Face Mask And Crowd Detection Using Pytorch and Multi-Task Cascade Convolutional Neural Network

Abstract

The whole world is facing the biggest problem of health due to the rapid transmission of coronavirus disease-2019 (COVID-19). According to the World Health Organization (WHO), the best way to prevent the spread of COVID-19 is wearing a mask and keep the distance. But there is huge neglect of the guidelines by people which is resulting in daily increase in an infected patient. It is very difficult to monitor the people manually in these areas. So, in this paper, we propose an idea to monitor people using the automation process to identify the people who are wearing the mask and who are not. Many new trained models are being devised using pre-existing datasets to make the algorithm as accurate as possible. These models have made it possible to extract even the pixel details. We aim to style a binary face classifier which may detect any face present within the frame regardless of its alignment. The proposed idea or module built by pre-trained model and using computer vision libraries in python. The proposed model trained and tested on Real-world Masked Face Recognition Data set (RMFRD). By this project, we can calculate the number of people who are not wearing the mask and don't follow social distancing. By using the pre-trained libraries, this module will be robust and will have a high accuracy rate. PU - SOC SCIENCE & NATURE PI - BHOPAL PA - C-52 HOUSING BOARD COLONY, KOHE FIZA, BHOPAL, MADHYA PRADESH 462 001, INDIA