Portal de Periódicos da CAPES

Real-Time Face Mask Detection Using Deep Learning

2023; Volume: 12; Issue: 40 Linguagem: Inglês

10.5120/ijais2023451944

2249-0868

Md. Towhidul Islam Robin, Md. Samiul Islam, Ahmed Abdal Shafi Rasel, Mehedi Hasan,

Face recognition and analysis

COVID-19 has made a severe impact on the world's economy, health, education sector, and so on.Although the infection rate of the COVID-19 virus has dropped significantly in recent days, there is no scope to ignore precautionary measures like wearing masks in public.In this paper we proposed, a real-time Deep Learning based face detection system using a balanced dataset of over 6000 images divided into two classes named 'With Mask' and 'Without Mask.The existing studies do not consider the data where people wear masks but not correctly, hence violating the safety measures.Our dataset is well-curated to detect people who do not wear masks properly.Our proposed CNN model not only overperformed other existing studies in terms of accuracy but also requires significantly less memory and time compared to other existing models.Our model achieved an accuracy rate of about 98.5% and it requires only 20 Mega Bytes of memory to deploy the model.