VISION-BASED FACE MASK DETECTION IN REAL TIME VIDEOS COMPUTER

Abstract: The recent COVID-19 pandemic has highlighted the necessity of using facemasks as a preventive tool against infectious illnesses. There has been a lot of interest in developing automated systems for real-time facemask detection to monitor compliance with facemask use. This work presents a visual method for identifying masks worn by individuals in live-action footage. The proposed system utilizes computer vision methods to analyze live footage captured by a camera or online cam. Firstly, face identification algorithms are used to locate human faces in the video footage. Next, a deep learning-based classifier is used to determine whether the recognized face is covered by a mask.To train the facemask detection algorithm, a large dataset of annotated photos of people wearing and not wearing facemasks is used. Transfer learning strategies are employed to perform accurate and efficient facemask categorization by leveraging pre-trained convolution neural networks (CNNs). The trained model is subsequently integrated into the pipeline used to analyze videos in real-time, enabling instantaneous facemask detection.