A variable patch approach with polling mechanism for intelligent human tracking

Abstract

Human tracking is one of the challenging and significant components of intelligent tracking systems. Even though the correlation filter-based trackers accomplished the competitive outcomes in terms of accuracy and robustness, there is a need to improve tracking performance. In this paper, the vote mapping of patched confidence methodology is used to enhance a tracking system's performance by continuously increasing the number of patches. The system aims to provide robustness to occlusion and global scene changes by utilizing numerous patches from an image's bounding box. The proposed variable patch approach with a polling mechanism provides robustness to the occlusion in real-life tracking scenarios. The extensive quantitative and qualitative evaluation of a few challenging sequences is carried out using various patches. The tracking performance of proposed technique is better than existing techniques in terms of precision and success rate.