An intelligent driver distraction degree detection method, device, equipment, medium and product

Abstract

The application discloses a kind of intelligent driver distraction degree detection method, device, equipment, medium and product, involve distraction driving behavior identification field, this method includes: obtaining the original driving data set containing the visual behavior parameter of driver, manual operation parameter and vehicle dynamic parameter;Using SF 2 SOG method filters key features in original driving data set to construct training data set;Through condensation clustering algorithm, give the label of distraction degree to training data;Build complete training data set containing training data set and the label of distraction degree corresponding to each training sample data, and based on complete training data set, train stacked ensemble model containing multiple basic classifiers such as random forest classifier, extreme gradient boosting tree classifier and adaptive boosting classifier and a meta-classifier, realize the accurate prediction of distraction degree.The application realizes the accurate, real-time identification of driver distraction state, helps to prevent traffic accidents and improve road safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of distracted driving behavior recognition, and in particular to a method and device for detecting the degree of distraction of an intelligent driver, a medium and a product. BACKGROUND

[0002] With the rapid development of intelligent driving technology, vehicle cabin design is undergoing unprecedented changes. In particular, the widespread use of in-vehicle information systems (IVIS) has greatly enriched the way information is exchanged during driving. These systems provide drivers with a more convenient and rich driving experience through large screens and multi-functional touch interfaces.

[0003] However, while enjoying the convenience brought by intelligent driving technology, the problem of driver distraction is increasingly prominent. Complex visual and manual tasks not only easily interfere with the driver's driving operation, but also significantly increase the risk of driving. Visual and manual distraction has been confirmed by authoritative agencies as an important factor leading to traffic accidents. Therefore, how to effectively identify and classify the degree of driver distraction has become a key to improving driving safety.

[0004] In related technologies, driver distraction recognition systems are mainly based on simple visual or manual task monitoring, and infer the distraction state of the driver by detecting the driver's eye movement behavior, driving operation and vehicle dynamics, etc. These methods mostly use a single classification model, such as Support Vector Machine (SVM), Random Forest (RF), etc. Although they can identify the degree of distraction to some extent, they have problems of insufficient recognition accuracy and poor generalization ability in the face of the complexity and diversity of driving situations. In addition, existing systems often only consider part of the driving behavior factors and fail to fully utilize multi-dimensional data for comprehensive analysis and classification, resulting in limited model performance. More critically, related distraction recognition methods, such as facial feature and eye movement data analysis based on Convolutional Neural Network (CNN), while improving the recognition accuracy to some extent, lack interpretability in the classification process, and cannot clearly reveal the influence of each input feature on the final classification result. This lack of interpretability in the classification method brings safety hazards in practical applications, because it is difficult to accurately understand why the model makes a particular classification decision, making it difficult to classify and adjust potential errors. SUMMARY

[0005] The application aims to provide an intelligent driver distraction degree detection method, device, equipment, medium and product, which can accurately and effectively identify and interpret the visual-manual distraction degree of a driver, thereby improving the safety and reliability of an intelligent driving system.

[0006] To achieve the above object, the application provides the following solutions.

[0007] In a first aspect, the application provides an intelligent driver distraction degree detection method, comprising:

[0008] obtaining an original driving data set; the original driving data set comprises state values of different driving parameters at different historical time points, and the different driving parameters include visual behavior parameters, manual operation parameters and dynamic parameters of a driving vehicle of a driver;

[0009] determining a preset number of driving parameters with high discriminability in the original driving data set as feature parameters by using an SF 2 SOG method, and constructing a training data set; the training data set comprises a plurality of training sample data, and each training sample data is composed of state values of the feature parameters of the same driver in the same historical time period;

[0010] performing unsupervised clustering on the training data set by using a condensed clustering algorithm to obtain a distraction degree label of each training sample data in the training data set; the distraction degree label includes high distraction, medium distraction and low distraction;

[0011] constructing a complete training data set by using the training data set and the distraction degree label corresponding to each training sample data in the training data set;

[0012] training a stacked ensemble model based on the complete training data set to obtain a trained stacked ensemble model; the stacked ensemble model comprises a plurality of different base classifiers and a meta-classifier; each of the plurality of different base classifiers is connected to the meta-classifier; the plurality of different base classifiers include a random forest classifier, an extreme gradient boosting tree classifier and a self-adaptive enhancement classifier, the base classifier is used for distraction degree prediction, and the meta-classifier is used for fusing the distraction degree prediction results output by each base classifier to obtain a fusion prediction result of the distraction degree;

[0013] inputting state values of each feature parameter of a to-be-detected driver in a preset time period into the trained stacked ensemble model to obtain a fusion prediction result of the distraction degree of the to-be-detected driver.

[0014] In a second aspect, the application provides an intelligent driver distraction degree detection device, comprising:

[0015] The data collection module is configured to obtain an original driving data set, wherein the original driving data set comprises state values of different driving parameters at different historical time points, and the different driving parameters comprise a visual behavior parameter of a driver, a manual operation parameter, and a dynamic parameter of a driving vehicle.

[0016] The feature importance evaluation module is configured to use the SF 2 The SOG method determines a preset number of driving parameters with high discriminability in the original driving data set as feature parameters, and constructs a training data set, wherein the training data set comprises a plurality of training sample data, and each training sample data is composed of state values of the feature parameters in the same historical time period of the same driver.

[0017] The unsupervised clustering module is configured to use a condensed clustering algorithm to perform unsupervised clustering on the training data set, and obtain a distraction degree label of each training sample data in the training data set, wherein the distraction degree label comprises high distraction, medium distraction, and low distraction.

[0018] The training data integration module is configured to integrate the training data set and the distraction degree label corresponding to each training sample data in the training data set to form a complete training data set.

[0019] The model training module is configured to train a stacked ensemble model based on the complete training data set, and obtain a trained stacked ensemble model, wherein the stacked ensemble model comprises a plurality of different base classifiers and a meta-classifier, the plurality of different base classifiers are connected to the meta-classifier, the plurality of different base classifiers comprise a random forest classifier, an extreme gradient boosting tree classifier, and a self-adaptive enhancement classifier, the base classifiers are used for distraction degree prediction, and the meta-classifier is used for fusing the distraction degree prediction results output by the base classifiers to obtain a fusion prediction result of the distraction degree.

[0020] The distraction detection module is configured to input state values of the feature parameters of a to-be-detected driver in a preset time period into the trained stacked ensemble model, and obtain a fusion prediction result of the distraction degree of the to-be-detected driver.

[0021] In a third aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the intelligent driver distraction degree detection method according to any one of the above.

[0022] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the intelligent driver distraction degree detection method according to any one of the above.

[0023] In a fifth aspect, the present application provides a computer program product comprising a computer program which, when executed by a processor, implements the steps of the intelligent driver distraction level detection method of any one of the above.

[0024] According to the specific embodiments provided in the present application, the present application has the following technical effects:

[0025] The present application provides an intelligent driver distraction level detection method, device, equipment, medium and product, by acquiring an original driving data set containing driver visual behavior parameters, manual operation parameters and driving vehicle dynamic parameters, and using the SF 2 The SOG method constructs a training data set by screening out high-discrimination feature parameters, solves the problem of inaccurate driving parameter selection in traditional methods, which affects the accuracy of distraction level detection, and realizes accurate capture and use of key driving parameters. Further, the present application uses a condensed clustering algorithm to perform unsupervised clustering on the training data set, and labels the distraction level of the training sample data, including high distraction, medium distraction and low distraction. This step solves the problem of high cost and low efficiency of manual labeling of distraction level labels, and realizes automatic and efficient classification of driver distraction states. Based on the complete training data set, the present application trains a stacked ensemble model comprising multiple base classifiers and a meta-classifier. Among them, the base classifiers such as random forest classifier, extreme gradient boosting tree classifier and adaptive boosting classifier are used for preliminary prediction of the distraction level, and the meta-classifier is used to fuse the prediction results of each base classifier to obtain more accurate and stable distraction level fusion prediction results. This design solves the problem of insufficient generalization ability and limited prediction accuracy of a single classifier model, and realizes accurate and real-time identification of the distraction level of the driver. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0027] Figure 1 A flowchart of an intelligent driver distraction level detection method provided by an embodiment of the present application is shown in the figure;

[0028] Figure 2 A flowchart of an intelligent driver distraction level detection method provided by another embodiment of the present application is shown in the figure;

[0029] Figure 3A flowchart of a feature parameter selection process using an L2,0 norm sparsity learning method is provided for an embodiment of the present application.

[0030] Figure 4 A flowchart of a stacked ensemble model training step is provided for an embodiment of the present application.

[0031] Figure 5 A functional module diagram of an intelligent driver distraction level detection device is provided for an embodiment of the present application.

[0032] Figure 6 A structural diagram of a computer device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0034] To make the above objectives, features and advantages of the present application more apparent, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0035] In an exemplary embodiment, as shown in Figure 1 and Figure 2 , an intelligent driver distraction level detection method is provided, which is executed by a computer device, specifically, can be executed by a terminal or a server, or can be executed by a terminal and a server together. In the embodiments of the present application, the method includes the following steps 101 to 106. Wherein:

[0036] Step 101, obtaining an original driving data set; the original driving data set includes state values of different driving parameters at different historical time points, and the different driving parameters include visual behavior parameters, manual operation parameters and dynamic parameters of a driving vehicle of a driver.

[0037] Step 102, determining a preset number of driving parameters with high discriminability in the original driving data set as feature parameters using an SF 2 SOG method, and constructing a training data set; the training data set includes a plurality of training sample data, and the training sample data is composed of state values of each feature parameter of the same driver in the same historical time period.

[0038] Step 103, unsupervised clustering is performed on the training data set by using a condensed clustering algorithm to obtain the distraction degree label of each training sample data in the training data set; the distraction degree label includes high distraction, medium distraction and low distraction.

[0039] Step 104, the training data set and the distraction degree label corresponding to each training sample data in the training data set constitute a complete training data set.

[0040] Step 105, based on the complete training data set, a stacked ensemble model is trained to obtain a trained stacked ensemble model; the stacked ensemble model includes a plurality of different base classifiers and a meta-classifier; the plurality of different base classifiers are connected with the meta-classifier; the plurality of different base classifiers include a random forest classifier, an extreme gradient boosting tree classifier and an adaptive boosting classifier, the base classifier is used for distraction degree prediction, and the meta-classifier is used for fusing the distraction degree prediction results output by each base classifier to obtain a fusion prediction result of the distraction degree.

[0041] Step 106, inputting the state values of each feature parameter of the to-be-detected driver in a preset time period into the trained stacked ensemble model to obtain a fusion prediction result of the distraction degree of the to-be-detected driver.

[0042] By implementing the above steps 101 to 106, the application integrates an efficient feature selection method, a clustering algorithm and a stacked ensemble learning model, which can improve the accuracy of driver distraction degree recognition and is suitable for complex environments of intelligent driving.

[0043] In another exemplary embodiment of the application, in order to collect multi-dimensional driving parameter data of the driver, step 101 specifically includes:

[0044] Through a driving simulation experiment, driving parameters of 35 participants are collected by using a driving simulator and a Tobii eye tracking device to constitute an original driving data set. The driving parameters include:

[0045] Visual behavior parameters of the driver: such as average pupil diameter (APD), total duration of glances (TDG), and glance hit rate (GH).

[0046] Manual operation parameters of the driver: such as the number of manual operations, operation time, hit rate, etc.

[0047] Dynamic parameters of the driving vehicle: such as steering wheel operation angle, vehicle speed, vehicle lateral position, etc.

[0048] To ensure the accuracy and integrity of the data, the original driving data set is preprocessed. The preprocessing steps include:

[0049] Fill in missing values: Use reasonable interpolation methods or default values to fill in missing parts in the data set to avoid the impact of data loss on subsequent analysis.

[0050] Delete blank values: completely remove meaningless blank records in the data set to ensure data purity and availability.

[0051] Text feature digitization: use the LabelEncoder tool in the sklearn library to digitize the text features in the data set (such as driver number, vehicle type, etc.) to facilitate subsequent data processing and analysis.

[0052] After the above preprocessing steps, a clear structure, complete data and easy-to-analyze driving data set is obtained, laying a solid foundation for subsequent research and analysis.

[0053] In another exemplary embodiment of the present application, in order to reduce data redundancy and noise and improve model training efficiency, step 102 specifically includes:

[0054] Because the collected driving parameter dimension is high, direct use may cause model overfitting, so the present application adopts SF 2 SOG (Sparse and Flexible Projection for Unsupervised Feature Selection, Sparse and Flexible Projection for Unsupervised Feature Selection) method for feature parameter selection.

[0055] SF 2 SOG is an unsupervised embedded feature selection method that can select the most discriminative features by adaptive graph and sparsity constraints while preserving the data manifold structure. Its formula is as follows:

[0056]

[0057] Where F represents a manifold embedding matrix composed of the state values of the selected preset number of driving parameters, each column represents a driving parameter, and each row represents the state values of each driving parameter of the same driver at the same historical moment. W represents a dimension reduction transformation matrix for mapping the manifold embedding matrix to a low-dimensional space. G represents an adaptive similarity matrix composed of the similarity between each driving parameter in F. Tr() represents the sum of the elements on the diagonal of the matrix. L GLaplace matrix of G, X represents an original state matrix constructed based on an original driving dataset, S represents an initial similarity matrix composed of similarities between each driving parameter in X, λ and γ represent a first hyperparameter and a second hyperparameter respectively, represents calculating F-norm and squaring, W T W = I m represents orthogonality constraint of W, I m represents m-dimensional unit matrix, ||W|| 2,0 represents sparsity constraint of W, h represents the number of selected features; g i is the i-th row vector of matrix G, represents the similarity of the i-th driving parameter with all driving parameters, J represents an all-1 matrix, g ij is the i-th row j-th column element in G, represents the similarity of the i-th driving parameter with the j-th driving parameter, T represents a transpose matrix.

[0058] As Figure 3 shown, the selection process of feature parameters is given, in which driving parameters from the original driving dataset are directly selected as feature parameters, 6 different types of driving parameters are taken as examples, each row represents a type of driving parameter, and numbers 1-6 represent 6 types of driving parameters, each column represents a training sample data, and the darkness of the circle color represents the importance of the driving parameter for the sample, and numbers 1, 2, 4 and 6 represent the numbers of selected feature parameters. After L2,0 norm is applied for sparsity calculation, a certain type of feature will be judged as unimportant, and thus will not be selected.

[0059] In the application of the present application, SF 2 The SOG method is used to select 15 most discriminative features from the original driving dataset. These features include pupil diameter, gaze duration, vehicle lateral position change, etc. The selected feature parameters provide more effective input for subsequent clustering and classification.

[0060] In another exemplary embodiment of the present application, step 103 specifically comprises:

[0061] An unsupervised clustering is performed on the training dataset by using a condensed clustering algorithm. The condensed clustering is a bottom-up hierarchical clustering algorithm, which can merge the most similar training sample data into a cluster by calculating the distance between the training sample data, until a predetermined number of clusters is reached.

[0062] In the present application, the distraction behaviors of the drivers are divided into three categories: high distraction, medium distraction and low distraction by using the condensed clustering algorithm. The clustering process is as follows:

[0063] Initialization: each training sample data in the training dataset is taken as a separate cluster.

[0064] Compute distance: Compute the distance between all clusters, using Euclidean distance as the distance metric.

[0065] Merge clusters: Merge the two closest clusters at a time until a predetermined number of clusters is reached (three clusters), resulting in the clustering result.

[0066] Generate labels: According to the clustering result, label each training sample data in the training data set with the corresponding degree of distraction label (high, medium, low). These clustering labels are used for subsequent training of the stacked ensemble model.

[0067] In another exemplary embodiment of the present application, step 105 specifically comprises:

[0068] A heuristic stacking ensemble model is used, which is a common ensemble learning method that combines multiple layers of models to improve classification results. In the present application, the heuristic stacking ensemble model includes the following parts:

[0069] Basic classifiers: Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Adaptive Boosting (Adaboost). Each basic classifier performs individual classification on the data.

[0070] Meta-classifier: Logistic Regression (LR), used to integrate the prediction results of the basic classifiers to generate a fused prediction result of the degree of distraction.

[0071] The basic classifiers are trained through 5-fold cross-validation to prevent overfitting and improve the generalization ability of the model. To further optimize the performance of the model, the Particle Swarm Optimization (PSO) algorithm is introduced to automatically adjust the hyperparameters of the basic classifiers.

[0072] The core idea of PSO is to search for the optimal solution by simulating the foraging behavior of bird flocks. It updates the position of particles by moving them in the search space to find the parameter combination that maximizes the objective function (classification accuracy).

[0073] In the present application, PSO is used to optimize the number of trees, maximum depth, and other hyperparameters of Random Forest and XGBoost. Through iterative search, the best parameter settings are obtained to ultimately improve the prediction accuracy of the heuristic stacking ensemble model.

[0074] Among them, implement this implementation, such as Figure 4The specific training process of the stacked model is shown as follows:

[0075] The training data set is divided into multiple subsets.

[0076] For each subset, a different base classifier is used for training and prediction, and a prediction result of the degree of distraction of each base classifier is obtained.

[0077] These prediction results are combined as new features into a new training data set.

[0078] A meta-classifier is trained using this new training data set.

[0079] Finally, the trained meta-classifier is used to make a fusion prediction of the degree of distraction. The input of the stacked ensemble model is SF 2 The SOG algorithm selects 15 most discriminative feature parameters, and the output is the result of the stacked ensemble model after predicting the degree of distraction. The label is the label of the degree of distraction generated by cluster analysis.

[0080] After the stacked ensemble model is trained, the degree of distraction of the driver is predicted, and the model is evaluated by accuracy (Accuracy, ACC), precision, recall and F1 score. The experiment shows that the stacked ensemble model of the present application has high accuracy in different difficulty distraction tasks, especially in high distraction scenarios, the classification accuracy is significantly better than traditional single model, and finally reaches 96.25% accuracy.

[0081] In another exemplary embodiment of the present application, after step 105, it further includes:

[0082] Based on the complete training data set, the SHAP value of each feature parameter is calculated; wherein the SHAP value represents the influence degree of the feature parameter on the fusion prediction result.

[0083] The present application adopts SHAP (SHapley Additive exPlanations) analysis. SHAP is based on cooperative game theory, and calculates the marginal contribution of features to explain the influence of each feature on the classification result. The specific analysis steps are as follows:

[0084] 1. For each input sample, the SHAP value of each feature parameter is calculated, which represents the influence degree of the feature parameter on the fusion prediction result.

[0085] The SHAP value of each feature The calculation formula is as follows:

[0086]

[0087] where, represents the SHAP value of the i'th feature parameter, {x M} represents a set of state values of M feature parameters, M = 1, 2, 3,..., N; M≠i', N represents the total number of feature parameters, P represents any feature subset of {x M} and x i′ represents the state value of the i'th feature parameter, f() represents the stacked ensemble model, P∪{x i′} represents adding the feature value x i′ of the i'th feature to the feature subset P.

[0088] 2. Visualize the SHAP value to show the contribution of each feature to different prediction results.

[0089] 3. Identify the most important feature parameters, such as the pupil diameter (APD) and the fluctuation of the vehicle lateral position, which have a significant impact on the prediction results of the degree of distraction through global and local analysis.

[0090] SHAP uses an additive feature attribution method to explain the prediction target, in which input feature parameters are combined to illustrate the contribution of each input feature parameter to the prediction of the stacked ensemble model. For a given sample x with feature vector i′ , represents the state value of the k'th feature parameter in the i'th sample, where k = 1, 2,..., n, the prediction of the model for this sample is represented as y i′ , and the average prediction of all samples is represented as The formula is as follows:

[0091]

[0092] The formula expresses the meaning of accumulating the effects of each feature on the prediction result, thereby generating a prediction value. For the prediction value y i′ of a given sample, it can be decomposed into the sum of the SHAP values of all feature parameters plus the average prediction value of all samples, which represents the contribution of the k'th feature parameter in the i'th sample to the prediction result (i.e., the SHAP value), by calculating the SHAP value of each feature parameter to explain how y i′ is contributed by each feature. If is greater than 0, it means that the feature has improved the prediction value and has a positive effect; if is less than 0, it means that the feature has reduced the prediction value and has a negative effect.

[0093] The application also provides an application scenario of the intelligent driver distraction degree detection method. Specifically, the intelligent driver distraction degree detection method provided in the embodiment can be applied in a driver state monitoring scenario of an intelligent driving assistance system. The driver state monitoring scenario of the intelligent driving assistance system includes a data collection link, a data analysis and processing link, and a state feedback and intervention link. The driver driving data enters the data analysis and processing link from the data collection link, and the intelligent driver distraction degree detection method provided in the embodiment belongs to the data analysis and processing link, in which a prediction result of the distraction degree of the driver is obtained. According to the prediction result, a warning is sent to the driver in various ways such as sound, vision, or touch in the state feedback and intervention link, reminding the driver to pay attention to driving safety. If the distraction degree of the driver is continuously high, the system can further take intervention measures such as automatically adjusting the vehicle speed, lane keeping assistance, and the like to ensure driving safety.

[0094] Based on the same inventive concept, the embodiment of the application also provides an intelligent driver distraction degree detection device for implementing the intelligent driver distraction degree detection method described above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more intelligent driver distraction degree detection device embodiments provided below can refer to the limitations of the intelligent driver distraction degree detection method described above, which will not be repeated here.

[0095] In one exemplary embodiment, as shown in Figure 5 An intelligent driver distraction degree detection device is provided, including:

[0096] The data collection module 201 is configured to obtain an original driving data set. The original driving data set includes state values of different driving parameters at different historical time points, and the different driving parameters include visual behavior parameters, manual operation parameters, and dynamic parameters of a driving vehicle of the driver.

[0097] The feature importance evaluation module 202 is configured to determine a preset number of driving parameters with high discriminability in the original driving data set as feature parameters by using the SF 2 SOG method, and construct a training data set. The training data set includes a plurality of training sample data, and each training sample data is composed of state values of the feature parameters of the same driver in the same historical time period.

[0098] The unsupervised clustering module 203 is configured to perform unsupervised clustering on the training data set by using a condensation clustering algorithm, and obtain a distraction degree label of each training sample data in the training data set. The distraction degree label includes high distraction, medium distraction, and low distraction.

[0099] The training data integration module 204 is configured to integrate the training data set and the label of the distraction degree corresponding to each training sample data in the training data set to form a complete training data set.

[0100] The model training module 205 is configured to train a stacked ensemble model based on the complete training data set to obtain a trained stacked ensemble model; the stacked ensemble model comprises a plurality of different base classifiers and a meta-classifier; the plurality of different base classifiers are connected to the meta-classifier; the plurality of different base classifiers comprise a random forest classifier, an extreme gradient boosting tree classifier and an adaptive boosting classifier, the base classifiers are used for distraction degree prediction, and the meta-classifier is used for fusing the prediction results of the distraction degree output by each base classifier to obtain a fused prediction result of the distraction degree.

[0101] The distraction detection module 206 is configured to input the state values of each feature parameter of the driver to be detected within a preset time period into the trained stacked ensemble model to obtain a fused prediction result of the distraction degree of the driver to be detected.

[0102] As an optional implementation, the intelligent driver distraction degree detection device provided in the application further comprises an interpretive analysis module 207.

[0103] The interpretive analysis module 207 is configured to calculate a SHAP value of each feature parameter based on the complete training data set; wherein the SHAP value represents the influence degree of the feature parameter on the fused prediction result.

[0104] The SHAP (Shapley additive explanation model) theory is adopted to interpret the prediction result of the distraction degree, to quantify the influence of each input feature parameter on the fused prediction result of the distraction degree, and to provide global and local interpretive analysis.

[0105] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and an internal structure diagram of the computer device can be as shown in Figure 6As shown in the figure. The computer device includes a processor, a memory, an input / output interface (I / O for short) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the data of the driver distraction degree detection. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with the terminal outside through the network connection. The computer program is executed by the processor to realize an intelligent driver distraction degree detection method.

[0106] Those skilled in the art can understand that, Figure 6 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than Figure 6 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than

[0107] In an exemplary embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by the processor to realize the steps in each of the above method embodiments.

[0108] In an exemplary embodiment, a computer program product is provided, which includes a computer program. The computer program is executed by the processor to realize the steps in each of the above method embodiments.

[0109] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0110] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, databases or other media used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0111] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0112] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0113] The principles and implementation modes of the present application are described by applying specific examples herein, and the above-mentioned embodiments are only used to help understand the method and its core idea of the present application; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range can be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A method for detecting the degree of driver distraction, characterized by, The intelligent driver distraction degree detection method comprises: obtaining an original driving data set; the original driving data set comprises state values of different driving parameters at different historical time points, and the different driving parameters comprise visual behavior parameters, manual operation parameters and dynamic parameters of a driving vehicle of a driver; SF 2 The SOG method determines a preset number of driving parameters with high discriminability in the original driving data set as feature parameters, and constructs a training data set; the training data set includes a plurality of training sample data, and the training sample data is composed of state values of each feature parameter in the same historical period of the same driver; the calculation formula of the discriminability of the driving parameters in the original driving data set is: ； wherein F represents a manifold embedding matrix composed of state values of a preset number of driving parameters selected, each column represents a driving parameter, and each row represents state values of each driving parameter of the same driver at the same historical time, W represents a dimension reduction transformation matrix for mapping the manifold embedding matrix to a low-dimensional space, G represents an adaptive similarity matrix composed of similarities between each driving parameter in F, Tr() represents a sum of elements on a diagonal line of a matrix, L G represents a Laplacian matrix of G, X represents an original state matrix constructed based on an original driving data set, S represents an initial similarity matrix composed of similarities between each driving parameter in X, and respectively represent a first hyperparameter and a second hyperparameter, represents calculating a norm of F and squaring, represents an orthogonality constraint of W, I m represents an m-dimensional unit matrix, represents a sparsity constraint of W, and h represents a selected number of features; is an i-th row vector of the matrix G, represents a similarity of the i-th driving parameter with all driving parameters, and J represents an all-1 matrix, is an i-th row and j-th column element in G, represents a similarity of the i-th driving parameter with the j-th driving parameter, T represents a transposed matrix; using a condensed clustering algorithm to perform unsupervised clustering on the training data set to obtain a distraction degree label of each training sample data in the training data set; the distraction degree label comprises high distraction, medium distraction and low distraction; constructing a complete training data set by using the training data set and the distraction degree label corresponding to each training sample data in the training data set; training a stacked ensemble model based on the complete training data set to obtain a trained stacked ensemble model; the stacked ensemble model comprises a plurality of different base classifiers and a meta-classifier; the plurality of different base classifiers are connected with the meta-classifier; the plurality of different base classifiers comprise a random forest classifier, an extreme gradient boosting tree classifier and an adaptive boosting classifier, the base classifier is used for distraction degree prediction, and the meta-classifier is used for fusing the distraction degree prediction results output by each base classifier to obtain a fusion prediction result of the distraction degree; inputting state values of each feature parameter of a driver to be detected within a preset time period into the trained stacked ensemble model to obtain a fusion prediction result of the distraction degree of the driver to be detected; obtaining the trained stacked ensemble model, and then further comprising: calculating a SHAP value of each feature parameter based on the complete training data set; wherein the SHAP value represents the influence degree of the feature parameter on the fusion prediction result; and the calculation formula of the SHAP value of the feature parameter is: ； wherein, SHAP value of the i-th feature parameter, set of state values of M feature parameters, N represents the total number of feature parameters, P represents any feature subset of P, state value of the i-th feature parameter, add the feature value of the i-th feature​​​​ 2. The intelligent driver distraction level detection method of claim 1, wherein using a condensed clustering algorithm to perform unsupervised clustering on the training data set to obtain a distraction degree label of each training sample data in the training data set, specifically comprising: each training sample data in the training data set is taken as a separate cluster; calculating the distance between all clusters, and using Euclidean distance as the distance measurement; merging two nearest clusters each time until a preset cluster number is reached to obtain a clustering result; labeling each training sample data in the training data set with a corresponding distraction degree label according to the clustering result.

3. The intelligent driver distraction level detection method of claim 1, wherein The base classifier is trained by a 5-fold cross-validation method, and the hyperparameters are optimized by a particle swarm optimization algorithm.

4. An intelligent driver drowsiness detection device, characterized by comprising: The intelligent driver distraction degree detection device applies the intelligent driver distraction degree detection method in any one of claims 1-3, and the intelligent driver distraction degree detection device comprises: a data acquisition module configured to obtain an original driving data set; the original driving data set comprises state values of different driving parameters at different historical time points, and the different driving parameters comprise visual behavior parameters, manual operation parameters and dynamic parameters of a driving vehicle of a driver; The feature importance evaluation module is configured to employ the SF 2 The SOG method determines a preset number of driving parameters with high discriminability in the original driving data set as feature parameters, and constructs a training data set; the training data set includes a plurality of training sample data, and each training sample data is composed of state values of each feature parameter in the same historical period of the same driver. an unsupervised clustering module configured to use a condensed clustering algorithm to perform unsupervised clustering on the training data set to obtain a distraction degree label of each training sample data in the training data set; the distraction degree label comprises high distraction, medium distraction and low distraction; The training data integration module is configured to integrate the training data set and the label of the distraction degree corresponding to each training sample data in the training data set to form a complete training data set. The model training module is configured to train a stacked ensemble model based on the complete training data set to obtain a trained stacked ensemble model; the stacked ensemble model comprises a plurality of different base classifiers and a meta-classifier; the plurality of different base classifiers are connected to the meta-classifier; the plurality of different base classifiers comprise a random forest classifier, an extreme gradient boosting tree classifier and an adaptive boosting classifier, the base classifiers are used for distraction degree prediction, and the meta-classifier is used for fusing the prediction results of the distraction degree output by each base classifier to obtain a fused prediction result of the distraction degree. The distraction detection module is configured to input the state values of the feature parameters of the driver to be detected within a preset time period into the trained stacked ensemble model to obtain a fused prediction result of the distraction degree of the driver to be detected.

5. A computer device comprising: A memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the intelligent driver distraction degree detection method of any one of claims 1-3.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the intelligent driver distraction degree detection method of any one of claims 1-3.

7. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the intelligent driver distraction degree detection method of any one of claims 1-3.

Images