Covid cases have been shooting up at an alarming pace and coronavirus has become a global threat for the entire human population. The most effective preventive measure according to WHO (World Health Organization) against this virus is masking in public places. A mask can even cut down the infection rate by 40%. For those involved in policy implementation, it is not a trivial task to identify individuals who either are without a mask or wearing it in an improper manner. To assist them in their

more »

... b, we have developed a transfer learning model that can assist in detecting if a person is wearing a mask, in case the mask is there on the face, the model then checks if it is being used in the right way or not. The paper proposes transfer learning model that fine-tunes an already trained state-of-the-art CNN architecture, ResNet50. The proposed model has been trained on a dataset uploaded on Kaggle and gives an accuracy of around 99.7%. The accuracy achieved is a significant improvement when compared with previously used approaches.