Research On Driver Emotion Recognition Based On Recognition Process And Facial Expression

There is a survey reports that accidents related to human factors account for 90%of all traffic accidents,of which more than 70% are caused by drivers.Therefore,monitoring drivers is an important approach to improve traffic safety,and emotions are crucial as an important factor affecting drivers’ driving performance.The emotion regulation of the drivers’ emotion state through the human-computer interaction in the car can reduce the negative impact of emotions on driving.Some existing researches mostly focus on the recognition of some negative discrete emotions of the driver(such as anger,sadness),and cannot realize the recognition of multiple discrete emotions and continuous emotions,and the engineering application is weak.Therefore,based on the prior knowledge of human cognitive process theory,the driver emotion recognition process is further modeled,and this paper proposes and implements a driver emotion recognition model based on cognitive process and facial expressions,which can identify various discrete drivers emotions and three-dimensional continuous emotions.Induced in this article,through the study and research of related theoretical and technical knowledge such as emotion definition,emotion classification,and the influence of emotion on driving,considering that the driver is in a high cognitive load state in the driving scene,the use of non-intrusive and non-contact measurement methods appears to be particularly importantly,according to the emotional cognitive process model,a driver emotion recognition method based on cognitive process and facial expressions is proposed.Due to the differences in the expressions of emotions in different scenarios,and there is currently no public drivers’ emotional facial expressions dataset in driving scenarios,this paper designs an experiment to collect the drivers’ emotional facial expression data set,and annotates the drivers’ emotions from the Differential Emotions Scale(DES)and Self-Assessment Manikins(SAM),and realizes the driver’s discrete emotions and dimension emotion recognition based on cognitive processes and facial expressions.By using various evaluation metrics on different facial expressions public datasets to validate,the driver emotion recognition method proposed in this paper is better than the currently widely used hybrid CNN(Convolutional Neural Networks)-LSTM(Long Short-term Memory)model,and it’s performance is much better than the baseline model,the effectiveness of the driver emotion recognition model is validated.