Automated Realtime Mask Availability Detection Using Neural Network

Abstract

The COVID-19 is highly known for its rapid spread. With over 65000 cases recorded on daily. The world started to face huge impacts and challenges. The World Health Organization strongly advice to wear a mask as measurement to fight back the huge spread of the virus. wearing a mask can reduce the infection for more than 80%. However, it’s quite challenging for authorities and governments to ensure the mask availability for people in public places. As there exist people who are careless about wearing the mask. The mask availability system is automated solution to monitor the mask availability. The system contributes with neural networks, image processing and computer vision to take part of mask detecting. The system is trained with over 1000 image of people with mask and people with no mask. The images are used to create trained model using convolutional neural network. The trained model has achieved 97% of accuracy. The system has many strengths and weaknesses. One of the strengths observed is that the system was able to detect all the people faces who are facing the camera and showed the results in the screen. And on the other hand, one of the weaknesses that the system algorithm can be only maintained by certified people who are aware of the technologies used in the system. Future recommendations are to add live monitoring to temperature and increase the data set size to achieve higher accuracy of detection.