Face Mask Detection Using CNN via Active Learning

The COVID 19 not only has changed the way we live but it also changed how hospitals operate. Security in hospitals now-a-days is given utmost priority. As one small mistake can easily change the life-saving hospital into a Coronavirus Hot spot. Face masks are by far the most important protection measure during the global pandemic COVID19 began to spread. The most effective preventative precaution against the corona virus COVID19 pandemic, according to the World Health Organization, is using a face mask in public settings. Face mask detection is performed by utilizing a deep learning model. In terms of image recognition models, Convolutional Neural Networks (CNNs) have established themselves as the dominating class. In this work we apply Active Learning with CNN using the five query strategies i.e. Largest Margin, Smallest Margin, Entropy Reduction, Least Confident and Random Sampling. Our experiments achieved an accuracy in the range of 62% to 72% and the gain achieved via active learning is in the range of 73% to 79%.