Performance evaluation of Abrupt motion trackers

Abstract

Abrupt motion tracking has gained special attention in visual tracking community in the past several years. However, the growing interest has not been accompanied by the development of criteria to evaluate the performance of different tracking algorithms. In this paper, we introduce an evaluation criterion for abrupt motion trackers. This criterion contains a set of trials that test the robustness of trackers on a variety of abrupt motions induced by different realworld conditions. Moreover, a new evaluation measure - Abrupt Capture Rate (ACR) is proposed to quantitatively evaluate the accuracy of different trackers. We demonstrate the effectiveness and validation of the proposed evaluation criteria experimentally on several trackers.