Citation

Yunos, Nur Syahidah and Yogarayan, Sumendra and Ahmad, Nazrul Muhaimin and Kannan, Subarmaniam (2025) A Comparison of CNN Architectures in Edge Computing Environments: Health Care Prospect. In: 6th International Conference on Advanced Information Technologies, ICAIT 2025, 3 November 2025, Yangon.

Text IEEE Xplore Full-Text PDF_2.pdf - Published Version Restricted to Repository staff only Download (4MB)

Abstract

Wearing a face mask has become an essential accountability for ensuring public safety in view of developing health regulations and the necessity for proactive safety measures. This research conducts a comparison of several lightweight neural network architectures, with a specific focus on their application in face mask detection for edge computing. Considering the intrinsic processing limits of edge devices such as raspberry pi, this research carefully evaluates the complexity of models and computational overhead. Through extensive evaluation using a balanced dataset of 11,792 images, we investigate the performance of architectures including CNN (Convolutional Neural Network), MobileNetV2, and EfficientNetB0. This includes model accuracy, Training time, inference time and model size. The results provided show that there are trade-offs, with some models has less time training but potentially less precise, and others being more accurate but slower. The work concludes by proposing potential future research directions, which include the investigation of novel architectures and advanced model optimization techniques.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Deep learning, edge computing
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics
Divisions:	Faculty of Information Science and Technology (FIST)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	17 Mar 2026 01:52
Last Modified:	17 Mar 2026 01:52
URII:	http://shdl.mmu.edu.my/id/eprint/15457

Downloads

Edit (login required)