Design And Implementation Of Monitoring System Of Protective Goggles Wearing Based On Deep Learning

The outbreak of COVID-19 in 2020 has brought huge threat to social development and people's lives and health.Wearing personal protective eqiupment such as masks and protective goggles is the key method to provent people especially medical staff from infection.Moreover,it is clearly requird to wear goggles to prevent eye damage in chemical,metallurgy related factoris and laboratories.However,the currnet monitoring method of wearing goggles is mainly based on manual inspection,which is inefficient and manpower-consuming.Therefore,the design and implementation of an automated monitoring system of protective goggles wearing is quite necessary when people entering or staying in specific medical or laboratory places.For the practical application of the system,this thesis proposes a protective goggles wearing detection algorithm based on deep learning.By optimizing the deep residual network structure and introducing the Dropout mechanism,the over-fitting phenomenon is reduced,and the classification accuracy of goggles wearing is improved.At the same time,we build a protective goggles wearing data set and propose a goggles wearing image synthesis algorithm based on facial feature points which improves the generalization ability of the protective goggles wearing detection algorithm.Combining with the classification results of goggles wearing in face and eye area,we design and implement a protective goggles wearing detection system.After experimental verification,the system meets the requirement of actual scenarios in terms of real-time and accuracy.