Improving the Detection of Mask-Wearing Mistakes by Deep Learning

Abstract

This study focuses on the detection of wearing mask errors after machine learning by a Multi-Layer Perceptron Mixer (MLP Mixer) applied to protect masks from COVID-19. To combat the spread of the COVID-19 pandemic, facemasks have become an essential accessory, so it's necessary to identify individuals who follow this health protection. In this case, the most successful face-detection method Viola-Jones was used combining different techniques, each in one step. To make decisions, image classification aims to detect the presence of masks in images using mathematical methods. The classy design involves partitioning the parameter space based on representative attributes for each class. For this purpose, we used MLP mixer which is a convolutional neural network, also known as CNNs or ConvNets, they constitute deep learning because it is much better at detecting similarities than by an integrated image-to-image comparison. The classification ratio is satisfactory to achieve maximum accuracy in detecting. However, the learning time for network convergence is prolonged due to changes in parameters.