Real-Time Face Mask Detection

Real-Time Face Mask Detection

Muskan Kumari

Dept.of Computer Science Engineering Parul University

Vadodara, India muskan.kumari24721@gmail.com

Lonka Nithin

Dept.of Computer Science Engineering Parul University

Vadodara, India 210303124697@paruluniversity.ac.in.

Kuppireddy Charan Teja Reddy

Dept.of Computer Science Engineering Parul University

Vadodara, India 210303124679@paruluniversity.ac.in

Lakshmisetty Eswara Harshitha

Dept.of Computer Science Engineering Parul University

Vadodara, India 210303124687@paruluniversity.ac.in

Kunisetty Sai Naveen

Dept.of Computer Science Engineering Parul University

Vadodara, India 210303124677@paruluniversity.ac.in

Abstract—Real-time face mask detection leverages computer vision and deep learning to identify individuals wearing masks in videos or live camera feeds. It involves two key steps: face detec- tion to locate human faces and mask detection using trained deep learning models to analyze the facial region for mask presence. This technology offers benefits like public health monitoring and security access control, but requires considerations for accuracy and real-time processing efficiency. Ongoing research focuses on improving these aspects for wider deployment.The widespread adoption of face masks as a preventive measure against infectious diseases has necessitated the development of efficient face mask detection systems. In this paper, we propose a real-time face mask detection system utilizing deep learning techniques. The system employs a convolutional neural network (CNN) architecture, specifically designed to accurately detect the presence or absence of face masks in live video streams. Initially, the proposed system preprocesses the input video frames to extract facial regions using a pre-trained face detection model. These facial regions are then fed into the CNN for classification into two categories: with mask and without mask. The CNN model is trained on a diverse dataset of annotated facial images with and without masks, ensuring robustness and generalization. To enhance real-time performance, we optimize the model architecture for efficient inference on resource-constrained devices, such as embedded systems and mobile devices. We leverage techniques such as model pruning, quantization, and parallelization to achieve low- latency inference without compromising accuracy. Experimental evaluations conducted on various real-world scenarios demon- strate the effectiveness and efficiency of the proposed system. The system achieves high accuracy in detecting face masks in real- time while maintaining low computational overhead. Moreover, extensive testing under different lighting conditions, angles, and occlusions validates its robustness and practical viability. Overall, the proposed real-time face mask detection system presents a scalable and deployable solution for ensuring compliance with face mask mandates in public spaces, contributing to public health efforts to mitigate the spread of infectious diseases