Safety helmet positioning and color recognition method and system based on deep learning

Abstract

The invention provides a safety helmet positioning and color recognition method and system based on deep learning, and the method comprises the steps: collecting image information, obtaining a movingobject image in the image information, and enabling the moving object image to serve as a region of interest; inputting the region of interest into a pre-established convolutional neural network model, and performing multi-target parallel detection on the image information according to a preset target classification; obtaining a parallel detection result, wherein the detection result at least comprises human head position information and safety helmet position information; judging whether a person wearing the safety helmet in the collected image information is judged according to the paralleldetection result, and when it is judged that the person does not wear the safety helmet, giving out alarm information; according to the invention, a designated area can be continuously and effectivelymonitored, safety accidents caused by the fact that personnel do not wear the safety helmet are prevented, manual intelligent analysis and processing are replaced, alarm signals are sent out in realtime, and the problems of high universality, low stability and low accuracy of security video monitoring are solved.

Description

Technical field [0001] The invention relates to the field of computer applications, in particular to a method and system for safety helmet positioning and color recognition based on deep learning. Background technique [0002] With the popularity of security surveillance cameras, identification requirements in various scenarios have also emerged, especially for the identification requirements of personnel safety. For construction sites, factories and other specific occasions, relevant staff and foreign visitors must wear safety helmets, and this type of area needs to be monitored by security surveillance cameras for 24 hours in real time. If a human body enters the area and does not wear the safety helmet correctly Then promptly remind or issue an alarm signal. [0003] The traditional security video surveillance method uses manual methods to distinguish large quantities of surveillance videos, but this method is difficult to work continuously for 24 hours, and it is easy to miss ...

AI Technical Summary

Problems solved by technology

The biggest limitation of this type of serial identification method is that if there is a problem with the first step of human identification, the subsequent helmet identification operation cannot be performed

In real scenes, the human body is easily occluded in the video screen, and the angle of the image acquisition device also has a great impact on the accuracy of human body recognition, which is likely to cause misidentification and missed detection

Therefore, the versatility, stability and accuracy of this serial method are not high

Therefore, a new monitoring method is needed to improve the generality, stability and accuracy of security video monitoring.

Images