A Novel Single Shot Multibox Detector based Face Mask Detection

Abstract

Face mask detection has gained significant importance in recent times due to the COVID-19 pandemic. This research work proposes a face mask detection method using the Single Shot MultiBox Detector (SSD) framework. The proposed method aims to accurately identify whether individuals are wearing masks or not in real-time scenarios. To begin with, a dataset comprising images of masked and unmasked faces is collected from kaggle and annotated with corresponding labels. The SSD model is then trained on this dataset, utilizing its ability to simultaneously perform object localization and classification. By leveraging the pre-trained weights and the architecture of SSD, the model is able to detect faces and classify them as either wearing masks or not. During the inference phase, the trained SSD model is applied to input images or video frames. The model analyzes the detected faces and assigns a label based on the presence or absence of masks. The output includes bounding box coordinates around the detected faces, along with the corresponding mask or non-mask label. Experimental evaluations are conducted on various real-world datasets to assess the performance of the proposed method. The results demonstrate the effectiveness and efficiency of the approach, achieving high accuracy in face mask detection.