Pedestrian Detection Fusion Method Based on Mean Shift

Abstract

Detecting pedestrians is a challenging task, which requires precise localization of pedestrians that appear in images and videos. Window-scanning based detection methods have demonstrated their promise by scanning the image densely with multi-scale detection window. However, an essential and critical issue, i.e., how to fuse these dense detections obtained through pedestrian detector and yield the final target detection, is not well addressed in the literature. This paper proposes and implements a general method for fusing pedestrian detections. In this method, detection fusion is regarded as a kernel density estimate and implemented through mean shift iterative procedure. Moreover, the notion of nearest neighbor consistency is adopted, which significantly accelerates the fusion procedure. Experimental results demonstrate the efficiency of the mean shift-based fusion method.