Distraction driving behavior identification method and system based on multi-source data combination

Abstract

The invention discloses a multi-source data joint distracted driving behavior identification method and system, and belongs to the technical field of cross-domain data joint analysis, and the method comprises the steps: enabling a vehicle GPS database to be connected to a traffic database, screening the GPS data of a vehicle, and obtaining a matched data set; performing data preprocessing on the matched data set to generate a GPS data set, and performing data alignment to form a joint feature data set; based on the joint feature data set, dividing a time sequence sample with a fixed time window, extracting statistical features, constructing a multi-classification prediction model, and outputting a distraction driving behavior recognition result; and uploading the distraction driving behavior identification result to a vehicle supervision platform. According to the invention, through multi-source data conjoint analysis, a three-level association mechanism and multi-dimensional feature fusion, and through efficient integration of cross-domain data, concealment and reliability are considered, the identification precision and efficiency are high, and an all-weather and low-cost scheme is provided for safety supervision of freight vehicles.

Description

Technical Field

[0001] The present invention belongs to the technical field of cross-domain data joint analysis, and specifically relates to a method and system for identifying distracted driving behavior by combining multi-source data. Background Art

[0002] Road transport accounts for a significant portion of both freight volume and passenger flow in the transportation industry, providing crucial support for economic and social development. However, trucks account for 32.04% of traffic accidents, far exceeding their fleet population. This necessitates further safety oversight of freight vehicles. However, while existing key vehicle monitoring platforms can effectively monitor fatigue and speeding, they remain limited in detecting distracted driving. Mainstream distracted driving detection methods based on camera capture and in-vehicle equipment are susceptible to environmental interference, are highly invasive, and can be easily circumvented. Therefore, accurately identifying distracted driving in freight vehicles by combining existing data without requiring additional equipment has become a pressing issue. Currently, existing technologies utilize GPS systems and monitoring platforms to identify distracted driving in freight vehicles. However, these methods currently lack cross-domain data analysis, such as mobile phone network data and vehicle GPS data, making non-invasive identification of distracted driving behavior difficult. This may require additional equipment, potentially disrupting the driver and prone to circumvention. Furthermore, the accuracy and efficiency of these methods require further improvement. Summary of the Invention

[0003] The present invention provides a method and system for identifying distracted driving behavior by combining multiple data sources. The aim is to address the current problems in identifying distracted driving in freight vehicles, such as the inability to conduct cross-domain joint analysis of data such as mobile phone mobile network data and vehicle GPS data, the difficulty in performing non-invasive distracted driving behavior identification, the need to add equipment, which may interfere with the driver, and the ease with which distracted driving behavior can be circumvented. Furthermore, the accuracy and efficiency of identification need to be further improved.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions: The present invention provides a method for identifying distracted driving behavior by combining multi-source data, comprising the following steps: S1. Connect the vehicle GPS database to the traffic database, use the vehicle license plate number, longitude and latitude as matching identifiers, filter the vehicle GPS data, and associate it with the owner's mobile phone number information to obtain a matching data set; S2. Preprocess the matching dataset, remove redundant features, and retain the owner's mobile phone number information, acquisition timestamp, and vehicle speed information to generate a GPS dataset. Align the data with the mobile phone communication database using timestamp and mobile phone number information to form a joint feature dataset. S3. Based on the joint feature dataset, the time series samples are divided into fixed time windows, and the statistical features of vehicle speed, traffic data packet length, and call behavior are extracted to build a multi-classification prediction model and output the distracted driving behavior identification results; S4. Upload the distracted driving behavior identification results to the vehicle supervision platform.

[0005] In some embodiments, in S1, screening the vehicle's GPS data specifically includes: Based on the key vehicle information dataset and key road section dataset in the traffic database, a combined query of license plate number, vehicle longitude and latitude is used to determine whether the vehicle is a key vehicle in the key road section. If it is, the vehicle's GPS data is retained and associated with the owner's mobile phone number; otherwise, it is removed.

[0006] In some embodiments, in S1, the association operation of the matching identifier is implemented through a three-level data association mechanism, specifically including: binding vehicle identity information based on the license plate number, determining the real-time geographic location based on the vehicle's longitude and latitude, and aligning the vehicle's operating status and communication behavior data based on timestamps.

[0007] In some embodiments, in S2, data preprocessing includes: checking missing values ​​on the matching data set and eliminating records containing missing values; checking outliers and eliminating abnormal speed data based on a reasonable threshold range of vehicle speed; retaining the owner's mobile phone number, collection timestamp, and vehicle speed information to generate a GPS data set.

[0008] In some implementations, in S2, the data alignment specifically includes: associating the vehicle owner's mobile phone number in the GPS data set with the communication data mobile phone number in the mobile phone communication database; based on the matching of the acquisition timestamp and the communication data timestamp, achieving time alignment of the vehicle operation status data with the call records and traffic data packet long data.

[0009] In some embodiments, in S3, the extraction of statistical features includes: extracting the maximum value, minimum value, average value, variance and preset quantiles for the time series data of vehicle speed; extracting the maximum value, minimum value, average value, variance and preset quantiles for the time series data of traffic data packet length; and for call behavior data, counting the number of calls and total call duration within the time window.

[0010] In some embodiments, in S3, constructing the multi-classification prediction model includes: combining vehicle speed statistical features, traffic data packet length statistical features, and call behavior features into a multi-dimensional classification feature vector; An ensemble learning algorithm is used to train a multi-classification model. The ensemble learning algorithm includes the XGBoost or LSTM algorithm. The driving status categories are output, including normal driving, making a phone call, and playing with the phone.

[0011] Furthermore, in S3, the multi-classification prediction model supports real-time warning function. When the driving status category output by the multi-classification prediction model is making a phone call or playing with a mobile phone, a distracted driving warning signal is generated; the warning signal and the recognition result are uploaded to the vehicle supervision platform simultaneously.

[0012] In some implementations, in S3, the time window is divided using a sliding time window mechanism: The preset time length is used as the window unit, and the vehicle operation time period is divided into sliding segments according to a fixed step size; each time window corresponds to an independent driving state classification result.

[0013] The present invention also provides a distracted driving behavior identification system based on multi-source data. The system includes a data acquisition module, a multi-source data joint analysis module, and a distracted driving behavior identification module, wherein: Data collection module: used to connect the vehicle GPS database to the traffic database, filter the vehicle GPS data based on the vehicle license plate number, longitude and latitude as matching identifiers, and associate it with the owner's mobile phone number information to obtain a matching data set; Multi-source data joint analysis module: This module is used to preprocess the matching data set, remove redundant features, retain the owner's mobile phone number information, acquisition timestamp, and vehicle speed information, generate a GPS data set, and align the data with the mobile communication database using timestamp and mobile phone number information to form a joint feature data set. Distracted Driving Behavior Identification Module: Based on the joint feature dataset, it divides the time series samples into fixed time windows, extracts the statistical features of vehicle speed, traffic data packet length, and call behavior, builds a multi-classification prediction model, outputs distracted driving behavior identification results, and uploads the distracted driving behavior identification results to the vehicle supervision platform.

[0014] Compared with the prior art, the present invention provides a method and system for identifying distracted driving behavior by combining multi-source data, which has the following beneficial effects: The present invention proposes a method for identifying distracted driving behavior using multi-source data. By combining multiple sources, such as vehicle GPS data and mobile phone network data, this method achieves non-invasive distracted driving behavior identification. This method eliminates the need for additional equipment, reduces interference with driver behavior, and improves the stealth and accuracy of monitoring. Through efficient data preprocessing and feature engineering, the present invention significantly reduces network bandwidth overhead, enabling real-time monitoring around the clock. Compared to existing technologies, the present invention can identify distracted driving behavior in real time with minimal data collection requirements, ensuring the traffic safety of freight vehicles to a certain extent. By combining multiple sources, such as vehicle GPS data and mobile phone communication data, the present invention constructs a secure and efficient identification model. This method overcomes the limitations of joint data analysis and significantly improves the accuracy and efficiency of distracted driving behavior identification through efficient cross-domain data integration. Through non-invasive, real-time, cross-domain data joint analysis, the present invention significantly improves the efficiency and accuracy of distracted driving behavior identification while reducing the system's computational and bandwidth overhead, demonstrating its high practicality and potential for widespread adoption. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings in the specification are used to provide further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0016] Figure 1 A flowchart of a method for identifying distracted driving behavior by combining multi-source data according to the present invention; Figure 2 Schematic diagram of the flow of a multi-source data cross-domain joint algorithm in a multi-source data joint distracted driving behavior identification method of the present invention; Figure 3 The figure is a flow chart of a distracted driving behavior recognition algorithm in a method for identifying distracted driving behavior using multi-source data in the present invention. DETAILED DESCRIPTION

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0018] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0019] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0020] In the description of the embodiments of the present invention, it should be noted that if the terms "upper," "lower," "horizontal," "inner," etc. appear, the orientation or positional relationship indicated is based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the inventive product is typically placed when in use. These terms are merely for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on the present invention. In addition, the terms "first," "second," etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0021] In addition, if the term "horizontal" appears, it does not mean that the component must be absolutely horizontal, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0022] In the description of the embodiments of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "disposed," "installed," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0023] like Figure 1 As shown, the present invention provides a method for identifying distracted driving behavior by combining multi-source data, comprising the following steps: S1. Connect the vehicle GPS database to the traffic database, use the vehicle license plate number, longitude and latitude as matching identifiers, filter the vehicle GPS data, and associate it with the owner's mobile phone number information to obtain a matching data set; S2. Preprocess the matching dataset, remove redundant features, and retain the owner's mobile phone number information, acquisition timestamp, and vehicle speed information to generate a GPS dataset. Align the data with the mobile phone communication database using timestamp and mobile phone number information to form a joint feature dataset. S3. Based on the joint feature dataset, the time series samples are divided into fixed time windows, and the statistical features of vehicle speed, traffic data packet length, and call behavior are extracted to build a multi-classification prediction model and output the distracted driving behavior identification results; S4. Upload the distracted driving behavior identification results to the vehicle supervision platform.

[0024] The present invention uses a method for identifying distracted driving behaviors by combining multi-source data. First, the GPS database is connected to the traffic database to match the GPS data corresponding to key vehicles in key sections. Secondly, the GPS data is connected to the data processing platform for data preprocessing and feature engineering, and the data processing results are connected to the mobile phone communication database. Then, distracted driving identification is completed on the mobile phone communication database side. Finally, the identification results are returned to the data processing platform and further uploaded to the vehicle supervision platform for user use. In the cross-domain data joint analysis, the network bandwidth overhead is significantly reduced through efficient preprocessing and feature engineering. The present invention fits the current status of existing network resources, minimizes the requirements for data collection, realizes non-invasive, all-weather real-time monitoring of distracted driving behaviors, and ensures the traffic safety of freight vehicles.

[0025] The following is a detailed description of a method and system for identifying distracted driving behavior using multi-source data in accordance with the present invention through specific embodiments.

[0026] like Figure 1 As shown, the present invention provides a method for identifying distracted driving behavior by combining multi-source data, and proposes a new distracted driving behavior identification technology by combining multi-source data. Since the use of mobile phones is the main cause of distracted driving, the present invention constructs a safe and efficient identification model for mobile phone mobile network data and vehicle GPS data, and significantly reduces network bandwidth overhead through efficient preprocessing and feature engineering in the cross-domain data joint analysis. The present invention minimizes the data collection requirements based on the current status of existing network resources, realizes non-invasive, all-weather real-time monitoring of distracted driving behavior, and ensures the traffic safety of freight vehicles. The specific process includes the following steps: Step 1: The GPS database is connected to the traffic database to match the GPS data corresponding to key vehicles in key sections.

[0027] Step 2: Connect the GPS data to the data processing platform for data preprocessing and feature engineering, and connect the data processing results to the mobile phone communication database.

[0028] Step 3: Complete distracted driving identification on the mobile phone communication database side.

[0029] Step 4: The recognition results are now returned to the data processing platform and further uploaded to the vehicle supervision platform for user use.

[0030] like Figure 2 As shown, in some embodiments, in step 1, in [ , ]( ) time range, the vehicle GPS database and mobile phone communication database collected corresponding data samples respectively. Vehicle samples, of which the original GPS data are features, denoted as ; Mobile phone communication raw data includes Call characteristics and data packet length characteristics , recorded as ; The traffic database contains key vehicle information datasets and key road section datasets . GPS dataset Access the traffic database to match vehicle information. is the license plate number, is the longitude of the vehicle, is the latitude of the vehicle, and these three features are used as matching identifiers in and Query to determine whether the sample is a key vehicle in a key section. If it exists, keep the sample and add The car owner’s mobile phone number information is collected, otherwise the sample will be eliminated.

[0031] It should be noted that all actions of acquiring signals, information or data in the present invention are performed in compliance with the corresponding data protection laws and policies of the country where they are located and with the authorization given by the owner of the corresponding device.

[0032] like Figure 2 As shown, in some embodiments, in step 2, the matching successful data set First, we check for missing values ​​and outliers, and then remove redundant features in the GPS data that are not relevant to prediction, retaining only the owner's mobile phone number, acquisition timestamp, and vehicle speed information, thereby reducing the feature dimension. The final GPS dataset is recorded as .Will Access to mobile phone communication database. is the mobile phone number corresponding to the communication data, For the corresponding timestamp, use in and in Perform data alignment to obtain a joint dataset .

[0033] like Figure 3 As shown, in some embodiments, in step 3, the distracted driving recognition model phase is executed. Divide the time series samples of each vehicle within the observation time range, and the vehicle corresponds to one of the categories 0: normal driving, 1: making a phone call, and 2: playing with the phone in each window time. The overall label set is recorded as . Extract key features of the three parts of the data. For time series data such as vehicle speed and traffic data packet length, calculate their maximum (minimum), average, median, variance, quantile, etc. Statistical features, denoted as For call data, the number of calls and total call duration within the time window are counted and recorded as . Merge three types of data, namely ,based on Build a multi-classification prediction model and get the prediction results based on the scores of each class.

[0034] The present invention also provides a distracted driving behavior identification system based on multi-source data. The system includes a data acquisition module, a multi-source data joint analysis module, and a distracted driving behavior identification module, wherein: Data collection module: used to connect the vehicle GPS database to the traffic database, filter the vehicle GPS data based on the vehicle license plate number, longitude and latitude as matching identifiers, and associate the owner's mobile phone number information to obtain a matching data set Multi-source data cross-domain joint module: used to realize multi-source data fusion and association analysis of GPS database, mobile communication database and traffic database, specifically including: screening GPS data of key vehicles traveling on key sections of roads in the traffic database through triple matching identifiers of license plate number, longitude and latitude; establishing association index between GPS data and mobile communication data based on the owner's mobile phone number; using timestamp alignment technology to achieve temporal matching between vehicle operation status and communication behavior, constructing a joint feature data set containing vehicle speed, call records and traffic data, and completing pre-processing operations such as data cleaning, redundant feature elimination and missing value processing.

[0035] Distracted Driving Behavior Recognition Module: This module is used to intelligently identify distracted driving behaviors based on a joint feature dataset. Specifically, it divides vehicle operating time periods through a sliding time window mechanism; extracts statistical features such as maximum value, mean, and variance from time-series data such as vehicle speed and traffic packet length; extracts frequency and duration features from call behavior; fuses multi-dimensional features to construct classification feature vectors, and uses machine learning algorithms to train multi-classification models (normal / making a call / playing with the phone); outputs distracted driving identification results in real time and pushes them to the regulatory platform, while also supporting dynamic model updates and identification threshold configuration functions.

[0036] In the multi-source data cross-domain joint module of the present invention, a three-level data association mechanism (license plate number-geographic location-time stamp) is adopted to achieve accurate matching of heterogeneous data sources. Dynamic geographic fencing technology is used to automatically identify vehicles traveling on key roads, and a spatiotemporal association between GPS trajectory data and communication behavior is established based on mobile phone number binding. At the same time, data cleaning, feature dimensionality reduction and time series alignment functions are integrated to form a standardized joint feature dataset.

[0037] The distracted driving behavior recognition module of this invention utilizes a multi-granularity feature engineering system (time series statistical features + behavior frequency features + traffic packet features). It employs a dynamic segmentation method using a sliding time window and an ensemble learning algorithm to achieve a three-level classification of driving status (0 / 1 / 2). This ensemble learning algorithm can employ either the extreme gradient boosting (XGBoost) algorithm or the long short-term memory (LSTM) algorithm. The distracted driving behavior recognition module provides real-time warnings, confidence assessment, and online model updates. Both modules utilize a distributed message queue for high-concurrency data processing, forming a closed-loop distracted driving monitoring solution.

[0038] In summary, the present invention provides a method and system for identifying distracted driving behavior by combining multi-source data. Through the joint analysis of multi-source data, non-invasive real-time monitoring of distracted driving behavior of freight vehicles is achieved. Based on a three-level association mechanism of license plate number, geographic location and timestamp, the present invention accurately matches the GPS data of key vehicles in key sections with the communication data of the owner's mobile phone. There is no need to install additional in-vehicle equipment, which reduces interference with driving behavior. Redundant features are eliminated through data preprocessing, and statistical features of vehicle speed, traffic packet length and call behavior are extracted in combination with a sliding time window, which significantly reduces network bandwidth overhead and adapts to existing network resources. In addition, the present invention integrates multi-dimensional features to construct a classification model, realizes three-level classification of driving status, and improves recognition accuracy and efficiency. At the same time, dynamic geographic fence technology automatically demarcates the monitoring area, and distributed message queues ensure real-time transmission and processing of high-concurrency data, forming a closed-loop monitoring. The present invention takes into account both concealment and reliability, and provides all-weather, low-cost technical support for the safety supervision of freight vehicles.

[0039] Finally, it should be noted that the above is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any ordinary technician in this industry can smoothly implement the present invention as shown in the specification and described above. Any equivalent changes, modifications and evolutions made by using the technical content disclosed above are all equivalent embodiments of the present invention. At the same time, any equivalent changes, modifications and evolutions made to the above embodiments based on the essential technology of the present invention are still within the scope of protection of the technical solution of the present invention.

Claims

1. A method for identifying distracted driving behavior using multi-source data, characterized in that: The steps include: S1. Connect the vehicle GPS database to the traffic database, use the vehicle license plate number, longitude and latitude as matching identifiers, filter the vehicle GPS data, and associate it with the owner's mobile phone number information to obtain a matching data set; S2. Preprocess the matching dataset, remove redundant features, and retain the owner's mobile phone number information, acquisition timestamp, and vehicle speed information to generate a GPS dataset. Align the data with the mobile phone communication database using timestamp and mobile phone number information to form a joint feature dataset. S3. Based on the joint feature dataset, the time series samples are divided into fixed time windows, and the statistical features of vehicle speed, traffic data packet length, and call behavior are extracted to build a multi-classification prediction model and output the distracted driving behavior identification results; S4. Upload the distracted driving behavior identification results to the vehicle supervision platform.

2. The method for identifying distracted driving behavior using multi-source data according to claim 1, characterized in that: In S1, screening the vehicle's GPS data specifically includes: Based on the key vehicle information dataset and key road section dataset in the traffic database, a combined query of license plate number, vehicle longitude and latitude is used to determine whether the vehicle is a key vehicle in the key road section. If it is, the vehicle's GPS data is retained and associated with the owner's mobile phone number; otherwise, it is removed.

3. The method for identifying distracted driving behavior using multi-source data according to claim 1, characterized in that: In S1, the association operation of the matching identifier is implemented through a three-level data association mechanism, specifically including: binding vehicle identity information based on the license plate number, determining the real-time geographic location based on the vehicle's longitude and latitude, and aligning the vehicle's operating status and communication behavior data based on timestamps.

4. The method for identifying distracted driving behavior using multi-source data according to claim 1, characterized in that: In S2, data preprocessing includes: checking missing values ​​on the matching data set and eliminating records containing missing values; checking outliers and eliminating abnormal speed data based on a reasonable threshold range of vehicle speed; retaining the owner's mobile phone number, collection timestamp and vehicle speed information to generate a GPS data set.

5. The method for identifying distracted driving behavior using multi-source data according to claim 1, characterized in that: In the S2, the data alignment specifically includes: associating the owner's mobile phone number in the GPS data set with the communication data mobile phone number in the mobile phone communication database; based on the matching of the acquisition timestamp and the communication data timestamp, realizing the time alignment of the vehicle operation status data with the call records and traffic data packet long data.

6. The method for identifying distracted driving behavior using multi-source data according to claim 1, characterized in that: In the S3, the extraction of statistical features includes: extracting the maximum value, minimum value, average value, variance and preset quantiles of the time series data of vehicle speed; extracting the maximum value, minimum value, average value, variance and preset quantiles of the time series data of traffic data packet length; and for call behavior data, counting the number of calls and total call duration within the time window.

7. The method for identifying distracted driving behavior using multi-source data according to claim 1, characterized in that: In said S3, the construction of the multi-classification prediction model includes: combining the vehicle speed statistical features, the traffic data packet length statistical features and the call behavior features into a multi-dimensional classification feature vector; An ensemble learning algorithm is used to train a multi-classification model. The ensemble learning algorithm includes the XGBoost or LSTM algorithm. The driving status categories are output, including normal driving, making a phone call, and playing with the phone.

8. The method for identifying distracted driving behavior using multi-source data according to claim 7, characterized in that: In S3, the multi-classification prediction model supports a real-time warning function. When the driving status category output by the multi-classification prediction model is making a phone call or playing with a mobile phone, a distracted driving warning signal is generated; The warning signal and recognition results are uploaded to the vehicle supervision platform simultaneously.

9. The method for identifying distracted driving behavior using multi-source data according to claim 1, characterized in that: In S3, the time window is divided using a sliding time window mechanism: The preset time length is used as the window unit, and the vehicle operation time period is divided into sliding segments according to a fixed step size; each time window corresponds to an independent driving state classification result.

10. The system according to any one of claims 1 to 9, wherein the method for identifying distracted driving behavior by combining multi-source data is based on: The system includes a data acquisition module, a multi-source data joint analysis module, and a distracted driving behavior recognition module, wherein: Data collection module: used to connect the vehicle GPS database to the traffic database, filter the vehicle GPS data based on the vehicle license plate number, longitude and latitude as matching identifiers, and associate it with the owner's mobile phone number information to obtain a matching data set; Multi-source data joint analysis module: This module is used to preprocess the matching data set, remove redundant features, retain the owner's mobile phone number information, acquisition timestamp, and vehicle speed information, generate a GPS data set, and align the data with the mobile communication database using timestamp and mobile phone number information to form a joint feature data set. Distracted Driving Behavior Identification Module: Based on the joint feature dataset, it divides the time series samples into fixed time windows, extracts the statistical features of vehicle speed, traffic data packet length, and call behavior, builds a multi-classification prediction model, outputs distracted driving behavior identification results, and uploads the distracted driving behavior identification results to the vehicle supervision platform.

Images