A Pedestrian Recognition Method and System Based on Adaptive Depth Spatial Features

Abstract

The present invention provides a pedestrian recognition method and system for adaptive depth spatial features; the method includes the steps of: segmenting the pedestrian image according to the preset segmentation number n; establishing n groups of feature extraction models with the same number of blocks as the pedestrian image, And correspondingly extract feature information from n blocks of pedestrian image blocks; establish n groups of feature classifiers with the same number of feature information groups, and correspondingly classify n groups of feature information; calculate each The loss value generated in the group feature classification process; return n groups of loss values ​​to the corresponding feature extraction model and feature classifier to optimize the feature extraction model and feature classifier. The present invention can effectively extract more information of pedestrians, especially local information, and adopts the adaptive pedestrian feature classification in the form of loss value feedback, and finally tends to the optimal solution, which can guide the pedestrian feature extraction model to extract more depth spaces pedestrian feature information.

Description

technical field [0001] The present invention mainly relates to the technical field of image processing, in particular to a pedestrian recognition method and system for adaptive depth space features. Background technique [0002] With the popularization of computers and the emergence of smart devices, technology has an increasing influence on people. With the intensification of globalization, people's demand for identity verification is becoming stronger and stronger. There is a big gap between traditional personal identification codes, passwords, IC cards, security cards and other traditional identification technologies due to their unchanging operations and difficulty in remembering and the needs of the information age. Face recognition is a very popular topic in the field of biometric identification. It is a biometric identification technology that is expected to be widely used in social and economic activities and people's daily life after fingerprint identification. Ped...

AI Technical Summary

Problems solved by technology

[0003] With the development of society and the advancement of technology, technology has more and more influence on people, and people's demand for identification is getting stronger and stronger. Inconsistencies and other issues make it easy for the same pedestrian to be misjudged as the same pedestrian

In the past, the method of directly extracting features from the entire image was often used, which often did not extract enough more detailed feature information.

Images