Real time drowsiness detection by image processing from eye blinks
4
Issued Date
2025-01-01
Resource Type
Scopus ID
2-s2.0-105023708709
Journal Title
Proceedings 2025 5th International Conference on Computer Communication and Information Systems Cccis 2025
Start Page
37
End Page
41
Rights Holder(s)
SCOPUS
Bibliographic Citation
Proceedings 2025 5th International Conference on Computer Communication and Information Systems Cccis 2025 (2025) , 37-41
Suggested Citation
Nakbuppa P., Asavaskulkiet K., Boonton N., Norasarn N., Tausiesakul B., Tiptipakorn S. Real time drowsiness detection by image processing from eye blinks. Proceedings 2025 5th International Conference on Computer Communication and Information Systems Cccis 2025 (2025) , 37-41. 41. doi:10.1109/CCCIS64581.2025.00014 Retrieved from: https://repository.li.mahidol.ac.th/handle/123456789/113455
Title
Real time drowsiness detection by image processing from eye blinks
Author's Affiliation
Corresponding Author(s)
Other Contributor(s)
Abstract
Transportation plays a crucial role in modern society, and many individuals rely on cars for commuting, which introduces potential risks to their safety. In Thailand, drowsy driving accidents rank as the third leading cause of traffic incidents. Our objective is to enhance road safety through a project that involves developing a Python-based camera system capable of detecting drowsiness, designed to be compatible with compact devices such as the Raspberry Pi. This camera will monitor eye blinks to identify signs of fatigue. This research's objective is to develop a system that can detect drowsy drivers and provide timely alerts to prevent accidents. We utilize MediaPipe and OpenCV to detect faces and incorporate data from research, such as PERCLOS to identify drowsiness. Our camera can accurately count eye blinks with 98% accuracy for participants wearing glasses and 95% accuracy for those without glasses on windows. On the Raspberry Pi 4 model B, the results are 97.33% accuracy at 300 lx for participants without glasses, 95.27% accuracy at 50 lx for participants without glasses, and 96.93% accuracy at 0 lx for participants without glasses. Then, this research aims to develop a camera system capable of counting eye blinks, indicative of drowsiness, and triggering an alarm with a buzzer when a drowsy driver is detected.