A Pedestrian Re-identification Method Based on Deep Learning

Abstract

The invention discloses a pedestrian re-identification method based on deep learning, comprising: step 1: preprocessing a pedestrian picture data set, classifying the pictures according to pedestrian IDs, and enhancing the pictures; step 2: selecting a residual network as the basic Step 3: Use the BatchHard algorithm to build a triple loss function; Step 4: Input the enhanced pedestrian image dataset into the adjusted network for training, according to the change of the loss function, get The trained network model; Step 5: Input the pedestrian pictures and videos to be identified into the trained network model, and output the pedestrian re-identification information. The invention realizes pedestrian re-identification through the deep learning method, and can be used in the fields of pedestrian search, target tracking and monitoring.

Description

technical field [0001] The invention relates to computer vision processing, in particular to a pedestrian re-identification method based on deep learning, which can be applied to the fields of video surveillance, suspect tracking, personnel search and the like. Background technique [0002] With the rapid development of society, social and public safety issues have attracted more and more attention. Surveillance cameras have been installed in large public places, followed by massive video surveillance data. How to make good use of these massive data to make surveillance More efficient and intelligent technology has become a problem to be solved. Pedestrian re-identification is the core link in video surveillance. Pedestrian re-identification is a technology to determine whether pedestrians appearing under different surveillance videos belong to the same pedestrian. Traditional pedestrian re-identification The technology mainly adopts the method of manually extracting feature...

AI Technical Summary

Problems solved by technology

[0002] With the rapid development of society, people pay more and more attention to social public security issues. Surveillance cameras have been installed in large public places, followed by massive video surveillance data. How to make good use of these massive data to make surveillance More efficient and intelligent technology has become a problem to be solved. Pedestrian re-identification is the core link in video surveillance. Pedestrian re-identification is a technology to judge whether pedestrians appearing in different surveillance videos belong to the same pedestrian. Traditional pedestrian re-identification The technology mainly adopts the method of manual feature extraction, which has low efficiency and low accuracy.

In the prior art, there are pedestrian re-identification methods based on machine learning, but there is a problem of low re-identification accuracy caused by poor extraction of pedestrian visual features

Images