<![CDATA[The instant spread of COVID-19 has underscored the need for effective measures such as wearing face masks to control transmission. As a response, facemask detection systems using advanced machine learning techniques have become essential for ensuring compliance and public safety. This research focused on developing a system for detecting facemask usage using a hybridized approach comprising of Convolutional Neural Networks (CNN), Principal Component Analysis (PCA), and the Common Objects in Context (COCO) model. A hybridized detection model is often explored to enhance the precision and efficiency of previous methods that leveraged traditional machine learning or deep learning for the same task. Hence, this system effectively identifies whether individuals are properly wearing masks, not wearing masks at all, or wearing masks improperly from images and real-time video streams using bounding boxes. The results demonstrate that the hybrid approach achieves high accuracy in detecting various facemask conditions across different scenarios. Evaluation metrics such as Average Precision (AP) and Average Recall (AR) indicate the model's robustness, with a reported AP value of 70% and an AR value of 81%, primarily evaluated on larger objects within images. Further evaluations involving different individuals and types of facemasks revealed variability in detection accuracy, highlighting the model's effectiveness and areas for improvement. Nevertheless, the development and deployment of facemask detection systems are crucial for managing public health and ensuring safety in the face of ongoing and future pandemics.]]>
Copyrights © 2024
