Research on Pedestrian Detection Technology Based on Deep Learning Network Complexity

Pedestrian detection (P-D) is an significant part in the field of machine vision. The development of P-D is allong with the whole process of target detection. The P-D's objective is to accurately find the moving suspected people in the video, return its classification results. Due to the particularity of its detection targets, for the past few years, P-D technology has been broadly applied in monitoring equipment, unmanned driving, smart home systems and other fields, which has attracted widespread attention from the academic community and society. However, the human body itself is a non-rigid target, and when a limb movement occurs, the extracted target features are time-varying. This leads to the machine learning method based on hand-designed features, which has the disadvantages of robustness and low noise immunity.