Face mask detection using deep learning: An approach to reduce risk of Coronavirus spread

[Display omitted] •In this work, a deep learning based model for detecting masks over faces in public place to curtail community spread of Coronavirus is presented. The proposed model efficiently handles varying kinds of occlusions in dense situation by making use of ensemble of single and two stage detectors. The ensemble approach not only helps in achieving high accuracy but also improves detection speed considerably. The model is 98.2% accurate at mask detection with average inference times of 0.05 seconds per image.•The high accuracy of model is also due to highly balanced face mask centric dataset achieved through Random over-sampling with data augmentation over original MAFA dataset. Our technique reduces the imbalance ratio ρ = 11.82 (original) to ρ = 1.07.•The other factors that contributed towards achievement of highly efficient model include application of bounding box affine transformation and transfer learning. The bounding box transformation improves localization performance during mask detection. Transfer learning leads to good results by enabling use of powerful pre-trained model such as ResNet 50 being trained on large dataset like ImageNet.•The experiment is conducted with three most popular baseline models viz. ResNet50, AlexNet and MobileNet and explored the possibility of plug-in with proposed model to achieve highly accurate results in less inference time. It is observed that proposed technique achieves high accuracy (98.2%) when implemented with ResNet 50.•Besides, the results are also compared with recent public baseline model published as RetinaFaceMask [14] and improvements of 11.07% and 6.44% in Precision and Recall for mask detection are recorded. Effective strategies to restrain COVID-19 pandemic need high attention to mitigate negatively impacted communal health and global economy, with the brim-full horizon yet to unfold. In the absence of effective antiviral and limited medical resources, many measures are recommended by WHO to control the infection rate and avoid exhausting the limited medical resources. Wearing a mask is among the non-pharmaceutical intervention measures that can be used to cut the primary source of SARS-CoV2 droplets expelled by an infected individual. Regardless of discourse on medical resources and diversities in masks, all countries are mandating coverings over the nose and mouth in public. To contribute towards communal health, this paper aims to devise a highly accurate and real-time technique that ca