DriSafePh: An IoT Based Realtime Driver Drowsiness Detection System using Hybrid Machine learning Algorithm

Driver drowsiness contributes significantly to road accidents worldwide, and while drowsiness detection systems have already been implemented on higher-end cars, DriSafePh introduces an embedded system using a Raspberry Pi that can be implemented in any enclosed vehicle. DriSafePh aims to address th...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2024 14th International Conference on Software Technology and Engineering (ICSTE) S. 258 - 265
Hauptverfasser: Von C. Golosinda, Kenneth, Van D. Marcellones, Jasper, Ampongol, Luke Jellergil P., Magloyuan, Neil P., Cerna, Patrick D.
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 16.08.2024
Schlagworte:
Accuracy
Detection algorithms
drowsiness detection
eye closure
Hybrid Machine Learning Algorithm
Machine learning
Machine learning algorithms
Performance evaluation
Software
Software algorithms
Surveys
Text to speech
Vehicles
Yawn detection
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Driver drowsiness contributes significantly to road accidents worldwide, and while drowsiness detection systems have already been implemented on higher-end cars, DriSafePh introduces an embedded system using a Raspberry Pi that can be implemented in any enclosed vehicle. DriSafePh aims to address the prevalent issue of driver drowsiness on the road by developing a real-time driver drowsiness detection system and leveraging a hybrid machine learning algorithm that processes facial landmarks and gestures to accurately detect drowsiness, which has yet to be used in its field and will be compared to other algorithms already in use. Additionally, the researcher aims to provide a warning notification to the driver through a text-to-speech algorithm when the appropriate drowsiness level is detected and optimize the machine learning model's performance. The study conducted a survey to gather drivers' opinions on a drowsiness detection system and their personal experiences with drowsy driving. The results showed that the respondents viewed the working prototype positively. Additionally, the data revealed that more than 70% of the respondents had experienced drowsy driving, highlighting the need for such a device. The recommended approach of the DriSafePh model was evaluated using a performance matrix, and the detection for the face, eye, and mouth provides a high accuracy level of 96-99%. In the meantime, eye closure and yawn detection accuracy were able to match the other existing algorithms on the Raspberry Pi device with an accuracy of 96.89% and 85.93%, respectively.
DOI:10.1109/ICSTE63875.2024.00051