An Improved Digital Image Processing Based Driver Sleep Identification and Alert System Using Internet of Things with Smart Sensors Association

• Published in: 2024 Ninth International Conference on Science Technology Engineering and Mathematics (ICONSTEM) pp. 1 - 7
• Main Authors: N, Juliet, C, Prathap, S, Sujitha, V, Sandhiya, P, Surya Prakash, A, Mohana Priya
• Format: Conference Proceeding
• Language: English
• Published: IEEE 04.04.2024
• Subjects: Alert System; Artificial intelligence; CDDDS; Detectors; Digital Image Processing; Driver Sleep Identification; Feature extraction; Internet of Things; IoT; Real-time systems; SDSIM; Semantics; Sleep; Smart Sensors
• DOI: 10.1109/ICONSTEM60960.2024.10568610

Employing computer vision and artificial intelligence techniques, this project aims to develop and deploy a system that can identify when a driver is sleepy and provide a real-time warning. The camera-acquired driver footage had its facial and ocular pictures segmented using a Viola-Jones detector. A single image was created by combining the images of the left and right eyes. The result was a two-eye minimum-dimensional picture. Using Gabor filters, we were able to extract features from these photos. Images with open and closed eyes were classified using these attributes. Major traffic accidents are caused by drivers who are too sleepy, according to studies on the topic. Nowadays, sleepiness is mostly caused by drivers who are too exhausted to operate heavy vehicles. The primary factor contributing to the current upsurge in traffic accidents is, unfortunately, sleepiness. This escalates into a global crisis that requires immediate attention. Improving the ability to identify sleepiness in real time is the main objective of all technologies. A wide variety of sleepiness detectors based on various AI algorithms are already on the market. A related area of our study is driver drowsiness detection, which involves detecting tiredness in drivers by facial recognition and eye tracking. The system compares the extracted eye picture to the dataset. In order to assess the efficacy of the suggested method, which is known as the Semantic Driver Sleep Identification Model (SDSIM), it is cross-validated with an existing model known as the Conventional Driver Drowsiness Detection Scheme (CDDDS). The proposed approach follows the logic of an innovative technology called Internet of Things (IoT) and dataset helped the system determine that the eyes needed to be open for the system to continue tracking, and that if they were closed for a specific amount of time, an alarm might be sounded to notify the driver. We established a score that went down while the eyes were open and up when they were closed.