Comprehensive Driver State Recognition Based on Deep Learning and PERCLOS Criterion

This system is mainly used for driver's mental state identification. It is concerned with deep learning technology. Deep convolutional neural network and face feature point processing method are adopted to analyze the driver's face information. The system also uses information comprehensive decision-making to analyze, so as to identify the driver's seven kinds of emotion and driver's fatigue condition. At the beginning of the system, it should collect the image of the identified person and record the collection time. And then it uses face recognition algorithm to process and outputs face recognition results. it inputs face recognition results into deep neural network for processing to get expression recognition and fatigue recognition results. Finally, it records the identification result and the corresponding collection time into the database in order as the data set. The expression and fatigue data in the database are analyzed in the real time by comprehensive weighted assessment method. Thus, the result of mental state recognition of the identified person is obtained. This allows machines to efficiently perceive and analyze human emotions and fatigue levels as well as to interact with humans in a more efficient manner.