Design Of Construction Site Face Recognition And Wear Detection System Based On Deep Learnin

With the continuous development of AI technology,the traditional artificial safety supervision scheme is gradually replaced by intelligent monitoring system,which is also the inevitable trend of industrial intelligence development.Workers must wear special protective equipment when working in the site.Therefore,it is important to identify the identity of staff and monitor the wearing of safety equipment.This paper designs a monitoring system which can be used for face recognition at the entrance and detecting the wearing of safety equipment in the construction site.This paper completes the following research contents.Firstly,the face recognition algorithm at the entrance of the construction site is designed.According to the special environment of the construction site,a light preprocessing method based on HSV color space to adjust the value of V channel is constructed.Then,an improved face recognition model based on Mobile Face Net is constructed.The CBAM module of hybrid domain attention mechanism is embedded in the bottleneck part of the model to improve the detection accuracy of the network.Experiments show that the improved model has high accuracy after adding the attention mechanism module CBAM.Secondly,a monitoring system is designed to detect the wearing of safety equipment of staff in the construction site.The dataset is made according to the actual needs,and the enhancement is realized by using Mosaic-9 algorithm.The regression box filtering algorithm in YOLOv5 algorithm is replaced by WBF from NMS to solve the problems of missing detection caused by high overlap between targets.The Deep SORT tracking algorithm is used to solve the problems of repeated alarm caused by only relying on target detection and judgment.Finally,according to the actual needs of users and the actual application requirements of the monitoring system,the management system and visual interface are designed,the overall debugging of the system is completed,and the feasibility of the system is verified.The monitoring and management system designed in this paper meets the actual needs of site personnel identification and security equipment detection.