Based On Deep Learning And Physiological Signals Mixed Emotion Recognition Method

In recent years,with the development of road traffic,more and more traffic problems and traffic accidents continue to appear.Among them,the more severe ones,such as stealing the steering wheel,road rage,drunk driving,cause inattention,brain excitement,and night roads.Car accidents caused by shocks and other situations are not uncommon.These accidents related to drivers or passengers are closely related to their own emotional changes.Looking for a way to monitor the driver ’ s mood fluctuations.When it is monitored that the driver’s or passengers’ mood changes are beyond the normal range,the warning will be used in time.When the situation is worse,the intelligent driving system will take over the driving of the vehicle,which can be effective To reduce the occurrence of such incidents.In this paper,deep learning methods are used to build convolutional neural networks to recognize facial expressions,combined with physiological signal monitoring methods,to build a mixed emotion recognition model,facing the driver,to monitor the driver’s emotions.Emotion recognition has been developed for a long time at home and abroad,and it has been applied in the medical field in the early days.With the development of automobile intelligence,people have begun to pay attention to the impact of driver emotions on driving.Emotion recognition has gradually been used in the field of car safety application.In the early days,traditional machine learning methods were used to classify physiological signals in different emotional states.As deep learning methods gradually entered people’s field of vision,there were gradually related researches on facial expression recognition based on deep learning.In this paper,a simulated driving experiment is used to obtain a data set of emotions under driving conditions.According to the driver’s physiological signal fluctuation changes,the facial expression pictures of the corresponding time period are intercepted,and the subjective judgment of the volunteers is used as the emotional label to set the label on the picture,and the relevant data set is obtained after the picture is preprocessed.There are many types of neural network structures.This paper compares the common CNN network and Xception network.Through the comparison on the test set,the recognition rate of the common CNN network is 60%,while the recognition rate of the Xception network on the same test set reaches 69.33%.The experiment found that the Xception network is similar Performance is better than that.The Xception network was further optimized,and the recognition rate increased to74.67%.By combining physiological signal monitoring and building a mixed emotion recognition model,some of the recognition errors of the neural network were "corrected",and the accuracy rate on the test set was increased to 82.67%.Based on the well-built emotion recognition model,this article puts forward a solution strategy to solve the emotional out-of-control in a man-machine co-driving vehicle.It can effectively reduce traffic accidents caused by emotional changes.