An Ensemble Model for Face Mask Detection Using Faster RCNN with ResNet50

The computer vision which is an important aspect of Artificial Intelligence. The object detection is the most researchable area with deep learning algorithms. Now in the current COVID – 19 pandemics, the social distancing is a mandatory factor to prevent this transmission of this deadly virus. The government is struggling to handle the persons without wearing masks in public places. Our work concentrates on the object detection of face masks using the state-of-the-art methodologies like YOLO, SSD, RCNN, Fast RCNN and Faster RCNN with different backbone architectures like ResNet, MobileNet, etc. This paper brings out various ensemble methods by combining the state of art methodologies and compare those combinations to identify the best performance, in choice of the dataset of the application. We have obtained the highest performance benchmark with the usage of Faster RCNN – ResNet50 among the other ensemble methods. All the performance evaluation metrics are compared with one other with the same face mask detection image dataset. In this paper, we present a balancing collation of the ensemble methods of object detection algorithms.