Research on Multi-feature and Machine Learning Hierarchical Pedestrian Detection Method Based on Deep Learning

Abstract

Abstract With the rapid development of science and technology, computers have become an important way for people to obtain information. As an important means of transmitting visual information, image plays a very important role. Pedestrian detection is a key research field of machine vision and a key technology in intelligent video surveillance and intelligent transportation system. Pedestrian detection is the basis and premise of pedestrian tracking, behavior analysis, gait analysis and pedestrian identification, and a good pedestrian detection algorithm can provide strong support and guarantee for the latter. Pedestrian detection is not an independent problem, it is closely related to video pedestrian tracking, human behavior recognition and pedestrian identification. The method of machine learning is to let the program have the ability of learning. Through continuous training and iteration, the parameters of the model are optimized. Deep learning can give full play to the speed and rationality of the machine and improve the accuracy of pedestrian detection.