Some of the major causes that lead to road accidents are drowsiness, fatigue, or distraction of the driver. Drowsiness has become a growing problem with long distance drivers. The primary causes of drowsiness are long shifts at work and sleep deprivation. Sometimes, transport companies do not rotate drivers regularly between shifts. Thus, due to driving for an extended period, they feel sleepy. Such cases may lead to fatal accidents that may cause many casualties (for travel companies), tragic loss to a family, loss of goods (for goods transport companies), and may also cause massive economic losses. The signs of distraction and sleepiness in a driver can be discerned through their facial posture. The most popularly used methods to detect drowsy feeling of the driver are by calculating Eye Aspect Ratio (EAR) and Yawn Aspect Ratio (YAR) using landmarks of the face. The system activates alarm once the features extracted from the facial and body landmarks exceed generalized threshold value. In addition to EAR and YAR, the system includes features like Front Tilt Ratio, Side Tilt Angle, and Head-Shoulder Ratio. These features collectively contribute to detect the steadiness of driver. Experiments show that the algorithm has 91.8% detection accuracy. Our proposed solution aims to overcome the shortcomings in drowsiness detection.
An Image Processing Approach to Enhance Driver Distraction and Drowsiness Detection Algorithm
2024-03-01
3042282 byte
Conference paper
Electronic Resource
English