Driver Sleepiness Detection Algorithm Based on Relevance Vector Machine

Driver sleepiness is one of the most important causes traffic accidents. Efficient and stable algorithms are crucial for distinguishing nonfatigue from fatigue state. Relevance vector machine (RVM) as a leading-edge detection approach allows meeting this requirement represents potential solution state detection. To accurately effectively identify driver’s reduce number accidents caused by driver sleepiness, paper considers degree mouth opening eye multi-source related variables establishes classification non-fatigue states based on literature investigation. On basis, an RVM model automatic proposed. Twenty male respondents participated in data collection process total 1000 datasets driving status (half half fatigue) were obtained. The results recognition analysed different classifiers. show that accuracy RVM-driven classifiers with kernel functions was higher than 90%, which indicated mouth-opening index used work closely to Based obtained results, proposed identification method has improve accuracy. More importantly, it provides scientific theoretical basis development warning methods.