Enhancing Smart Infrastructure Monitoring in Response to Approaching Pandemics | ||||
| International Journal of Telecommunications | ||||
| Volume 05, Issue 01, January 2025, Page 1-14 PDF (1.67 MB) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/ijt.2025.351987.1077 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
Nourhan Osama Mohamed  1; khaled Abd El Salam Ali 2; Ahmed Magdy Mohamed2; Bassem ELhady ElNaghi2
| ||||
| 1Electrical Engineering Department, Faculty of Engineering, Suez Canal University, Ismailia, Egypt. | ||||
| 2Electrical Engineering Department, Faculty of Engineering, Suez Canal University, Ismailia, Egypt | ||||
| Abstract | ||||
| Airborne illnesses like chickenpox, influenza, and COVID-19 pose a major risk to public health since COVID-19 has killed about 7 million people. Wearing face masks has therefore become mandatory and significant. in order to prevent the spread of certain illnesses, particularly in healthcare institutions such as hospitals. This study introduces a scalable deep convolutional neural network (DCNN)-based face mask monitoring system that is better than manual surveillance, particularly in high-density settings. This study offers three methods: First, the pre-trained algorithms model, which included seven different algorithms and was optimized with hyperparameters to find optimal settings; the Darknet-53 algorithm performed the best among them, achieving an accuracy of 97.5%. The second was a customized DCNN model that achieved 96.4% accuracy in binary mask detection. The last suggested system is a hybrid model that improves the accuracy and stability of the model by using pre-trained algorithms as classifiers and a DCNN as a feature extractor. AlexNet and Darknet-53 were tested as classifiers in our study; Darknet-53's accuracy was 98%. | ||||
| Keywords | ||||
| Airborne Diseases; Deep Convolution Neural Network; Pre-trained Algorithms | ||||
|
Statistics Article View: 81 PDF Download: 71 |
||||
