Comparative Study of Various Face Detection Methods for Driver Drowsiness Detection

Driver drowsiness is a significant contributing factor to vehicular accidents, resulting in severe injuries, fatalities, and substantial financial ramifications. The prevalence of disability among individuals has also experienced an upward trend, thereby posing challenges to their overall survival. The detection of driver drowsiness at an early stage is a significant worry, requiring the development of an intelligent system capable of effectively identifying this condition. By doing so, the occurrence of crashes can be prevented. The primary attributes exhibited by an individual experiencing drowsiness encompass the gradual closure of eyelids, oscillatory movements of the head in either a forward or backward manner, and frequent occurrences of yawning that is captured by mounting a camera. The prevalence of utilizing behavioral measurements for the identification of driver drowsiness has increased due to improvements in artificial intelligence (AI) technology. Numerous contemporary face detection systems have been developed, possessing the capability to swiftly record, identify, and analyze facial images in real-time, while simultaneously extracting salient facial characteristics. This research presents a comparative review of different face detection approaches with the aim of enhancing the identification of facial features in order to identify driver drowsiness. The findings suggest that MTCNN (Multitask Convolutional Neural Networks) exhibit superior performance compared to alternative face identification methods.