Pedestrian Detection in Crowded Crowd Scene Based on Deep Learning

Abstract

Detection-based methods typically detect and locate each person on a crowd image by using a designed pedestrian target detector and obtain counting results by accumulating each detected person. However, these methods require a large amount of computational resources and are often limited by human occlusion and complex background in real scenes, resulting in inaccurate detection. Based on the characteristics of computer depth learning and population density distribution map, a more optimized pedestrian detection approach is proposed.