Driver emotion recognition and uncertainty evaluation method based on deep learning

Abstract

The invention relates to a driver emotion recognition and uncertainty evaluation method based on deep learning, and belongs to the field of driver emotion recognition. And the image enhancement module adopts a self-adaptive histogram equalization technology, so that the influence caused by non-uniform illumination in the vehicle is effectively solved, and the image quality is improved. The visual feature extraction module uses an improved Wide ResNet network, and by adding a spectrum normalization layer and a Gaussian process layer, not only can abundant emotional features be extracted, but also the distance relationship between the features can be perceived, and the processing ability of the model to complex and uncertain data can be enhanced. And the emotion recognition module classifies the probability density distribution obtained by the Gaussian process layer by using a Bayesian theory system, and calculates the Gaussian process posterior through random feature extension Laplacian approximation to realize efficient emotion recognition. By introducing an uncertainty evaluation function, the distance between the test sample and the training sample is effectively predicted, and decision optimization is provided for driver emotion recognition in an automatic driving environment.

Description

Technical Field

[0001] The present invention belongs to the field of driver emotion recognition and relates to a driver emotion recognition and uncertainty assessment method based on deep learning. Background Art

[0002] With the rapid development of modern society and the continuous advancement of computer technology, intelligent driving systems have become a global research hotspot. As an important research topic in the field of intelligent driving systems, driver emotion recognition has attracted the attention of the scientific research community. At the same time, with the increase in computing power and the decrease in computing costs, computer technology has made significant progress in the past few decades. This development is not only reflected in the rapid iteration of hardware, such as more powerful processors and larger storage devices, but also in the innovation of software and algorithms. In particular, in the field of deep learning, major breakthroughs have been made in recent years. Deep learning technology has demonstrated excellent performance in many fields such as image recognition and natural language processing.

[0003] However, the emergence of life pressure and mental health problems in modern society has a significant impact on the emotional state of drivers and traffic safety. The rapidly developing social environment and increasing work pressure have led to a large number of mental health problems, such as stress, anxiety and depression. Emotions are a kind of behavior, hormone and autonomous response pattern that promotes survival. Bad emotions not only affect the daily quality of an individual's life, but are also directly related to traffic safety and the emotional state of the driver. Especially when subjected to strong emotional fluctuations, their judgment and reaction speed may be affected, thereby increasing the risk of traffic accidents. For example, emotional fluctuations may cause the driver's attention to be distracted, slow to react, or even take risky driving behaviors. Therefore, monitoring and management of the driver's emotional state has become an important part of ensuring road traffic safety. With the advancement of deep learning and artificial intelligence technology, driver emotion recognition methods based on deep neural networks have emerged, providing the possibility of real-time monitoring and evaluation of the driver's emotional state, opening up new ways to improve traffic safety and optimize driving experience.

[0004] In the driver emotion recognition method. Yang's high-purity feature separation method, through two partial feature exchanges and reconstruction loss constraints, realizes the effective separation of expression-related features and irrelevant features, and improves the recognition accuracy. Wang realized the recognition of driver's anger and conducted risk assessment on drivers in this mood, achieving good recognition accuracy. However, at present, there are relatively few studies on combining uncertainty estimation with facial expression recognition. In order to solve the problem of lack of uncertainty assessment ability of expression state, it will be very helpful to apply uncertainty assessment to complex system reasoning in uncertain environment, such as adding uncertainty assessment to driver emotion analysis. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a driver emotion recognition and uncertainty assessment method based on deep learning. While identifying the driver's emotion, the distance between the new test example and the previously trained example is predicted to achieve the function of uncertainty assessment.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for driver emotion recognition and uncertainty assessment based on deep learning, the method comprising the following steps:

[0008] Adaptive histogram equalization is performed on the driver image to solve the problem caused by uneven lighting inside the car;

[0009] The improved Wide ResNet network is used to extract emotional features from driver images, where the improved Wide ResNet network includes:

[0010] At least four convolutional layers, each of which is followed by a spectral normalization (SN) layer;

[0011] Replace the dropout layer, fully connected layer, and softmax layer close to the output in the original Wide ResNet network with a Gaussian process GP layer;

[0012] The Bayesian theoretical system is used to classify the probability density distribution obtained through the Gaussian process layer, where the posterior of the Gaussian process is calculated using the random feature extended Laplace approximation.

[0013] Further, the adaptive histogram equalization processing includes:

[0014] Split the image into blocks of equal size;

[0015] Calculate the gray level contained in each sub-block;

[0016] Count the number of gray levels contained in each gray level;

[0017] Calculate the image histogram of the sub-image;

[0018] Calculate the transformation function and merge the sub-blocks into the entire image.

[0019] Furthermore, the improved Wide ResNet network includes:

[0020] The first convolutional layer has 64 3X3 convolution kernels;

[0021] The second convolutional layer has 128 3X3 convolution kernels;

[0022] The third convolutional layer has 256 3X3 convolution kernels;

[0023] The fourth convolutional layer has 512 3X3 convolution kernels;

[0024] The convolutional layers are connected with a 2X2 pooling layer maxpool, with a moving step of 16;

[0025] Introduce residual modules to ensure gradient correlation between levels;

[0026] The output feature map is nonlinearly transformed using the PReLU activation function.

[0027] Further, the implementation method of the spectrum normalization SN layer is:

[0028] Initialization: Select a random initial vector;

[0029] Iterative calculation: use the initial vector and the parameter matrix to perform matrix multiplication; normalize the result vector; perform matrix multiplication of the normalized result vector and the parameter matrix; normalize the result vector;

[0030] After K iterations, the approximate value of the maximum singular value is:

[0031] where σ l (W) is the lth singular value of the matrix W, indicating the degree of expansion and contraction of the matrix in a certain direction; W represents a matrix, and K represents the number of iterations; is the lth left singular vector u obtained after the Kth iteration i The transpose of is the lth right singular vector v after the Kth iteration l ;

[0032] Divide each parameter matrix by it for normalization purposes.

[0033] Furthermore, the Gaussian process GP layer is implemented as follows:

[0034] Training data: There are N training samples where y i is the label of the i-th training sample, x i is the input feature of the i-th training sample;

[0035] Hidden representation: For each input x i Transformed into a hidden representation h through a hidden representation function i =h(x i );h(x i ) is the input feature x i The result after mapping to the latent space;

[0036] Conditional Gaussian process: Based on these hidden representations h i , define a conditional Gaussian process g, whose output at N hidden representation points is a vector g N×1 =[g(h1),...,g(h N )] T ;

[0037] Gaussian process prior with RBF kernel: Gaussian process model assumes g N×1 It follows a multivariate normal distribution with mean zero and covariance matrix σ 2 K, where the elements of the covariance matrix K are calculated using the RBF kernel function:

[0038] RBF kernel function: The RBF kernel function is defined as: K(x,x′)=exp(-γ‖xx′‖ 2 ), where x and x′ are two input vectors, ‖xx′‖ is the Euclidean distance between x and x′; γ is a parameter that controls the width of the kernel function;

[0039] Posterior distribution: Given the observed data The posterior distribution of g is calculated using the Bayesian formula: Among them, p(g) is the prior distribution and p(D|g) is the likelihood function of the observed data.

[0040] Furthermore, the dropout layer, fully connected layer, and softmax layer close to the output in the original Wide ResNet network are replaced with the Gaussian process GP layer, and the loss function formula is: Where N is the number of samples, C is the number of categories, and γ i,j represents the actual probability of labeling the i-th sample as the j-th class, μ i,j represents the average probability of labeling the i-th sample as the j-th class predicted by GP.

[0041] A driver emotion recognition and uncertainty assessment system based on deep learning, the system comprising:

[0042] An image input device, used for acquiring an image of the driver;

[0043] An image enhancement module, used for performing adaptive histogram equalization processing on the driver image;

[0044] Visual feature extraction module, used to extract emotional features from images using the improved Wide ResNet network;

[0045] The emotion recognition module is used to classify the probability density distribution obtained through the Gaussian process layer, where the posterior of the Gaussian process is calculated using the random feature extended Laplace approximation;

[0046] The image input device head is responsible for capturing the driver image and inputting the image data into the image enhancement module;

[0047] The image enhancement module performs adaptive histogram equalization on the input driver image to solve the problem caused by uneven lighting inside the car, and outputs the processed image;

[0048] The processed image is passed as input to the visual feature extraction module;

[0049] The visual feature extraction module uses the improved Wide ResNet network to extract emotional features in the image and passes the extracted features as input to the emotion recognition module;

[0050] The emotion recognition module uses the Bayesian theoretical system to classify the probability density distribution obtained through the Gaussian process layer and outputs the driver emotion recognition results.

[0051] Furthermore, the functions of the image enhancement module, the visual feature extraction module and the emotion recognition module are implemented by a deep learning model.

[0052] Furthermore, the image input device is a camera.

[0053] The beneficial effects of the present invention are:

[0054] (1) Through the improved Wide ResNet network, the present invention can effectively extract the emotional features in the driver's facial image and combine it with the Gaussian process layer for classification, thereby achieving high-precision driver emotion recognition.

[0055] (2) The present invention introduces an uncertainty assessment function, which provides a confidence assessment for the driver emotion recognition results by predicting the distance between the test sample and the training sample, helping the decision-making system to identify potential risks and take corresponding measures, such as reminding the driver to pay attention to safe driving or taking auxiliary driving measures.

[0056] (3) The introduction of the spectral normalization layer and the Gaussian process layer enhances the model’s ability to process complex and uncertain data, enabling the system to maintain a high level of recognition accuracy when faced with interference factors such as lighting changes and differences in driver characteristics.

[0057] (4) The random feature expansion Laplace approximation of the Gaussian process layer enables the model to maintain a high generalization ability even with less training data and adapt to different driving environments and driver characteristics.

[0058] (5) The present invention can monitor the driver's emotional state in real time and, combined with the uncertainty assessment results, promptly identify potential risks, provide safety protection for the automatic driving system, and reduce the probability of traffic accidents.

[0059] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0061] Figure 1 It is the Wide Resnet model of the SNGP collection;

[0062] Figure 2 This is the overall framework diagram of the driver emotion recognition method and uncertainty assessment method. DETAILED DESCRIPTION

[0063] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0064] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0065] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0066] See also Figure 1 and Figure 2 , this method is divided into three stages:

[0067] The first stage denoises the input image.

[0068] The second stage uses the WideResnet network to extract image features as input for the subsequent classifier.

[0069] In the third stage, the image features are placed into the Gaussian process layer for classification and recognition, and uncertainty recognition is performed.

[0070] The present invention mainly includes the following three modules:

[0071] 1. Image enhancement module

[0072] The image enhancement module mainly solves the problem of uneven lighting in the driver's image by adaptive histogram equalization, and performs histogram equalization in local areas to compensate for local lighting changes. The specific steps include:

[0073] (1) Divide the image into blocks of equal size

[0074] (2) Calculate the gray level r contained in each sub-block k

[0075] (3) Count the number n contained in each gray level k

[0076] (4) Calculate the image histogram of the sub-image

[0077] (5) Calculate the transformation function, that is:

[0078]

[0079] Merge the sub-blocks into the entire image.

[0080] This method can effectively handle the problem of uneven lighting in images while retaining more image details.

[0081] 2. Visual feature extraction module:

[0082] We use the WideResNet network to extract emotional features from images. The Wide ResNet model extracts image features. Wide ResNet is a variant of ResNet, which improves performance by adjusting the width (i.e., the number of channels or filters per layer) and depth (i.e., the number of network layers) of the network. Wide ResNet achieves improved accuracy by increasing the width of the network layer rather than the depth.

[0083] The uncertainty assessment function is implemented using the spectral normalized neural Gaussian process. Spectral-normalized Neural Gaussian Process (SNGP) is a method that improves the distance perception ability of modern neural networks by adding a weight normalization step during training and replacing the output layer with a Gaussian process. It aims to improve the ability of neural networks to handle uncertainty, especially in the prediction of novel or out-of-distribution (OOD) data, and can significantly enhance the performance of the model when dealing with complex and uncertain data.

[0084] The improved Wide ResNet has 4 convolutional layers, namely: the first convolutional layer, the second convolutional layer, the third convolutional layer, and the fourth convolutional layer. The first convolutional layer has 64 3X3 convolutional kernels, the second convolutional layer has 128 3X3 convolutional kernels, and the accumulation layer 3 has 256 3X3 convolutional kernels. The fourth layer has 512 3X3 convolutional kernels. The pooling layer between every two layers is a 2X2 pooling layer (maxpool) with a moving step of 16. The residual module is introduced to ensure the gradient correlation between levels. Then, the output feature map will be nonlinearly transformed using the activation function. The activation function uses the PReLU activation function, which has better effects in the classification field. The formula is as follows:

[0085]

[0086] Uncertainty-aware Wide ResNet: Using the attractive properties of SN (Spectral Normalization) and GP (Gaussian Process), we develop an uncertainty-aware Wide ResNet, namely SNGP-G, whose architecture has two key differences compared to the original WideResNet:

[0087] 1) Each convolutional layer in the original architecture is followed by a SN layer. These convolutional layers include: the convolutional layers in the main network, such as Conv Layer 1 and Conv Layer 2, which make the residual layer have distance perception capabilities.

[0088] 2) Replace the original dropout layer, fully connected layer, and softmax layer close to the output with the GP layer to enable it to have the ability to calculate distance.

[0089] The realization of spectral normalization is to solve the singular value and obtain the spectral norm of each layer parameter matrix by using the "power iteration method" to approximate the solution. The iterative process is as follows:

[0090] (1) Initialization: Select a random initial vector Typically, it is a randomly generated Gaussian vector from a standard normal distribution.

[0091] (2) Calculation

[0092] (3) Normalization

[0093] (4) Calculation

[0094] (5) Normalization

[0095] (6) Convergence: After K iterations, the approximate value of the maximum singular value is:

[0096]

[0097] After obtaining the spectral norm, the parameters on each parameter matrix are divided by it to achieve normalization and make the convolution part satisfy the Lipschitz continuity condition. 1-Lipschitz refers to a function f:R n →R is called a 1-Lipschitz function if for all x,y∈R n , which satisfies the following inequality:

[0098] |f(x)-f(y)|≤|xy|

[0099] |f(x)-f(y)|≤|xy|

[0100] This means that the function f will not change faster than the input, that is, the increment of the function f will not exceed the absolute value of the increment of the input point. Such functions are very important in optimization, machine learning, and numerical analysis because they ensure the stability and controllability of the output.

[0101] 3. Emotion Recognition Module:

[0102] In the emotion recognition module, the features obtained by WideResNet are used as input features. With the help of the Bayesian theoretical system, the probability density distribution obtained by the Gaussian process layer is classified using the knowledge of the potential class conditional probability density function and the maximum likelihood estimation in the Gaussian process layer.

[0103] In order to ensure that it can adapt to more environments, this paper uses the random feature extension Laplace approximation of the Gaussian process to realize the Gaussian process posterior. This can have high scalability with relatively few changes to the training model, and effectively calculate the uncertainty of the input image while ensuring a faster speed. And in the Gaussian process layer, for a test input x * Its posterior variance is proportional to its distance L2 to the training data point in latent space. Specifically:

[0104] (1) Training data: Assume we have N training samples where y i is the corresponding label, x i are input features.

[0105] (2) Hidden representation: Each input x i Through a hidden representation function h i Convert to hidden representation h i =h(x i ).

[0106] (3) Conditional Gaussian process: Based on these hidden representations h i , define a conditional Gaussian process g, whose output at N hidden representation points is a vector g N×1 =[g(h1),...,g(h N )] T .

[0107] (4) Gaussian process prior with RBF kernel: This Gaussian process model assumes that g N×1 It follows a multivariate normal distribution with mean zero and covariance matrix σ 2 K, where the elements of the covariance matrix K are calculated using the RBF kernel function:

[0108]

[0109] (5) RBF kernel function: The RBF kernel function is defined as:

[0110]

[0111] Here, h i,h j Indicates hidden indication h i and h j The Euclidean distance between .

[0112] (6) Posterior distribution: Given the observed data We calculate the posterior distribution of g using the Bayesian formula:

[0113]

[0114] Among them, p(g) is the prior distribution mentioned above, and p(D|g) is the likelihood function of the observed data.

[0115] After replacing the last three layers in the original Wide ResNet with Gaussian process layers, the loss function formula is as follows:

[0116]

[0117] Where N is the number of samples, C is the number of categories, and γ i,j represents the actual probability of labeling the i-th sample as the j-th class, μ i,j represents the average probability of labeling the i-th sample as the j-th class predicted by GP.

[0118] Example

[0119] 1. Evaluation indicators:

[0120] This work uses recall, precision, and F1-Score as performance evaluation indicators. Recall refers to the proportion of samples that the model correctly identifies as positive among all samples that are actually positive. It reflects the model's coverage of actual positive samples. Precision refers to the proportion of samples that the model predicts as positive among all samples that are actually positive. It reflects the accuracy of the model's prediction of positive samples. F1 score is the harmonic mean of precision and recall. It takes into account both precision and recall and is an indicator that strikes a balance between the two.

[0121] 2. Experimental data:

[0122] The dataset used in this work is MLI-DER (Multiple Light Intensities DriverEmotionRecognition). The MLI-DER dataset is a high-quality dataset specifically used to study driver emotion recognition. The dataset is designed to overcome the interference of different driver characteristics (such as gender, skin color, age, appearance) and changes in in-car lighting conditions on emotion recognition. The dataset contains 37 subjects (12 females and 25 males) with an age range of 21 to 37 years old. A total of 442 video samples were obtained, with a frame rate of 20fps and a resolution of 640x480 pixels. The key frames extracted from the video formed 4451 image samples.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A driver emotion recognition and uncertainty assessment method based on deep learning, characterized by: The method comprises the following steps: Adaptive histogram equalization is performed on the driver image to solve the problem caused by uneven lighting inside the car; The improved Wide ResNet network is used to extract emotional features from driver images, where the improved Wide ResNet network includes: At least four convolutional layers, each of which is followed by a spectral normalization (SN) layer; Replace the dropout layer, fully connected layer, and softmax layer close to the output in the original Wide ResNet network with a Gaussian process GP layer; The Bayesian theoretical system is used to classify the probability density distribution obtained through the Gaussian process layer, where the posterior of the Gaussian process is calculated using the random feature extended Laplace approximation.

2. The method for driver emotion recognition and uncertainty assessment based on deep learning according to claim 1, characterized in that: The adaptive histogram equalization process comprises: Split the image into blocks of equal size; Calculate the gray level contained in each sub-block; Count the number of gray levels contained in each gray level; Calculate the image histogram of the sub-image; Calculate the transformation function and merge the sub-blocks into the entire image.

3. The method for driver emotion recognition and uncertainty assessment based on deep learning according to claim 1, characterized in that: The improved Wide ResNet network includes: The first convolutional layer has 64 3X3 convolution kernels; The second convolutional layer has 128 3X3 convolution kernels; The third convolutional layer has 256 3X3 convolution kernels; The fourth convolutional layer has 512 3X3 convolution kernels; The convolutional layers are connected with a 2X2 pooling layer maxpool, with a moving step of 16; Introduce residual modules to ensure gradient correlation between levels; The output feature map is nonlinearly transformed using the PReLU activation function.

4. The method for driver emotion recognition and uncertainty assessment based on deep learning according to claim 1, characterized in that: The implementation method of the spectrum normalization SN layer is: Initialization: Select a random initial vector; Iterative calculation: use the initial vector and the parameter matrix to perform matrix multiplication; normalize the result vector; perform matrix multiplication of the normalized result vector and the parameter matrix; normalize the result vector; After K iterations, the approximate value of the maximum singular value is: where σ l (W) is the lth singular value of the matrix W, indicating the degree of expansion and contraction of the matrix in a certain direction; W represents a matrix, and K represents the number of iterations; is the lth left singular vector u obtained after the Kth iteration i The transpose of is the lth right singular vector v after the Kth iteration l ; Divide each parameter matrix by it for normalization purposes.

5. The method for driver emotion recognition and uncertainty assessment based on deep learning according to claim 1, characterized in that: The Gaussian process GP layer is implemented as follows: Training data: There are N training samples where y i is the label of the i-th training sample, x i is the input feature of the i-th training sample; Hidden representation: For each input x i Transformed into a hidden representation h through a hidden representation function i =h(x i );h(x i ) is the input feature x i The result after mapping to the latent space; Conditional Gaussian process: Based on these hidden representations h i , define a conditional Gaussian process g, whose output at N hidden representation points is a vector g N×1 =[g(h1),...,g(h N )] T ; Gaussian process prior with RBF kernel: Gaussian process model assumes g N×1 It follows a multivariate normal distribution with mean zero and covariance matrix σ 2 K, where the elements of the covariance matrix K are calculated using the RBF kernel function: RBF kernel function: The RBF kernel function is defined as: K(x,x′)=exp(-γ‖xx′‖ 2 ), where x and x′ are two input vectors, ‖xx′‖ is the Euclidean distance between x and x′; γ is a parameter that controls the width of the kernel function; Posterior distribution: Given the observed data The posterior distribution of g is calculated using the Bayesian formula: Among them, p(g) is the prior distribution and p(D|g) is the likelihood function of the observed data.

6. The method for driver emotion recognition and uncertainty assessment based on deep learning according to claim 1, characterized in that: The dropout layer, fully connected layer, and softmax layer close to the output in the original Wide ResNet network are replaced with the Gaussian process GP layer, and the loss function formula is: Where N is the number of samples, C is the number of categories, and γ i,j represents the actual probability of labeling the i-th sample as the j-th class, μ i,j represents the average probability of labeling the i-th sample as the j-th class predicted by GP.

7. A driver emotion recognition and uncertainty assessment system based on deep learning, characterized by: The system includes: An image input device, used for acquiring an image of the driver; An image enhancement module, used for performing adaptive histogram equalization processing on the driver image; Visual feature extraction module, used to extract emotional features from images using the improved Wide ResNet network; The emotion recognition module is used to classify the probability density distribution obtained through the Gaussian process layer, where the posterior of the Gaussian process is calculated using the random feature extended Laplace approximation; The image input device head is responsible for capturing the driver image and inputting the image data into the image enhancement module; The image enhancement module performs adaptive histogram equalization on the input driver image to solve the problem caused by uneven lighting inside the car, and outputs the processed image; The processed image is passed as input to the visual feature extraction module; The visual feature extraction module uses the improved Wide ResNet network to extract emotional features in the image and passes the extracted features as input to the emotion recognition module; The emotion recognition module uses the Bayesian theoretical system to classify the probability density distribution obtained through the Gaussian process layer and outputs the driver emotion recognition results.

8. The driver emotion recognition and uncertainty assessment system based on deep learning according to claim 7, characterized in that: The functions of the image enhancement module, visual feature extraction module and emotion recognition module are implemented by a deep learning model.

9. The driver emotion recognition and uncertainty assessment system based on deep learning according to claim 7, characterized in that: The image input device is a camera.

Images