to ambiguities that create confusion about which track belongs to and follows a target, we will evaluate the performance of tracking algorithms using well-known metrics from different categories such as track cardinality metrics, track accuracy and time metrics. The results of performance evaluation will be used for the decision which of tested tracking algorithms is better to use for multiple-target tracking purposes.