Title: Comparative model study for driver drowsiness detection system: a road safety mechanism for smart cities 5.0 in developing countries of South-Asian continent

Authors: Rohit Rastogi; Manvi Tyagi; K. Pranav; Pranjal Vaidywan

Addresses: Department of CSE, ABES Engineering College, Ghaziabad, UP, 201009, India ' Department of CSE, ABES Engineering College, Ghaziabad, UP, 201009, India ' Department of CSE, ABES Engineering College, Ghaziabad, UP, 201009, India ' Department of CSE, ABES Engineering College, Ghaziabad, UP, 201009, India

Abstract: In this research work, the author team implemented a driver drowsiness detection system using three different models: VGG16, InceptionV3, and Xception. Transfer learning was employed, leveraging the prior knowledge encoded in these pre-trained models to enhance the efficiency of our system. The author team utilised Haar Cascade for face and eyes detection from the video stream. After detecting eyes, the trained model predicted whether the eyes were open or closed. If the person's eyes remained closed for an extended period, an alarm would go off, waking up the driver. Live testing on the laptop itself demonstrated that the Xception model provided tremendous results, reinforcing the decision to use transfer learning with the Xception model over VGG16 and InceptionV3.While our comparison included three models, it's essential to acknowledge that transfer learning offers a powerful approach to leverage pre-existing knowledge and enhance the performance of machine learning models for specific tasks, such as driver drowsiness detection.

Keywords: computer vision; datasets; deep learning; drowsiness; Haar cascade; InceptionV3; neural networks; OpenCV; transfer learning; VGG16.

DOI: 10.1504/IJSSE.2026.156328

International Journal of System of Systems Engineering, 2026 Vol.16 No.4, pp.409 - 449

Received: 23 Dec 2023
Accepted: 18 Apr 2024
Published online: 13 Sep 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article