Mask Wearing Specification Detection Based on Cascaded Convolutional Neural Network

Abstract

Standardizing the wearing of masks is an important part of public health management. In order to solve the problem of low detection accuracy of wearing masks in complex light environment in public places, a mask detection method based on cascaded convolution neural network is proposed. Face location is carried out by using MTCNN algorithm, and the obtained face parts are classified at the first level to determine whether to wear a mask or not, and the Fast R-CNN algorithm is used to detect the part below the eyes of the object wearing the mask, and the second-level classification is used to determine whether the mask is standard or not. The experimental results show that in a multifaceted and complex background environment, the correct rates of testing whether to wear a mask and whether the mask is standard are 93.88% and 91.75% respectively, which can effectively detect whether the mask is standard or not.