2025, Vol. 6, Issue 2, Part A
Real-time mask detection system using deep learning-based image classification
Author(s): Mohammed Nasih Ismael
Abstract: This paper addresses the worldwide need for improved public health monitoring due to the current pandemic by proposing the development of a real-time face mask detection system built with deep learning image classification. We built a custom dataset of 7,553 facial images with half of the dataset containing facial images of individuals wearing a mask and the other half without a mask to eliminate bias. We then trained a Convolutional Neural Network (CNN) model with TensorFlow utilizing this dataset, which resulted in a validation accuracy of 94% and an AUC score of 0.98; indicating strong generalization performance for this model. We also evaluated the model with multiple performance type metrics such as precision, recall, F1-score, and ROC-AUC metrics. We also presented these findings visually through confusion matrices and learning curves. A primary difference in this solution that relied on classification is the input to our models are pre-cropped facial images of individuals instead of relying on object detection. In the end, the proposed system provides a lightweight, efficient, and accurate method that can be deployed in real time in controlled environments. In summary, the results suggest that the CNN based classification can be used for successful automated face mask detection and provides a basis for further application to surveillance systems or embedded devices.
DOI: 10.22271/27084531.2025.v6.i2a.95
Pages: 17-27 | Views: 69 | Downloads: 28
Download Full Article: Click Here
How to cite this article:
Mohammed Nasih Ismael. Real-time mask detection system using deep learning-based image classification. Int J Res Circuits Devices Syst 2025;6(2):17-27. DOI: 10.22271/27084531.2025.v6.i2a.95
