Giap, Yo Ceng and Muljono, Muljono and Affandy, Affandy and Basuki, Ruri Suko and Dewi, Deshinta Arrova (2026) Establishing SSD-MobileNetV2 as a Robust Baseline for Driver Drowsiness Detection Toward IoT-Ready in-Driving Safety Systems. International Journal of Transport Development and Integration, 10 (2). pp. 441-453.
![[img]](https://repositori.buddhidharma.ac.id/style/images/fileicons/text.png) |
Text (Article)
Establishing SSD-MobileNetV2 as a Robust Baseline for Driver Drowsiness Detection Toward IoT-Ready in-Driving Safety Systems.pdf - Published Version Download (995kB) |
|
Text (Turnitin)
Turnitin - Establishing SSD-MobileNetV2.pdf - Published Version Download (987kB) |
|
Text
Korespondensi Jurnal.pdf Download (2MB) |
Abstract
Driver drowsiness is one of the major reasons behind road accidents, emphasizing the need for accurate and efficient fatigue detection systems that can help monitor practical in-vehicle environments. While significant progress has been made in visual fatigue detection based on deep learning, many previous studies have been performed using a single dataset for training or controlled environments for testing. In this paper, we examine the reliability of lightweight driver-monitoring architectures for vision-based driver drowsiness detection based on three heterogeneous public datasets, i.e., Yawning Detection Dataset (YawDD), Driver Drowsiness Dataset (DDD), and National Tsing Hua University Drowsy Driving Dataset (NTHU-DDD), which cover different lighting conditions, facial characteristics, and head poses as encountered in driving scenarios. Among the considered architectures, Single Shot Detector (SSD)-MobileNetV2 was the most consistent, yielding an accuracy of 92%, precision of 93%, recall of 92%, and F1-score of 92% while also being computationally lighter than the other considered architectures. Reliability of the proposed architecture was statistically validated using the McNemar Test and 95% Confidence Intervals (CI). Our results show that SSD-MobileNetV2 could be a promising baseline for future lightweight drivermonitoring systems for heterogeneous driving environments.
| Item Type: | Article |
|---|---|
| Uncontrolled Keywords: | Driver drowsiness detection; IoT-ready safety systems; Lightweight deep learning; Intelligenttransportation systems; In-vehicle monitoring; Transportation services; Vision-based driver monitoring |
| Subjects: | 000 Karya Umum > 006 Metode Komputer Tertentu > 006.242 Kode Bar > 006.3 Kecerdasan Buatan 000 Karya Umum > 006 Metode Komputer Tertentu > 006.3 Kecerdasan Buatan |
| Divisions: | Fakultas Sains & Teknologi > Teknik Informatika |
| Depositing User: | Hariyanto Rie |
| Date Deposited: | 29 Aug 2026 09:51 |
| Last Modified: | 31 Aug 2026 04:06 |
| URI: | https://repositori.buddhidharma.ac.id//id/eprint/3482 |
Actions (login required)
 |
View Item |