WO2021115133A1 - 驾驶行为识别方法、装置、电子设备及存储介质 - Google Patents

驾驶行为识别方法、装置、电子设备及存储介质 Download PDF

Info

Publication number
WO2021115133A1
WO2021115133A1 PCT/CN2020/131953 CN2020131953W WO2021115133A1 WO 2021115133 A1 WO2021115133 A1 WO 2021115133A1 CN 2020131953 W CN2020131953 W CN 2020131953W WO 2021115133 A1 WO2021115133 A1 WO 2021115133A1
Authority
WO
WIPO (PCT)
Prior art keywords
vector
node
driving behavior
trajectory
target
Prior art date
Application number
PCT/CN2020/131953
Other languages
English (en)
French (fr)
Inventor
曾思敏
张旭
郑越
许强
Original Assignee
平安科技(深圳)有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2021115133A1 publication Critical patent/WO2021115133A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/50Context or environment of the image
    • G06V20/52Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G06V20/54Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/044Recurrent networks, e.g. Hopfield networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/049Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/50Context or environment of the image
    • G06V20/59Context or environment of the image inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
    • G06V20/597Recognising the driver's state or behaviour, e.g. attention or drowsiness

Definitions

  • This application relates to the field of artificial intelligence technology, and in particular to a driving behavior recognition method, device, electronic equipment, and storage medium.
  • the existing driving behavior recognition method is to identify whether the driver has dangerous driving behaviors by installing a variety of sensors on the vehicle, such as: section speed measurement, etc.
  • the inventor realized that the method of installing sensors on the vehicle or on the road The installation of sensors on the top increases the difficulty of the implementation of the program.
  • the accuracy of determining whether the driver has dangerous driving behavior through the sensor also depends on the performance of the sensor.
  • the first aspect of the present application provides a driving behavior recognition method, the driving behavior recognition method includes:
  • the first vector and the second vector are fused to obtain a target vector, and the target vector is input into a pre-built binary classification model to obtain a driving recognition result.
  • a second aspect of the present application provides an electronic device including a processor and a memory, and the processor is configured to execute computer-readable instructions stored in the memory to implement the following steps:
  • the first vector and the second vector are fused to obtain a target vector, and the target vector is input into a pre-built binary classification model to obtain a driving recognition result.
  • a third aspect of the present application provides a computer-readable storage medium having at least one computer-readable instruction stored thereon, and the at least one computer-readable instruction is executed by a processor to implement the following steps:
  • the first vector and the second vector are fused to obtain a target vector, and the target vector is input into a pre-built binary classification model to obtain a driving recognition result.
  • a fourth aspect of the present application provides a driving behavior recognition device, where the driving behavior recognition includes:
  • the determining unit is configured to receive a driving behavior recognition request, and determine the driver to be tested according to the driving behavior recognition request;
  • a creating unit configured to create a trajectory node network based on the multiple longitudes, the multiple latitudes, and the multiple time points;
  • a conversion unit configured to convert the driving behavior data into a first vector, and convert the trajectory characteristic information into a second vector
  • the input unit is used for fusing the first vector and the second vector to obtain a target vector, and inputting the target vector into a pre-built binary classification model to obtain a driving recognition result.
  • this application determines the driver to be tested according to the driving behavior recognition request, can accurately determine the driver to be tested, and obtain the driving behavior data and navigation data of the driver to be tested, based on the longitude and The latitude and the time point create a trajectory node network.
  • the navigation data By converting the navigation data into a trajectory node network, not only can the navigation data be viewed intuitively, but at the same time, it can also pave the way for the rapid extraction of feature information. Extracting trajectory feature information in the navigation data can reduce the time spent in navigation data analysis, thereby improving recognition efficiency.
  • the driving behavior data is converted into a first vector
  • the trajectory feature information is converted into a second vector
  • the The first vector and the second vector are used to obtain a target vector, so that the target vector has the characteristics of the driving behavior data and the trajectory feature information, so as to improve the comprehensiveness of the target vector, and input the target vector
  • the driving recognition result is obtained, and the accuracy of driving behavior recognition is improved by analyzing the comprehensive target vector.
  • this application does not need to install sensors in the vehicle or install sensors on the road To this end, the program’s practicability has been improved.
  • Fig. 1 is a flowchart of an embodiment of the driving behavior recognition method of the present application.
  • Fig. 1a is a flowchart of an embodiment of the present application for determining the driver to be tested.
  • Fig. 1b is a flowchart of an embodiment of creating a trajectory node network in the present application.
  • Fig. 1c is a flowchart of an embodiment of extracting trajectory feature information according to the present application.
  • Fig. 1d is a flowchart of an embodiment of generating a target vector in the present application.
  • Fig. 1e is a flowchart of another embodiment of the driving behavior recognition method of the present application.
  • Figure 1f is a flowchart of an embodiment of generating multiple random walk sequences in the present application.
  • Fig. 2 is a functional block diagram of an embodiment of the driving behavior recognition device of the present application.
  • FIG. 3 is a schematic structural diagram of an electronic device implementing an embodiment of a driving behavior recognition method according to the present application.
  • FIG. 1 it is a flowchart of an embodiment of the driving behavior recognition method of the present application. According to different needs, the order of the steps in the flowchart can be changed, and some steps can be omitted.
  • the driving behavior recognition method can be applied to smart traffic scenarios, so as to promote the construction of smart cities.
  • the driving behavior recognition method is applied to one or more electronic devices.
  • the electronic device is a device that can automatically perform numerical calculation and/or information processing in accordance with pre-set or stored readable instructions, and its hardware includes But it is not limited to microprocessors, application specific integrated circuits (ASICs), programmable gate arrays (Field-Programmable Gate Arrays, FPGAs), digital processors (Digital Signal Processors, DSPs), embedded devices, etc.
  • the electronic device may be any electronic product that can perform human-computer interaction with the user, such as a personal computer, a tablet computer, a smart phone, a personal digital assistant (PDA), a game console, an interactive network television ( Internet Protocol Television, IPTV), smart wearable devices, etc.
  • a personal computer a tablet computer
  • a smart phone a personal digital assistant (PDA)
  • PDA personal digital assistant
  • IPTV interactive network television
  • smart wearable devices etc.
  • the electronic device may also include a network device and/or user equipment.
  • the network device includes, but is not limited to, a single network server, a server group composed of multiple network servers, or a cloud composed of a large number of hosts or network servers based on cloud computing.
  • the network where the electronic device is located includes, but is not limited to: the Internet, a wide area network, a metropolitan area network, a local area network, a virtual private network (Virtual Private Network, VPN), etc.
  • VPN Virtual Private Network
  • a driving behavior recognition request is received, and the driver to be tested is determined according to the driving behavior recognition request.
  • the driving behavior recognition request can be triggered by an employee of the insurance company, or can be triggered automatically within a preset time.
  • the preset time may be a time point or a time period.
  • the data information carried in the driving behavior identification request includes, but is not limited to: a preset tag, an identification code, a request code, and the like.
  • the preset label may be name; the identification code may be 23458abd; and the request code may be 001.
  • Fig. 1a is a flowchart of an embodiment of the present application for determining a driver to be tested.
  • the electronic device determining the driver to be tested according to the driving behavior recognition request includes:
  • S101 Analyze the driving behavior recognition request by using the any idle thread to obtain data information carried in the driving behavior recognition request;
  • S102 Obtain a preset label from a label table, where the preset label refers to a pre-defined label
  • S104 Determine the driver to be tested according to the identification code.
  • the preset label is a pre-defined identifier, for example, the preset label may be the ID of the driver to be tested.
  • the identification code may be an ID card.
  • the data information includes: "ID, 2568871515; name, Li Sheng”.
  • the preset tag is acquired as ID
  • the electronic device acquires the information corresponding to the ID from "ID, 2568871515; name, Li Sheng” as 2568871515, and determines 2568871515 as the identification code.
  • the driving behavior recognition request is parsed through an idle thread. Since there is no need to wait for the thread to process other requests, the parsing speed of the driving behavior recognition request is increased, and the identification code can be accurately determined through the preset label, and then the identification code can be accurately determined. Describe the driver to be tested.
  • the electronic device determining the driver to be tested according to the identification code includes:
  • the electronic device obtains the user corresponding to the identification code from the public security library as the driver to be tested.
  • the driver to be tested can be accurately determined.
  • the driving behavior data includes: driving speed, whether the driver's terminal device is in a communication state when the vehicle is running, and the like.
  • each time point refers to the time that the vehicle driven by the driver to be tested stays at the corresponding longitude and latitude.
  • the electronic device acquiring the driving behavior data and navigation data of the driver to be tested includes one or a combination of the following methods:
  • the driving behavior data of the driver to be tested and the navigation data can be acquired in a variety of ways.
  • the trajectory node network is composed of multiple trajectory nodes.
  • Each trajectory node is composed of longitude, latitude, and time point.
  • the trajectory node can be expressed as (66°W, 36°N, 16:08).
  • Fig. 1b is a flowchart of an embodiment of creating a trajectory node network in the present application.
  • that the electronic device creates a trajectory node network based on the multiple longitudes, the multiple latitudes, and the multiple time points includes:
  • the electronic device fills the abscissa of each node, the ordinate of each node, and the vertical coordinate of each node into blank nodes in sequence to obtain multiple trajectory nodes, and enters the multiple trajectory nodes into the pre-division
  • the adjacent trajectory nodes entered in the table are connected to each other to obtain the trajectory node network.
  • the pre-divided table is divided according to longitude and latitude.
  • the trajectory node contains Time attributes and location attributes help improve the recognition accuracy of dangerous driving behaviors.
  • the trajectory characteristic information refers to the characteristic of the trajectory node in the trajectory node network.
  • the electronic device extracting trajectory feature information from the trajectory node network includes:
  • S131 Combine the multiple random walk sequences to obtain a target sequence, where the target sequence includes multiple target nodes;
  • S132 Determine the position of each target node in the target sequence, and convert each target node into a node vector according to the position;
  • S133 Multiply each node vector by the first preset matrix, and calculate the average value of the vector obtained after the multiplication to obtain an intermediate vector;
  • S134 Multiply the intermediate vector points by a second preset matrix to obtain a target matrix, where each column vector in the target matrix represents a vector corresponding to each target node;
  • S135 Calculate the predicted probability of each target node in the target matrix by using an activation function
  • the electronic device uses each vertex trajectory node in the trajectory node network as a starting point to perform a random walk until it reaches the end trajectory node to obtain the multiple random walk sequences.
  • Node 1 and node 2 are vertex trajectory nodes, respectively, node 7 and node 8 are end trajectory nodes respectively, the node 1 is used to perform a random walk until the node 7 is reached, and the random walk sequence 1 is obtained as: 1 -3-5-7; Use the node 1 to perform a random walk until the node 8 is reached, and the random walk sequence 2 is obtained: 1-3-4-6-8; Use the node 2 to perform a random walk Random walk until the node 7 is reached, and the random walk sequence 3 is obtained as: 2-3-5-7; the node 2 is used for random walk until the node 8 is walked, and the random walk sequence 3 is obtained.
  • Wandering sequence 4 is: 2-3-4-6-8.
  • the electronic device merges the multiple random walks according to the generation order of the random walk sequence, and following the above example, the generation order of the multiple random walk sequences is: random walk sequence 1, random walk Sequence 2, random walk sequence 3, random walk sequence 4, get the target sequence 1-3-5-7-1-3-4-6-8-2-3-5-7-2-3-4 -6-8.
  • the target node refers to a trajectory node.
  • the position of the node 3 in the target sequence is: serial number 2, serial number 6, serial number 11, and serial number 15, and the position of node 3 corresponds to
  • the vector value is determined to be 1, and the vector value corresponding to the remaining positions is determined to be 0, and the node vector of the node 3 is obtained as (0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0).
  • the first preset matrix is a matrix set according to an application scenario
  • the second preset matrix is a column matrix set according to an application scenario.
  • the trajectory feature information can be accurately extracted from the trajectory node network.
  • the electronic device performs numerical vector processing on the driving behavior data to obtain a first vector, and the electronic device uses the SkipGram framework to convert the trajectory feature information into a second vector.
  • the vector By converting the driving behavior data into a first vector and converting the trajectory feature information into a second vector, the vector can be directly input into a binary classification model for driving behavior analysis, which facilitates the analysis of driving behavior recognition.
  • the driving recognition result may also be stored in a node of a blockchain.
  • the target vector is a vector obtained by concatenating the first vector after special diagnosis extraction and the second vector after feature extraction.
  • the driving recognition result includes safe driving and dangerous driving.
  • FIG. 1d is a flowchart of an embodiment of generating a target vector in the present application.
  • the fusion of the first vector and the second vector by the electronic device to obtain the target vector includes:
  • S150 Perform feature extraction on the first vector by using a bidirectional long and short-term memory network to obtain a first forward vector corresponding to the forward long-short-term memory network and a first reverse vector corresponding to the reverse long- and short-term memory network;
  • S151 Perform feature extraction on the second vector by using a bidirectional long and short-term memory network to obtain a second forward vector corresponding to the forward long- and short-term memory network and a second reverse vector corresponding to the reverse long- and short-term memory network;
  • the dimensions of the first vector and the second vector can be reduced, thereby improving the efficiency of determining the driving recognition result.
  • the forward vector, the first reverse vector, the second forward vector, and the second reverse vector enable the target vector to have the characteristics of the first vector and the second vector, and then It can improve the recognition accuracy of driving behavior.
  • Fig. 1e is a flowchart of another embodiment of the driving behavior recognition method of the present application. This embodiment is improved on the basis of the driving behavior recognition method shown in FIG. 1 to FIG. 1d. As shown in FIG. 1e, this embodiment inputs the target vector to the pre-built in step S15 shown in FIG. Before the two-category model, the method may include the following steps:
  • S20 Use web crawler technology to obtain historical driving data from a preset website, and convert the historical driving data into a driving data vector;
  • the historical driving data includes data corresponding to dangerous driving, and also includes data corresponding to safe driving.
  • Adjusting the learner through the driving data vector in the verification data set can improve the accuracy of the two-class model.
  • the electronic device when the driving recognition result is dangerous driving, the electronic device sends preset safety information to the terminal device.
  • the driver to be tested can be reminded to pay attention to safety in time.
  • the method when the driving recognition result is dangerous driving, the method further includes:
  • the dangerous driving information of the driver to be tested can be stored, so that the insurance company can determine the premium of the driver to be tested.
  • the dangerous driving Information which can prevent the dangerous driving information from being tampered with, and improve the safety of the dangerous driving information.
  • this application can determine the driver to be tested according to the driving behavior recognition request when a driving behavior recognition request is received, and can accurately determine the driver to be tested from the driving behavior recognition request, and obtain The driving behavior data and navigation data of the driver to be tested, wherein the navigation data includes multiple longitudes, multiple latitudes, and multiple time points, and a trajectory node is created based on the longitude, the latitude, and the time point
  • the network by converting the navigation data into a network of trajectory nodes, can not only visually refer to the navigation data, but also pave the way for the rapid extraction of feature information.
  • Extracting trajectory feature information from the trajectory node network can reduce navigation data The time spent in analysis, thereby improving the recognition efficiency, converting the driving behavior data into a first vector, and converting the trajectory feature information into a second vector, fusing the first vector and the second vector, Obtain a target vector, make the target vector have the characteristics of the driving behavior data and the trajectory feature information, improve the comprehensiveness of the target vector, and input the target vector into a pre-built two-class model to obtain The driving recognition result improves the accuracy of driving behavior recognition by analyzing the comprehensive target vector.
  • this application does not need to install sensors in the vehicle or install sensors on the road, the implementation of the solution is improved. Sex.
  • the driving behavior recognition device 11 includes a determination unit 110, an acquisition unit 111, a creation unit 112, an extraction unit 113, a conversion unit 114, an input unit 115, a division unit 116, a training unit 117, an adjustment unit 118, a detection unit 119, and a generation unit 120.
  • the module/unit referred to in this application refers to a series of computer-readable instruction segments that can be executed by the processor 13 and can complete fixed functions, and are stored in the memory 12. In this embodiment, the functions of each module/unit will be described in detail in subsequent embodiments.
  • the determining unit 110 determines the driver to be tested according to the driving behavior recognition request.
  • the driving behavior recognition request can be triggered by an employee of the insurance company, or can be triggered automatically within a preset time.
  • the preset time may be a time point or a time period.
  • the data information carried in the driving behavior identification request includes, but is not limited to: a preset tag, an identification code, a request code, and the like.
  • the preset label may be name; the identification code may be 23458abd; and the request code may be 001.
  • the determining unit 110 determining the driver to be tested according to the driving behavior recognition request includes:
  • the driver to be tested is determined according to the identification code.
  • the preset label is a pre-defined identifier, for example, the preset label may be the ID of the driver to be tested.
  • the identification code may be an ID card.
  • the data information includes: "ID, 2568871515; name, Li Sheng”.
  • the preset tag is acquired as ID
  • the determining unit 110 acquires the information corresponding to the ID from "ID, 2568871515; name, Li Sheng" as 2568871515, and determines 2568871515 as the identification code.
  • the driving behavior recognition request is parsed through an idle thread. Since there is no need to wait for the thread to process other requests, the parsing speed of the driving behavior recognition request is increased, and the identification code can be accurately determined through the preset label, and then the identification code can be accurately determined. Describe the driver to be tested.
  • the determining unit 110 determining the driver to be tested according to the identification code includes:
  • the determining unit 110 obtains the user corresponding to the identification code from the public security library as the driver to be tested.
  • the driver to be tested can be accurately determined.
  • the obtaining unit 111 obtains the driving behavior data and navigation data of the driver to be tested, where the navigation data includes multiple longitudes, multiple latitudes, and multiple time points.
  • the driving behavior data includes: driving speed, whether the driver's terminal device is in a communication state when the vehicle is running, and the like.
  • each time point refers to the time that the vehicle driven by the driver to be tested stays at the corresponding longitude and latitude.
  • the obtaining unit 111 obtains the driving behavior data and navigation data of the driver to be tested, including one or a combination of the following methods:
  • the acquisition unit 111 determines the license plate associated with the driver to be tested, acquires the vehicle-mounted system associated with the license plate, and acquires information corresponding to the first identifier from the vehicle-mounted system as the driving behavior Data, and obtain information corresponding to the second identifier as the navigation data.
  • the acquisition unit 111 determines the terminal device associated with the driver to be tested, acquires the driving behavior data from the terminal device, and determines the location data collection device associated with the driver to be tested, And derive the navigation data from the position data collection device.
  • the driving behavior data of the driver to be tested and the navigation data can be acquired in a variety of ways.
  • the creating unit 112 creates a trajectory node network based on the multiple longitudes, the multiple latitudes, and the multiple time points.
  • the trajectory node network is composed of multiple trajectory nodes.
  • Each trajectory node is composed of longitude, latitude, and time point.
  • the trajectory node can be expressed as (66°W, 36°N, 16:08).
  • the creating unit 112 creates a trajectory node network based on the multiple longitudes, the multiple latitudes, and the multiple time points, including:
  • the abscissa of each node, the ordinate of each node, and the vertical coordinate of each node are merged into a trajectory node to obtain the trajectory node network.
  • the creation unit 112 fills the abscissa of each node, the ordinate of each node, and the vertical coordinate of each node into blank nodes in sequence to obtain multiple trajectory nodes, and enters the multiple trajectory nodes into the preset In the divided table, the trajectory nodes that are adjacent to each other in the input table are connected to each other to obtain the trajectory node network.
  • the pre-divided table is divided according to longitude and latitude.
  • the trajectory node contains Time attributes and location attributes help improve the recognition accuracy of dangerous driving behaviors.
  • the extraction unit 113 extracts trajectory feature information from the trajectory node network.
  • the trajectory characteristic information refers to the characteristic of the trajectory node in the trajectory node network.
  • the extraction unit 113 extracting trajectory feature information from the trajectory node network includes:
  • Target sequence includes multiple target nodes
  • the target node with the largest predicted probability is determined as the trajectory feature information.
  • the extraction unit 113 uses each vertex trajectory node in the trajectory node network as a starting point to perform a random walk until it reaches the end trajectory node to obtain the multiple random walk sequences.
  • Node 1 and node 2 are vertex trajectory nodes, respectively, node 7 and node 8 are end trajectory nodes respectively, the node 1 is used to perform a random walk until the node 7 is reached, and the random walk sequence 1 is obtained as: 1 -3-5-7; Use the node 1 to perform a random walk until the node 8 is reached, and the random walk sequence 2 is obtained: 1-3-4-6-8; Use the node 2 to perform a random walk Random walk until the node 7 is reached, and the random walk sequence 3 is obtained as: 2-3-5-7; the node 2 is used for random walk until the node 8 is walked, and the random walk sequence 3 is obtained.
  • Wandering sequence 4 is: 2-3-4-6-8.
  • the extraction unit 113 merges the multiple random walks according to the generation order of the random walk sequence, and following the above example, the generation order of the multiple random walk sequences is: random walk sequence 1, random walk Walk sequence 2, random walk sequence 3, random walk sequence 4, get the target sequence 1-3-5-7-1-3-4-6-8-2-3-5-7-2-3- 4-6-8.
  • the target node refers to a trajectory node.
  • the position of the node 3 in the target sequence is: serial number 2, serial number 6, serial number 11, and serial number 15, and the position of node 3 corresponds to
  • the vector value is determined to be 1, and the vector value corresponding to the remaining positions is determined to be 0, and the node vector of the node 3 is obtained as (0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0).
  • the first preset matrix is a matrix set according to an application scenario
  • the second preset matrix is a column matrix set according to an application scenario.
  • the trajectory feature information can be accurately extracted from the trajectory node network.
  • the conversion unit 114 converts the driving behavior data into a first vector, and converts the trajectory characteristic information into a second vector.
  • the conversion unit 114 performs numerical vector processing on the driving behavior data to obtain a first vector, and the conversion unit 114 uses the SkipGram framework to convert the trajectory feature information into a second vector. vector.
  • the vector By converting the driving behavior data into a first vector and converting the trajectory feature information into a second vector, the vector can be directly input into a binary classification model for driving behavior analysis, which facilitates the analysis of driving behavior recognition.
  • the input unit 115 fuses the first vector and the second vector to obtain a target vector, and inputs the target vector into a pre-built binary classification model to obtain a driving recognition result.
  • the driving recognition result may also be stored in a node of a blockchain.
  • the target vector is a vector obtained by concatenating the first vector after special diagnosis extraction and the second vector after feature extraction.
  • the driving recognition result includes safe driving and dangerous driving.
  • the input unit 115 fusing the first vector and the second vector to obtain a target vector includes:
  • the first forward vector, the first reverse vector, the second forward vector, and the second reverse vector are spliced to obtain the target vector.
  • the dimensions of the first vector and the second vector can be reduced, thereby improving the efficiency of determining the driving recognition result.
  • the forward vector, the first reverse vector, the second forward vector, and the second reverse vector enable the target vector to have the characteristics of the first vector and the second vector, and then It can improve the recognition accuracy of driving behavior.
  • the obtaining unit 111 before inputting the target vector to the pre-built binary classification model, uses web crawler technology to obtain historical driving data from a preset website, and combines the historical driving data. Conversion of driving data into driving data vector;
  • the dividing unit 116 divides the driving data vector to obtain a training data set and a verification data set
  • the training unit 117 trains the driving data vector in the training data set to obtain a learner
  • the adjustment unit 118 adjusts the learner according to the driving data vector in the verification data set to obtain the two-class model.
  • the historical driving data includes data corresponding to dangerous driving, and also includes data corresponding to safe driving.
  • Adjusting the learner through the driving data vector in the verification data set can improve the accuracy of the two-class model.
  • the electronic device when the driving recognition result is dangerous driving, the electronic device sends preset safety information to the terminal device.
  • the driver to be tested can be reminded to pay attention to safety in time.
  • the detection unit 119 detects the length of time during which the driving recognition result is dangerous driving;
  • the generating unit 120 When detecting that the duration is greater than or equal to the configured time, the generating unit 120 acquires the request number of the driving behavior identification request, and generates dangerous driving information according to the driver to be tested, the duration, and the request number;
  • the encryption unit 121 uses a symmetric encryption algorithm to encrypt the dangerous driving information to obtain the target ciphertext;
  • the storage unit 122 stores the target ciphertext.
  • the dangerous driving information of the driver to be tested can be stored, so that the insurance company can determine the premium of the driver to be tested.
  • the dangerous driving Information which can prevent the dangerous driving information from being tampered with, and improve the safety of the dangerous driving information.
  • this application can determine the driver to be tested according to the driving behavior recognition request when a driving behavior recognition request is received, and can accurately determine the driver to be tested from the driving behavior recognition request, and obtain The driving behavior data and navigation data of the driver to be tested, wherein the navigation data includes multiple longitudes, multiple latitudes, and multiple time points, and a trajectory node is created based on the longitude, the latitude, and the time point
  • the network by converting the navigation data into a network of trajectory nodes, can not only visually refer to the navigation data, but also pave the way for the rapid extraction of feature information.
  • Extracting trajectory feature information from the trajectory node network can reduce navigation data The time spent in analysis, thereby improving the recognition efficiency, converting the driving behavior data into a first vector, and converting the trajectory feature information into a second vector, fusing the first vector and the second vector, Obtain a target vector, make the target vector have the characteristics of the driving behavior data and the trajectory feature information, improve the comprehensiveness of the target vector, and input the target vector into a pre-built two-class model to obtain The driving recognition result improves the accuracy of driving behavior recognition by analyzing the comprehensive target vector.
  • this application does not need to install sensors in the vehicle or install sensors on the road, the implementation of the solution is improved. Sex.
  • FIG. 3 it is a schematic structural diagram of an electronic device implementing an embodiment of the driving behavior recognition method of the present application.
  • the electronic device 1 includes, but is not limited to, a memory 12, a processor 13, and computer-readable instructions stored in the memory 12 and running on the processor 13 , Such as driving behavior to recognize readable instructions.
  • the schematic diagram is only an example of the electronic device 1 and does not constitute a limitation on the electronic device 1. It may include more or less components than those shown in the figure, or a combination of certain components, or different components. Components, for example, the electronic device 1 may also include an input/output device, a network access device, a bus, and the like.
  • the processor 13 may be a central processing unit (Central Processing Unit, CPU), other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc.
  • the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.
  • the processor 13 is the computing core and control center of the electronic device 1 and connects the entire electronic device with various interfaces and lines. Each part of 1, and executes the operating system of the electronic device 1 and various installed application readable instructions, readable instruction codes, etc.
  • the processor 13 executes the operating system of the electronic device 1 and various installed application readable instructions.
  • the processor 13 executes the application-readable instructions to implement the steps in the above embodiments of the driving behavior recognition method, for example, the steps shown in FIG. 1.
  • the computer-readable instructions may be divided into one or more modules/units, and the one or more modules/units are stored in the memory 12 and executed by the processor 13 to Complete this application.
  • the one or more modules/units may be a series of computer-readable instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer-readable instructions in the electronic device 1.
  • the computer-readable instructions may be divided into a determination unit 110, an acquisition unit 111, a creation unit 112, an extraction unit 113, a conversion unit 114, an input unit 115, a division unit 116, a training unit 117, an adjustment unit 118, and a detection unit. 119.
  • the memory 12 may be used to store the computer-readable instructions and/or modules.
  • the processor 13 runs or executes the computer-readable instructions and/or modules stored in the memory 12 and calls the computer-readable instructions and/or modules stored in the memory 12
  • the data inside realizes various functions of the electronic device 1.
  • the memory 12 may mainly include a storage readable instruction area and a storage data area, wherein the storage readable instruction area may store an operating system and application readable instructions required by at least one function (such as a sound playback function, an image playback function, etc.) Etc.; the data storage area can store data created according to the use of electronic devices, etc.
  • the memory 12 may include non-volatile and volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a Secure Digital (SD) card, a flash memory card ( Flash Card), at least one magnetic disk storage device, flash memory device, or other storage device.
  • non-volatile and volatile memory such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a Secure Digital (SD) card, a flash memory card ( Flash Card), at least one magnetic disk storage device, flash memory device, or other storage device.
  • the memory 12 may be an external memory and/or an internal memory of the electronic device 1. Further, the memory 12 may be a memory in a physical form, such as a memory stick, a TF card (Trans-flash Card), and so on.
  • TF card Trans-flash Card
  • the integrated module/unit of the electronic device 1 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium, which may be non-volatile. It can also be volatile.
  • this application implements all or part of the processes in the above-mentioned embodiments and methods, and can also be completed by instructing relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium.
  • the computer-readable instruction when executed by the processor, it can implement the steps of the foregoing method embodiments.
  • the computer-readable instruction includes computer-readable instruction code
  • the computer-readable instruction code may be in the form of source code, object code, executable file, or some intermediate form.
  • the computer-readable medium may include: any entity or device capable of carrying the computer-readable instruction code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory, random access memory, etc. .
  • the blockchain referred to in this application is a new application mode of computer technology such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information for verification. The validity of the information (anti-counterfeiting) and the generation of the next block.
  • the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.
  • the memory 12 in the electronic device 1 stores multiple readable instructions to implement a driving behavior recognition method, and the processor 13 can execute the multiple readable instructions to achieve:
  • the first vector and the second vector are fused to obtain a target vector, and the target vector is input into a pre-built binary classification model to obtain a driving recognition result.
  • modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional modules in the various embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional modules.

Abstract

本申请涉及人工智能技术领域,提供一种驾驶行为识别方法、装置、电子设备及存储介质,该方法能够根据驾驶行为识别请求确定待测驾驶员,获取待测驾驶员的驾驶行为数据及导航数据,其中,导航数据包括多个经度、多个纬度及多个时间点,基于多个经度、多个纬度及多个时间点创建轨迹节点网络,并提取轨迹特征信息,将驾驶行为数据转换为第一向量,并将轨迹特征信息转换为第二向量,融合第一向量及第二向量,得到目标向量,并将目标向量输入至预先构建的二分类模型中,得到驾驶识别结果,本申请能够提高驾驶行为识别效率及驾驶行为识别的准确性。此外,本申请还涉及区块链技术,所述驾驶识别结果可存储于区块链中。

Description

驾驶行为识别方法、装置、电子设备及存储介质
本申请要求于2020年09月30日提交中国专利局,申请号为202011059578.6,发明名称为“驾驶行为识别方法、装置、电子设备及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及人工智能技术领域,尤其涉及一种驾驶行为识别方法、装置、电子设备及存储介质。
背景技术
随着社会经济的发展,越来越多的人开始拥有私家车,交通事故的发生频次也随之增加,为此,越来越多的人会通过购买交通保险来降低自己的风险损失。然而,危险驾驶行为(如超速驾驶、开车打电话、急速过弯等)会增加交通事故的发生概率,为保险公司带来一定的损失。为此,有效地识别出参保人员的危险驾驶行为对保险公司来说有着重大的意义。
现有的驾驶行为识别方式是通过在车辆上安装多种传感器来识别驾驶人员是否发生危险驾驶行为,如:区段测速等,然而,发明人意识到,在车辆上安装传感器的方式或者在道路上安装传感器是为方案的执行增加了难度,同时,通过传感器确定驾驶人员是否发生危险驾驶行为的准确性还依赖于传感器的性能。
因此,如何构建一种可执行的驾驶行为识别方案,以提高驾驶行为识别效率,成为有待解决的技术问题。
发明内容
鉴于以上内容,有必要提供一种驾驶行为识别方法、装置、电子设备及存储介质,能够提高驾驶行为识别效率及驾驶行为识别的准确性。
本申请的第一方面提供一种驾驶行为识别方法,所述驾驶行为识别方法包括:
接收到驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
从所述轨迹节点网络中提取轨迹特征信息;
将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
本申请的第二方面提供一种电子设备,所述电子设备包括处理器和存储器,所述处理器用于执行所述存储器中存储的计算机可读指令以实现以下步骤:
接收到驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
从所述轨迹节点网络中提取轨迹特征信息;
将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
本申请的第三方面提供一种计算机可读存储介质,所述计算机可读存储介质上存储有至少一个计算机可读指令,所述至少一个计算机可读指令被处理器执行以实现以下步骤:
接收到驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
从所述轨迹节点网络中提取轨迹特征信息;
将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
本申请的第四方面提供一种驾驶行为识别装置,所述驾驶行为识别包括:
确定单元,用于接收到驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
获取单元,用于获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
创建单元,用于基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
提取单元,用于从所述轨迹节点网络中提取轨迹特征信息;
转换单元,用于将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
输入单元,用于融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
由以上技术方案可以看出,本申请根据驾驶行为识别请求确定待测驾驶员,能够准确确定待测驾驶员,获取所述待测驾驶员的驾驶行为数据及导航数据,基于所述经度、所述纬度及所述时间点创建轨迹节点网络,通过将所述导航数据转换为轨迹节点网络,不仅能够直观查阅导航数据,同时,还能够为特征信息的快速提取作铺垫,从所述轨迹节点网络中提取轨迹特征信息,能够减少导航数据分析时所耗费的时间,进而提高识别效率,将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量,融合所述第一向量及所述第二向量,得到目标向量,使所述目标向量具有所述驾驶行为数据及所述轨迹特征信息的特征,提高所述目标向量的全面性,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果,通过分析具有全面性的目标向量,提高了驾驶行为识别的准确性,另外,由于本申请无需在车辆内安装传感器或者无需在道路上安装传感器,为此,提高了方案的可执行性。
附图说明
图1是本申请驾驶行为识别方法的一实施例的流程图。
图1a是本申请确定待测驾驶员的一实施例的流程图。
图1b是本申请创建轨迹节点网络的一实施例的流程图。
图1c是本申请提取轨迹特征信息的一实施例的流程图。
图1d是本申请生成目标向量的一实施例的流程图。
图1e是本申请驾驶行为识别方法的另一实施例的流程图。
图1f是本申请生成多个随机游走序列的一实施例的流程图。
图2是本申请驾驶行为识别装置的一实施例的功能模块图。
图3是本申请实现驾驶行为识别方法的一实施例的电子设备的结构示意图。
具体实施方式
为了使本申请的目的、技术方案和优点更加清楚,下面结合附图和具体实施例对本申请进行详细描述。
如图1所示,是本申请驾驶行为识别方法的一实施例的流程图。根据不同的需求,该流程图中步骤的顺序可以改变,某些步骤可以省略。
所述驾驶行为识别方法可应用于智慧交通场景中,从而推动智慧城市的建设。所述驾驶行为识别方法应用于一个或者多个电子设备中,所述电子设备是一种能够按照事先设定或存储的可读指令,自动进行数值计算和/或信息处理的设备,其硬件包括但不限于微处理器、专用集成电路(Application Specific Integrated Circuit,ASIC)、可编程门阵列(Field-Programmable Gate Array,FPGA)、数字处理器(Digital Signal Processor,DSP)、嵌入式设备等。
所述电子设备可以是任何一种可与用户进行人机交互的电子产品,例如,个人计算机、平板电脑、智能手机、个人数字助理(Personal Digital Assistant,PDA)、游戏机、交互式网络电视(Internet Protocol Television,IPTV)、智能穿戴式设备等。
所述电子设备还可以包括网络设备和/或用户设备。其中,所述网络设备包括,但不限于单个网络服务器、多个网络服务器组成的服务器组或基于云计算(Cloud Computing)的由大量主机或网络服务器构成的云。
所述电子设备所处的网络包括,但不限于:互联网、广域网、城域网、局域网、虚拟专用网络(Virtual Private Network,VPN)等。
S10,接收到驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员。
在本申请的至少一个实施例中,所述驾驶行为识别请求可以由保险公司的工作人员触发,也可以在预设时间内自动触发。
其中,所述预设时间可以是一个时间点,也可以是一个时间段。
在本申请的至少一个实施例中,所述驾驶行为识别请求携带的数据信息包括,但不限于:预设标签、身份识别码、请求编码等。例如,所述预设标签可以是name;所述身份识别码可以是23458abd;所述请求编码可以是001。
参见图1a,图1a是本申请确定待测驾驶员的一实施例的流程图。在本申请的至少一个实施例中,所述电子设备根据所述驾驶行为识别请求确定待测驾驶员,包括:
S100,从预设线程池中获取任意闲置线程;
S101,利用所述任意闲置线程解析所述驾驶行为识别请求,得到所述驾驶行为识别请求携带的数据信息;
S102,从标签表中获取预设标签,所述预设标签是指预先定义好的标签;
S103,从所述数据信息中获取与所述预设标签对应的信息,作为身份识别码;
S104,根据所述身份识别码确定所述待测驾驶员。
其中,所述预设标签是预先定义好的标识,例如:所述预设标签可以是所述待测驾驶员的ID。
进一步地,所述身份识别码可以是身份证。
例如:所述数据信息包括:“ID,2568871515;name,李生”。获取预设标签为ID,所述电子设备从“ID,2568871515;name,李生”中获取与ID对应的信息为2568871515,将2568871515确定为身份识别码。
通过闲置线程解析所述驾驶行为识别请求,由于无需等待线程处理其他请求,因此,提高所述驾驶行为识别请求的解析速度,进而通过预设标签能够准确确定所述身份识别码,进而准确确定所述待测驾驶员。
具体地,所述电子设备根据所述身份识别码确定所述待测驾驶员包括:
所述电子设备从公安库中获取与所述身份识别码对应的用户作为所述待测驾驶员。
由于身份识别码具有唯一性,因此,能够准确确定出所述待测驾驶员。
S11,获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点。
在本申请的至少一个实施例中,所述驾驶行为数据包括:驾驶速度、车辆行驶时驾驶员的终端设备是否处于通信状态等。
在本申请的至少一个实施例中,每个时间点是指所述待测驾驶员驾驶的车辆在相应的经度及纬度停留的时间。
在本申请的至少一个实施例中,所述电子设备获取所述待测驾驶员的驾驶行为数据及导航数据,包括以下一种或者多种方式的组合:
(1)确定与所述待测驾驶员关联的车牌,获取与所述车牌关联的车载系统,从所述车载系统中获取与第一标识对应的信息,作为所述驾驶行为数据,并获取与第二标识对应的信息,作为所述导航数据。
(2)确定与所述待测驾驶员关联的终端设备,并从所述终端设备中获取所述驾驶行为数据,确定与所述待测驾驶员关联的位置数据采集装置,并从所述位置数据采集装置导出所述导航数据。
通过上述实施方式,能够通过多种方式获取所述待测驾驶员的驾驶行为数据及所述导航数据。
S12,基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络。
在本申请的至少一个实施例中,所述轨迹节点网络是由多个轨迹节点构成的。每个轨迹节点都由经度、纬度及时间点组成,例如,轨迹节点可以表示为(66°W,36°N,16:08)。
参见图1b,图1b是本申请创建轨迹节点网络的一实施例的流程图。在本申请的至少一个实施例中,所述电子设备基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络,包括:
S120,将所述多个经度中的每个经度确定为节点横坐标,将所述多个纬度中的每个纬度确定为节点纵坐标,并将所述多个时间点中的每个时间点确定为节点竖坐标;
S121,将每个节点横坐标、每个节点纵坐标及每个节点竖坐标融合成轨迹节点,得到所述轨迹节点网络。
具体地,所述电子设备将每个节点横坐标、每个节点纵坐标及每个节点竖坐标依次填充至空白节点中,得到多个轨迹节点,并将所述多个轨迹节点录入到预先划分好的表格中,将录入表格中的互为相邻的轨迹节点相互连接,得到所述轨迹节点网络。
其中,所述预先划分好的表格是依据经度与维度划分的。
可以理解的是,在多落石地带,车辆停放的时间越长,发生危险驾驶的可能性越大,由此,通过将时间点与经度、纬度融合成轨迹节点,使所述轨迹节点中包含有时间属性及位置属性,有利于提高危险驾驶行为的识别精度。
S13,从所述轨迹节点网络中提取轨迹特征信息。
在本申请的至少一个实施例中,所述轨迹特征信息是指所述轨迹节点网络中的轨迹节点的特征。
参见图1c,图1c是本申请提取轨迹特征信息的一实施例的流程图。在本申请的至少一个实施例中,所述电子设备从所述轨迹节点网络中提取轨迹特征信息包括:
S130,以所述轨迹节点网络中每个顶点轨迹节点作为起点进行随机游走,得到多个随机游走序列;
S131,合并所述多个随机游走序列,得到目标序列,所述目标序列中包括多个目标节点;
S132,确定每个目标节点在所述目标序列中的位置,并根据所述位置将每个目标节点转换为节点向量;
S133,将每个节点向量分别与第一预设矩阵相乘,并计算相乘后得到的向量的平均值,得到中间向量;
S134,将所述中间向量点乘第二预设矩阵,得到目标矩阵,所述目标矩阵中每列向量表征每个目标节点对应的向量;
S135,采用激活函数计算所述目标矩阵中每个目标节点的预测概率;
S136,将预测概率最大的目标节点确定为所述轨迹特征信息。
具体地,所述电子设备以所述轨迹节点网络中每个顶点轨迹节点作为起点进行随机游走,直至游走至末端轨迹节点,得到所述多个随机游走序列。
例如:参见图1f,图1f是本申请生成多个随机游走序列的一实施例的流程图。节点1及节点2分别为顶点轨迹节点,节点7及节点8分别为末端轨迹节点,以所述节点1进行随机游走,直至游走至所述节点7,得到随机游走序列1为:1-3-5-7;以所述节点1进行随机游走,直至游走至所述节点8,得到随机游走序列2为:1-3-4-6-8;以所述节点2进行随机游走,直至游走至所述节点7,得到随机游走序列3为:2-3-5-7;以所述节点2进行随机游走,直至游走至所述节点8,得到随机游走序列4为:2-3-4-6-8。
进一步地,所述电子设备按照随机游走序列的生成顺序合并所述多个随机游走,承接上述例子,所述多个随机游走序列的生成顺序为:随机游走序列1、随机游走序列2、随机游走序列3、随机游走序列4,得到目标序列为1-3-5-7-1-3-4-6-8-2-3-5-7-2-3-4-6-8。
更进一步地,所述目标节点是指轨迹节点。承接上述例子,当目标节点为节点3时,确定出所述节点3在所述目标序列中的位置有:序号2、序号6、序号11及序号15,将所述节点3所在的位置对应的向量值确定为1,其余位置对应的向量值确定为0,得到所述节点3的节点向量为(0,1,0,0,0,1,0,0,0,0,1,0,0,0,1,0,0,0)。
更进一步地,所述第一预设矩阵是根据应用场景设置的矩阵,所述第二预设矩阵是根据应用场景设置的列矩阵。
通过上述实施方式,能够准确地从所述轨迹节点网络中提取所述轨迹特征信息。
S14,将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量。
在本申请的至少一个实施例中,所述电子设备对所述驾驶行为数据进行数值化向量处理,得到第一向量,所述电子设备利用SkipGram框架将所述轨迹特征信息转换为第二向量。
通过将所述驾驶行为数据转换为第一向量,及将所述轨迹特征信息转换为第二向量,能够将向量直接输入二分类模型进行驾驶行为分析,便于驾驶行为识别的分析。
S15,融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
需要强调的是,为进一步保证上述驾驶识别结果的私密和安全性,上述驾驶识别结果还可以存储于一区块链的节点中。
在本申请的至少一个实施例中,所述目标向量是通过拼接特诊提取后的第一向量与特征提取后的第二向量而得到的向量。
在本申请的至少一个实施例中,所述驾驶识别结果包括安全驾驶、危险驾驶。
参见图1d,图1d是本申请生成目标向量的一实施例的流程图。在本申请的至少一个实施例中,所述电子设备融合所述第一向量及所述第二向量,得到目标向量包括:
S150,利用双向长短期记忆网络对所述第一向量进行特征抽取,得到与正向长短期记忆网络对应的第一正向向量,及与反向长短期记忆网络对应的第一反向向量;
S151,利用双向长短期记忆网络对所述第二向量进行特征抽取,得到与正向长短期记忆网络对应的第二正向向量,及与反向长短期记忆网络对应的第二反向向量;
S152,拼接所述第一正向向量、所述第一反向向量、所述第二正向向量及所述第二反向向量,得到所述目标向量。
通过对所述第一向量及所述第二向量进行特征抽取,能够降低所述第一向量及所述第二向量的维度,进而提高所说驾驶识别结果的确定效率,通过拼接所述第一正向向量、所述第一反向向量、所述第二正向向量及所述第二反向向量,能够使所述目标向量具有所述第一向量及所述第二向量的特性,进而能够提高驾驶行为的识别精度。
进一步的,参见图1e,图1e是本申请驾驶行为识别方法的另一实施例的流程图。本实施例是在图1至图1d所示的驾驶行为识别方法的基础进行改进得到的,如图1e所示,本实施例在图1所示的步骤S15将所述目标向量输入至预先构建的二分类模型之前,所述还可包括如下步骤:
S20,采用网络爬虫技术从预设网站上获取历史驾驶数据,并将所述历史驾驶数据转换为驾驶数据向量;
S21,划分所述驾驶数据向量,得到训练数据集及验证数据集;
S22,训练所述训练数据集中的驾驶数据向量,得到学习器;
S23,根据所述验证数据集中的驾驶数据向量调整所述学习器,得到所述二分类模型。
其中,所述历史驾驶数据包括危险驾驶时对应的数据,也包括安全驾驶时对应的数据。
通过所述验证数据集中的驾驶数据向量对所述学习器进行调整,能够提高所述二分类模型的准确度。
在本申请的至少一个实施例中,当所述驾驶识别结果为危险驾驶时,所述电子设备向所述终端设备发送预设安全信息。
通过向所述终端设备发送所述预设安全信息,能够及时提醒所述待测驾驶员注意安全。
在本申请的至少一个实施例中,当所述驾驶识别结果为危险驾驶时,所述方法还包括:
检测所述驾驶识别结果为危险驾驶的时长;
当检测到所述时长大于或者等于配置时间时,获取所述驾驶行为识别请求的请求编号,并根据所述待测驾驶员、所述时长及所述请求编号生成危险驾驶信息;
采用对称加密算法加密所述危险驾驶信息,得到目标密文;
存储所述目标密文。
通过上述实施方式,能够在所述时长大于所述配置时间时,存储所述待测驾驶员的危险驾驶信息,以便保险公司确定所述待测驾驶员的保费,此外,通过加密所述危险驾驶信息,能够避免所述危险驾驶信息被篡改,提高所述危险驾驶信息的安全性。
由以上技术方案可以看出,本申请能够当接收到驾驶行为识别请求时,根据所述驾驶行为识别请求确定待测驾驶员,从所述驾驶行为识别请求中能够准确确定待测驾驶员,获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点,基于所述经度、所述纬度及所述时间点创建轨迹节点网络,通过将所述导航数据转换为轨迹节点网络,不仅能够直观查阅导航数据,同时,还能够为特征信息的快速提取作铺垫,从所述轨迹节点网络中提取轨迹特征信息,能够减少导航数据分析时所耗费的时间,进而提高识别效率,将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量,融合所述第一向量及所述第二向量,得到目标向量,使所述目标向量具有所述驾驶行为数据及所述轨迹特征信息的特征,提高所述目标向量的全面性,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果,通过分析具有全面性的目标向量,提高了驾驶行为识别的准确性,另外,由于本申请无需在车辆内安装传感器或者无需在道路上安装传感器,为此,提高了方案的可执行性。
如图2所示,是本申请驾驶行为识别装置的一实施例的功能模块图。所述驾驶行为识别装置11包括确定单元110、获取单元111、创建单元112、提取单元113、转换单元114、输入单元115、划分单元116、训练单元117、调整单元118、检测单元119、生成单元120、加密单元121及存储单元122。本申请所称的模块/单元是指一种能够被处理器13所执行,并且能够完成固定功能的一系列计算机可读指令段,其存储在存储器12中。在本实施例中,关于各模块/单元的功能将在后续的实施例中详述。
当接收到驾驶行为识别请求时,确定单元110根据所述驾驶行为识别请求确定待测驾驶员。
在本申请的至少一个实施例中,所述驾驶行为识别请求可以由保险公司的工作人员触发,也可以在预设时间内自动触发。
其中,所述预设时间可以是一个时间点,也可以是一个时间段。
在本申请的至少一个实施例中,所述驾驶行为识别请求携带的数据信息包括,但不限于:预设标签、身份识别码、请求编码等。例如,所述预设标签可以是name;所述身份识别码可以是23458abd;所述请求编码可以是001。
在本申请的至少一个实施例中,所述确定单元110根据所述驾驶行为识别请求确定待测驾驶员,包括:
从预设线程池中获取任意闲置线程;
利用所述任意闲置线程解析所述驾驶行为识别请求,得到所述驾驶行为识别请求携带的数据信息;
从标签表中获取预设标签,所述预设标签是指预先定义好的标签;
从所述数据信息中获取与所述预设标签对应的信息,作为身份识别码;
根据所述身份识别码确定所述待测驾驶员。
其中,所述预设标签是预先定义好的标识,例如:所述预设标签可以是所述待测驾驶员的ID。
进一步地,所述身份识别码可以是身份证。
例如:所述数据信息包括:“ID,2568871515;name,李生”。获取预设标签为ID,所述确定单元110从“ID,2568871515;name,李生”中获取与ID对应的信息为2568871515,将2568871515确定为身份识别码。
通过闲置线程解析所述驾驶行为识别请求,由于无需等待线程处理其他请求,因此,提高所述驾驶行为识别请求的解析速度,进而通过预设标签能够准确确定所述身份识别码,进而准确确定所述待测驾驶员。
具体地,所述确定单元110根据所述身份识别码确定所述待测驾驶员包括:
所述确定单元110从公安库中获取与所述身份识别码对应的用户作为所述待测驾驶员。
由于身份识别码具有唯一性,因此,能够准确确定出所述待测驾驶员。
获取单元111获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点。
在本申请的至少一个实施例中,所述驾驶行为数据包括:驾驶速度、车辆行驶时驾驶员的终端设备是否处于通信状态等。
在本申请的至少一个实施例中,每个时间点是指所述待测驾驶员驾驶的车辆在相应的经度及纬度停留的时间。
在本申请的至少一个实施例中,所述获取单元111获取所述待测驾驶员的驾驶行为数据及导航数据,包括以下一种或者多种方式的组合:
(1)所述获取单元111确定与所述待测驾驶员关联的车牌,获取与所述车牌关联的车载系统,从所述车载系统中获取与第一标识对应的信息,作为所述驾驶行为数据,并获取与第二标识对应的信息,作为所述导航数据。
(2)所述获取单元111确定与所述待测驾驶员关联的终端设备,并从所述终端设备中获取所述驾驶行为数据,确定与所述待测驾驶员关联的位置数据采集装置,并从所述位置数据采集装置导出所述导航数据。
通过上述实施方式,能够通过多种方式获取所述待测驾驶员的驾驶行为数据及所述导航数据。
创建单元112基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络。
在本申请的至少一个实施例中,所述轨迹节点网络是由多个轨迹节点构成的。每个轨迹节点都由经度、纬度及时间点组成,例如,轨迹节点可以表示为(66°W,36°N,16:08)。
在本申请的至少一个实施例中,所述创建单元112基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络,包括:
将所述多个经度中的每个经度确定为节点横坐标,将所述多个纬度中的每个纬度确定为 节点纵坐标,并将所述多个时间点中的每个时间点确定为节点竖坐标;
将每个节点横坐标、每个节点纵坐标及每个节点竖坐标融合成轨迹节点,得到所述轨迹节点网络。
具体地,所述创建单元112将每个节点横坐标、每个节点纵坐标及每个节点竖坐标依次填充至空白节点中,得到多个轨迹节点,并将所述多个轨迹节点录入到预先划分好的表格中,将录入表格中的互为相邻的轨迹节点相互连接,得到所述轨迹节点网络。
其中,所述预先划分好的表格是依据经度与维度划分的。
可以理解的是,在多落石地带,车辆停放的时间越长,发生危险驾驶的可能性越大,由此,通过将时间点与经度、纬度融合成轨迹节点,使所述轨迹节点中包含有时间属性及位置属性,有利于提高危险驾驶行为的识别精度。
提取单元113从所述轨迹节点网络中提取轨迹特征信息。
在本申请的至少一个实施例中,所述轨迹特征信息是指所述轨迹节点网络中的轨迹节点的特征。
在本申请的至少一个实施例中,所述提取单元113从所述轨迹节点网络中提取轨迹特征信息包括:
以所述轨迹节点网络中每个顶点轨迹节点作为起点进行随机游走,得到多个随机游走序列;
合并所述多个随机游走序列,得到目标序列,所述目标序列中包括多个目标节点;
确定每个目标节点在所述目标序列中的位置,并根据所述位置将每个目标节点转换为节点向量;
将每个节点向量分别与第一预设矩阵相乘,并计算相乘后得到的向量的平均值,得到中间向量;
将所述中间向量点乘第二预设矩阵,得到目标矩阵,所述目标矩阵中每列向量表征每个目标节点对应的向量;
采用激活函数计算所述目标矩阵中每个目标节点的预测概率;
将预测概率最大的目标节点确定为所述轨迹特征信息。
具体地,所述提取单元113以所述轨迹节点网络中每个顶点轨迹节点作为起点进行随机游走,直至游走至末端轨迹节点,得到所述多个随机游走序列。
例如:参见图1f,图1f是本申请生成多个随机游走序列的一实施例的流程图。节点1及节点2分别为顶点轨迹节点,节点7及节点8分别为末端轨迹节点,以所述节点1进行随机游走,直至游走至所述节点7,得到随机游走序列1为:1-3-5-7;以所述节点1进行随机游走,直至游走至所述节点8,得到随机游走序列2为:1-3-4-6-8;以所述节点2进行随机游走,直至游走至所述节点7,得到随机游走序列3为:2-3-5-7;以所述节点2进行随机游走,直至游走至所述节点8,得到随机游走序列4为:2-3-4-6-8。
进一步地,所述提取单元113按照随机游走序列的生成顺序合并所述多个随机游走,承接上述例子,所述多个随机游走序列的生成顺序为:随机游走序列1、随机游走序列2、随机游走序列3、随机游走序列4,得到目标序列为1-3-5-7-1-3-4-6-8-2-3-5-7-2-3-4-6-8。
更进一步地,所述目标节点是指轨迹节点。承接上述例子,当目标节点为节点3时,确定出所述节点3在所述目标序列中的位置有:序号2、序号6、序号11及序号15,将所述节点3所在的位置对应的向量值确定为1,其余位置对应的向量值确定为0,得到所述节点3的节点向量为(0,1,0,0,0,1,0,0,0,0,1,0,0,0,1,0,0,0)。
更进一步地,所述第一预设矩阵是根据应用场景设置的矩阵,所述第二预设矩阵是根据应用场景设置的列矩阵。
通过上述实施方式,能够准确地从所述轨迹节点网络中提取所述轨迹特征信息。
转换单元114将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量。
在本申请的至少一个实施例中,所述转换单元114对所述驾驶行为数据进行数值化向量处理,得到第一向量,所述转换单元114利用SkipGram框架将所述轨迹特征信息转换为第二向量。
通过将所述驾驶行为数据转换为第一向量,及将所述轨迹特征信息转换为第二向量,能够将向量直接输入二分类模型进行驾驶行为分析,便于驾驶行为识别的分析。
输入单元115融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
需要强调的是,为进一步保证上述驾驶识别结果的私密和安全性,上述驾驶识别结果还可以存储于一区块链的节点中。
在本申请的至少一个实施例中,所述目标向量是通过拼接特诊提取后的第一向量与特征提取后的第二向量而得到的向量。
在本申请的至少一个实施例中,所述驾驶识别结果包括安全驾驶、危险驾驶。
在本申请的至少一个实施例中,所述输入单元115融合所述第一向量及所述第二向量,得到目标向量包括:
利用双向长短期记忆网络对所述第一向量进行特征抽取,得到与正向长短期记忆网络对应的第一正向向量,及与反向长短期记忆网络对应的第一反向向量;
利用双向长短期记忆网络对所述第二向量进行特征抽取,得到与正向长短期记忆网络对应的第二正向向量,及与反向长短期记忆网络对应的第二反向向量;
拼接所述第一正向向量、所述第一反向向量、所述第二正向向量及所述第二反向向量,得到所述目标向量。
通过对所述第一向量及所述第二向量进行特征抽取,能够降低所述第一向量及所述第二向量的维度,进而提高所说驾驶识别结果的确定效率,通过拼接所述第一正向向量、所述第一反向向量、所述第二正向向量及所述第二反向向量,能够使所述目标向量具有所述第一向量及所述第二向量的特性,进而能够提高驾驶行为的识别精度。
在本申请的至少一个实施例中,在将所述目标向量输入至预先构建的二分类模型之前,所述获取单元111采用网络爬虫技术从预设网站上获取历史驾驶数据,并将所述历史驾驶数据转换为驾驶数据向量;
划分单元116划分所述驾驶数据向量,得到训练数据集及验证数据集;
训练单元117训练所述训练数据集中的驾驶数据向量,得到学习器;
调整单元118根据所述验证数据集中的驾驶数据向量调整所述学习器,得到所述二分类模型。
其中,所述历史驾驶数据包括危险驾驶时对应的数据,也包括安全驾驶时对应的数据。
通过所述验证数据集中的驾驶数据向量对所述学习器进行调整,能够提高所述二分类模型的准确度。
在本申请的至少一个实施例中,当所述驾驶识别结果为危险驾驶时,所述电子设备向所述终端设备发送预设安全信息。
通过向所述终端设备发送所述预设安全信息,能够及时提醒所述待测驾驶员注意安全。
在本申请的至少一个实施例中,当所述驾驶识别结果为危险驾驶时,检测单元119检测所述驾驶识别结果为危险驾驶的时长;
当检测到所述时长大于或者等于配置时间时,生成单元120获取所述驾驶行为识别请求的请求编号,并根据所述待测驾驶员、所述时长及所述请求编号生成危险驾驶信息;
加密单元121采用对称加密算法加密所述危险驾驶信息,得到目标密文;
存储单元122存储所述目标密文。
通过上述实施方式,能够在所述时长大于所述配置时间时,存储所述待测驾驶员的危险驾驶信息,以便保险公司确定所述待测驾驶员的保费,此外,通过加密所述危险驾驶信息,能够避免所述危险驾驶信息被篡改,提高所述危险驾驶信息的安全性。
由以上技术方案可以看出,本申请能够当接收到驾驶行为识别请求时,根据所述驾驶行为识别请求确定待测驾驶员,从所述驾驶行为识别请求中能够准确确定待测驾驶员,获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点,基于所述经度、所述纬度及所述时间点创建轨迹节点网络,通过将所述导航数据转换为轨迹节点网络,不仅能够直观查阅导航数据,同时,还能够为特征信息的快速提取作铺垫,从所述轨迹节点网络中提取轨迹特征信息,能够减少导航数据分析时所耗费的时间,进而提高识别效率,将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量,融合所述第一向量及所述第二向量,得到目标向量,使所述目标向量具有所述驾驶行为数据及所述轨迹特征信息的特征,提高所述目标向量的全面性,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果,通过分析具有全面性的目标向量,提高了驾驶行为识别的准确性,另外,由于本申请无需在车辆内安装传感器或者无需在道路上安装传感器,为此,提高了方案的可执行性。
如图3所示,是本申请实现驾驶行为识别方法的一实施例的电子设备的结构示意图。
在本申请的一个实施例中,所述电子设备1包括,但不限于,存储器12、处理器13,以及存储在所述存储器12中并可在所述处理器13上运行的计算机可读指令,例如驾驶行为识别可读指令。
本领域技术人员可以理解,所述示意图仅仅是电子设备1的示例,并不构成对电子设备1的限定,可以包括比图示更多或更少的部件,或者组合某些部件,或者不同的部件,例如所述电子设备1还可以包括输入输出设备、网络接入设备、总线等。
所述处理器13可以是中央处理单元(Central Processing Unit,CPU),还可以是其他通用处理器、数字信号处理器(Digital Signal Processor,DSP)、专用集成电路(Application Specific Integrated Circuit,ASIC)、现场可编程门阵列(Field-Programmable Gate Array,FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等,所述处理器13是所述电子设备1的运算核心和控制中心,利用各种接口和线路连接整个电子设备1的各个部分,及执行所述电子设备1的操作系统以及安装的各类应用可读指令、可读指令代码等。
所述处理器13执行所述电子设备1的操作系统以及安装的各类应用可读指令。所述处理器13执行所述应用可读指令以实现上述各个驾驶行为识别方法实施例中的步骤,例如图1所示的步骤。
示例性的,所述计算机可读指令可以被分割成一个或多个模块/单元,所述一个或者多个模块/单元被存储在所述存储器12中,并由所述处理器13执行,以完成本申请。所述一个或多个模块/单元可以是能够完成特定功能的一系列计算机可读指令段,该指令段用于描述所述计算机可读指令在所述电子设备1中的执行过程。例如,所述计算机可读指令可以被分割成确定单元110、获取单元111、创建单元112、提取单元113、转换单元114、输入单元115、划分单元116、训练单元117、调整单元118、检测单元119、生成单元120、加密单元121及存储单元122。
所述存储器12可用于存储所述计算机可读指令和/或模块,所述处理器13通过运行或执行存储在所述存储器12内的计算机可读指令和/或模块,以及调用存储在存储器12内的数据,实现所述电子设备1的各种功能。所述存储器12可主要包括存储可读指令区和存储数据区,其中,存储可读指令区可存储操作系统、至少一个功能所需的应用可读指令(比如声音播放功能、图像播放功能等)等;存储数据区可存储根据电子设备的使用所创建的数据等。此外,存储器12可以包括非易失性和易失性存储器,例如硬盘、内存、插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)、至少一个磁盘存储器件、闪存器件、或其他存储器件。
所述存储器12可以是电子设备1的外部存储器和/或内部存储器。进一步地,所述存储器12可以是具有实物形式的存储器,如内存条、TF卡(Trans-flash Card)等等。
所述电子设备1集成的模块/单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中,该计算机存储介质可以是非易失性,也可以是易失性的。基于这样的理解,本申请实现上述实施例方法中的全部或部分流程,也可以通过计算机可读指令来指令相关的硬件来完成,所述的计算机可读指令可存储于一计算机可读存储介质中,该计算机可读指令在被处理器执行时,可实现上述各个方法实施例的步骤。
其中,所述计算机可读指令包括计算机可读指令代码,所述计算机可读指令代码可以为源代码形式、对象代码形式、可执行文件或某些中间形式等。所述计算机可读介质可以包括:能够携带所述计算机可读指令代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器、随机存取存储器等。
本申请所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain),本质上是一个去中心化的数据库,是一串使用密码学方法相关联产生的数据块,每一个数据块中包含了一批次网络交易的信息,用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
结合图1,所述电子设备1中的所述存储器12存储多个可读指令以实现一种驾驶行为识别方法,所述处理器13可执行所述多个可读指令从而实现:
接收到驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
从所述轨迹节点网络中提取轨迹特征信息;
将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
具体地,所述处理器13对上述可读指令的具体实现方法可参考图1对应实施例中相关步骤的描述,在此不赘述。
在本申请所提供的几个实施例中,应该理解到,所揭露的系统,装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,所述模块的划分,仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式。
所述作为分离部件说明的模块可以是或者也可以不是物理上分开的,作为模块显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。
另外,在本申请各个实施例中的各功能模块可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用硬件加软件功能模块的形式实现。
因此,无论从哪一点来看,均应将实施例看作是示范性的,而且是非限制性的,本申请的范围由所附权利要求而不是上述说明限定,因此旨在将落在权利要求的等同要件的含义和范围内的所有变化涵括在本申请内。不应将权利要求中的任何附关联图标记视为限制所涉及的权利要求。
此外,显然“包括”一词不排除其他单元或步骤,单数不排除复数。所述的多个单元或装置也可以由一个单元或装置通过软件或者硬件来实现。第一、第二等词语用来表示名称,而并不表示任何特定的顺序。
最后应说明的是,以上实施例仅用以说明本申请的技术方案而非限制,尽管参照实施例对本申请进行了详细说明,本领域的普通技术人员应当理解,可以对本申请的技术方案进行修改或等同替换,而不脱离本申请技术方案的精神和范围。

Claims (20)

  1. 一种驾驶行为识别方法,其中,所述驾驶行为识别方法包括:
    接收驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
    获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
    基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
    从所述轨迹节点网络中提取轨迹特征信息;
    将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
    融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
  2. 根据权利要求1所述的驾驶行为识别方法,其中,所述根据所述驾驶行为识别请求确定待测驾驶员,包括:
    从预设线程池中获取任意闲置线程;
    利用所述任意闲置线程解析所述驾驶行为识别请求,得到所述驾驶行为识别请求携带的数据信息;
    从标签表中获取预设标签,所述预设标签是指预先定义好的标签;
    从所述数据信息中获取与所述预设标签对应的信息,作为身份识别码;
    根据所述身份识别码确定所述待测驾驶员。
  3. 根据权利要求1所述的驾驶行为识别方法,其中,所述获取所述待测驾驶员的驾驶行为数据及导航数据,包括以下一种或者多种方式的组合:
    确定与所述待测驾驶员关联的车牌,获取与所述车牌关联的车载系统,从所述车载系统中获取与第一标识对应的信息,作为所述驾驶行为数据,并获取与第二标识对应的信息,作为所述导航数据;及/或
    确定与所述待测驾驶员关联的终端设备,并从所述终端设备中获取所述驾驶行为数据,确定与所述待测驾驶员关联的位置数据采集装置,并从所述位置数据采集装置导出所述导航数据。
  4. 根据权利要求1所述的驾驶行为识别方法,其中,所述基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络,包括:
    将所述多个经度中的每个经度确定为节点横坐标,将所述多个纬度中的每个纬度确定为节点纵坐标,并将所述多个时间点中的每个时间点确定为节点竖坐标;
    将每个节点横坐标、每个节点纵坐标及每个节点竖坐标融合成轨迹节点,得到所述轨迹节点网络。
  5. 根据权利要求1所述的驾驶行为识别方法,其中,所述从所述轨迹节点网络中提取轨迹特征信息,包括:
    以所述轨迹节点网络中每个顶点轨迹节点作为起点进行随机游走,得到多个随机游走序列;
    合并所述多个随机游走序列,得到目标序列,所述目标序列中包括多个目标节点;
    确定每个目标节点在所述目标序列中的位置,并根据所述位置将每个目标节点转换为节点向量;
    将每个节点向量分别与第一预设矩阵相乘,并计算相乘后得到的向量的平均值,得到中间向量;
    将所述中间向量点乘第二预设矩阵,得到目标矩阵,所述目标矩阵中每列向量表征每个目标节点对应的向量;
    采用激活函数计算所述目标矩阵中每个目标节点的预测概率;
    将预测概率最大的目标节点确定为所述轨迹特征信息。
  6. 根据权利要求1所述的驾驶行为识别方法,其中,所述融合所述第一向量及所述第二向量,得到目标向量,包括:
    利用双向长短期记忆网络对所述第一向量进行特征抽取,得到与正向长短期记忆网络对应的第一正向向量,及与反向长短期记忆网络对应的第一反向向量;
    利用双向长短期记忆网络对所述第二向量进行特征抽取,得到与正向长短期记忆网络对应的第二正向向量,及与反向长短期记忆网络对应的第二反向向量;
    拼接所述第一正向向量、所述第一反向向量、所述第二正向向量及所述第二反向向量,得到所述目标向量。
  7. 根据权利要求1所述的驾驶行为识别方法,其中,在将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果之前,所述方法还包括:
    采用网络爬虫技术从预设网站上获取历史驾驶数据,并将所述历史驾驶数据转换为驾驶数据向量;
    划分所述驾驶数据向量,得到训练数据集及验证数据集;
    训练所述训练数据集中的驾驶数据向量,得到学习器;
    根据所述验证数据集中的驾驶数据向量调整所述学习器,得到所述二分类模型。
  8. 一种驾驶行为识别装置,其中,所述驾驶行为识别装置包括:
    确定单元,用于接收到驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
    获取单元,用于获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
    创建单元,用于基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
    提取单元,用于从所述轨迹节点网络中提取轨迹特征信息;
    转换单元,用于将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
    输入单元,用于融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
  9. 一种电子设备,其中,所述电子设备包括处理器和存储器,所述处理器用于执行存储器中存储的至少一个计算机可读指令以实现以下步骤:
    接收驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
    获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
    基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
    从所述轨迹节点网络中提取轨迹特征信息;
    将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
    融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
  10. 根据权利要求9所述的电子设备,其中,在所述根据所述驾驶行为识别请求确定待测驾驶员时,所述处理器执行所述至少一个计算机可读指令以实现以下步骤:
    从预设线程池中获取任意闲置线程;
    利用所述任意闲置线程解析所述驾驶行为识别请求,得到所述驾驶行为识别请求携带的数据信息;
    从标签表中获取预设标签,所述预设标签是指预先定义好的标签;
    从所述数据信息中获取与所述预设标签对应的信息,作为身份识别码;
    根据所述身份识别码确定所述待测驾驶员。
  11. 根据权利要求9所述的电子设备,其中,在所述获取所述待测驾驶员的驾驶行为数 据及导航数据时,所述处理器执行所述至少一个计算机可读指令以实现以下步骤:
    确定与所述待测驾驶员关联的车牌,获取与所述车牌关联的车载系统,从所述车载系统中获取与第一标识对应的信息,作为所述驾驶行为数据,并获取与第二标识对应的信息,作为所述导航数据;及/或
    确定与所述待测驾驶员关联的终端设备,并从所述终端设备中获取所述驾驶行为数据,确定与所述待测驾驶员关联的位置数据采集装置,并从所述位置数据采集装置导出所述导航数据。
  12. 根据权利要求9所述的电子设备,其中,在所述基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络时,所述处理器执行所述至少一个计算机可读指令以实现以下步骤:
    将所述多个经度中的每个经度确定为节点横坐标,将所述多个纬度中的每个纬度确定为节点纵坐标,并将所述多个时间点中的每个时间点确定为节点竖坐标;
    将每个节点横坐标、每个节点纵坐标及每个节点竖坐标融合成轨迹节点,得到所述轨迹节点网络。
  13. 根据权利要求9所述的电子设备,其中,在所述从所述轨迹节点网络中提取轨迹特征信息时,所述处理器执行所述至少一个计算机可读指令以实现以下步骤:
    以所述轨迹节点网络中每个顶点轨迹节点作为起点进行随机游走,得到多个随机游走序列;
    合并所述多个随机游走序列,得到目标序列,所述目标序列中包括多个目标节点;
    确定每个目标节点在所述目标序列中的位置,并根据所述位置将每个目标节点转换为节点向量;
    将每个节点向量分别与第一预设矩阵相乘,并计算相乘后得到的向量的平均值,得到中间向量;
    将所述中间向量点乘第二预设矩阵,得到目标矩阵,所述目标矩阵中每列向量表征每个目标节点对应的向量;
    采用激活函数计算所述目标矩阵中每个目标节点的预测概率;
    将预测概率最大的目标节点确定为所述轨迹特征信息。
  14. 根据权利要求9所述的电子设备,其中,在所述融合所述第一向量及所述第二向量,得到目标向量时,所述处理器执行所述至少一个计算机可读指令以实现以下步骤:
    利用双向长短期记忆网络对所述第一向量进行特征抽取,得到与正向长短期记忆网络对应的第一正向向量,及与反向长短期记忆网络对应的第一反向向量;
    利用双向长短期记忆网络对所述第二向量进行特征抽取,得到与正向长短期记忆网络对应的第二正向向量,及与反向长短期记忆网络对应的第二反向向量;
    拼接所述第一正向向量、所述第一反向向量、所述第二正向向量及所述第二反向向量,得到所述目标向量。
  15. 一种计算机可读存储介质,其中,所述计算机可读存储介质存储有至少一个计算机可读指令,所述至少一个计算机可读指令被处理器执行时实现以下步骤:
    接收驾驶行为识别请求,根据所述驾驶行为识别请求确定待测驾驶员;
    获取所述待测驾驶员的驾驶行为数据及导航数据,其中,所述导航数据包括多个经度、多个纬度及多个时间点;
    基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络;
    从所述轨迹节点网络中提取轨迹特征信息;
    将所述驾驶行为数据转换为第一向量,并将所述轨迹特征信息转换为第二向量;
    融合所述第一向量及所述第二向量,得到目标向量,并将所述目标向量输入至预先构建的二分类模型中,得到驾驶识别结果。
  16. 根据权利要求15所述的存储介质,其中,在所述根据所述驾驶行为识别请求确定 待测驾驶员时,所述至少一个计算机可读指令被处理器执行以实现以下步骤:
    从预设线程池中获取任意闲置线程;
    利用所述任意闲置线程解析所述驾驶行为识别请求,得到所述驾驶行为识别请求携带的数据信息;
    从标签表中获取预设标签,所述预设标签是指预先定义好的标签;
    从所述数据信息中获取与所述预设标签对应的信息,作为身份识别码;
    根据所述身份识别码确定所述待测驾驶员。
  17. 根据权利要求15所述的存储介质,其中,在所述获取所述待测驾驶员的驾驶行为数据及导航数据时,所述至少一个计算机可读指令被处理器执行以实现以下步骤:
    确定与所述待测驾驶员关联的车牌,获取与所述车牌关联的车载系统,从所述车载系统中获取与第一标识对应的信息,作为所述驾驶行为数据,并获取与第二标识对应的信息,作为所述导航数据;及/或
    确定与所述待测驾驶员关联的终端设备,并从所述终端设备中获取所述驾驶行为数据,确定与所述待测驾驶员关联的位置数据采集装置,并从所述位置数据采集装置导出所述导航数据。
  18. 根据权利要求15所述的存储介质,其中,在所述基于所述多个经度、所述多个纬度及所述多个时间点创建轨迹节点网络时,所述至少一个计算机可读指令被处理器执行以实现以下步骤:
    将所述多个经度中的每个经度确定为节点横坐标,将所述多个纬度中的每个纬度确定为节点纵坐标,并将所述多个时间点中的每个时间点确定为节点竖坐标;
    将每个节点横坐标、每个节点纵坐标及每个节点竖坐标融合成轨迹节点,得到所述轨迹节点网络。
  19. 根据权利要求15所述的存储介质,其中,在所述从所述轨迹节点网络中提取轨迹特征信息时,所述至少一个计算机可读指令被处理器执行时以实现以下步骤:
    以所述轨迹节点网络中每个顶点轨迹节点作为起点进行随机游走,得到多个随机游走序列;
    合并所述多个随机游走序列,得到目标序列,所述目标序列中包括多个目标节点;
    确定每个目标节点在所述目标序列中的位置,并根据所述位置将每个目标节点转换为节点向量;
    将每个节点向量分别与第一预设矩阵相乘,并计算相乘后得到的向量的平均值,得到中间向量;
    将所述中间向量点乘第二预设矩阵,得到目标矩阵,所述目标矩阵中每列向量表征每个目标节点对应的向量;
    采用激活函数计算所述目标矩阵中每个目标节点的预测概率;
    将预测概率最大的目标节点确定为所述轨迹特征信息。
  20. 根据权利要求15所述的存储介质,其中,在所述融合所述第一向量及所述第二向量,得到目标向量时,所述至少一个计算机可读指令被处理器执行以实现以下步骤:
    利用双向长短期记忆网络对所述第一向量进行特征抽取,得到与正向长短期记忆网络对应的第一正向向量,及与反向长短期记忆网络对应的第一反向向量;
    利用双向长短期记忆网络对所述第二向量进行特征抽取,得到与正向长短期记忆网络对应的第二正向向量,及与反向长短期记忆网络对应的第二反向向量;
    拼接所述第一正向向量、所述第一反向向量、所述第二正向向量及所述第二反向向量,得到所述目标向量。
PCT/CN2020/131953 2020-09-30 2020-11-26 驾驶行为识别方法、装置、电子设备及存储介质 WO2021115133A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011059578.6 2020-09-30
CN202011059578.6A CN112200046B (zh) 2020-09-30 2020-09-30 驾驶行为识别方法、装置、电子设备及存储介质

Publications (1)

Publication Number Publication Date
WO2021115133A1 true WO2021115133A1 (zh) 2021-06-17

Family

ID=74007243

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/131953 WO2021115133A1 (zh) 2020-09-30 2020-11-26 驾驶行为识别方法、装置、电子设备及存储介质

Country Status (2)

Country Link
CN (1) CN112200046B (zh)
WO (1) WO2021115133A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113269179A (zh) * 2021-06-24 2021-08-17 中国平安人寿保险股份有限公司 数据处理方法、装置、设备及存储介质
CN113968234A (zh) * 2021-11-29 2022-01-25 蔡定海 一种车辆辅助驾驶控制方法、装置及车载控制器
CN114463984A (zh) * 2022-03-02 2022-05-10 智道网联科技(北京)有限公司 车辆轨迹显示方法及相关设备
CN114529871A (zh) * 2022-02-21 2022-05-24 创新奇智(上海)科技有限公司 一种酒驾识别方法、装置,电子设备及存储介质
CN114757304A (zh) * 2022-06-10 2022-07-15 北京芯盾时代科技有限公司 一种数据识别方法、装置、设备及存储介质
CN116524723A (zh) * 2023-06-27 2023-08-01 天津大学 一种货车轨迹异常识别方法及系统

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113807457A (zh) * 2021-09-26 2021-12-17 北京市商汤科技开发有限公司 确定路网表征信息的方法、装置、设备及存储介质
CN114187489B (zh) * 2021-12-14 2024-04-30 中国平安财产保险股份有限公司 车辆异常驾驶风险检测方法、装置、电子设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109214107A (zh) * 2018-09-26 2019-01-15 大连海事大学 一种船舶航行行为在线预测方法
CN110631594A (zh) * 2019-10-24 2019-12-31 成都大成均图科技有限公司 基于复杂轨迹网络划分模型的离线地图匹配方法和系统
CN111310583A (zh) * 2020-01-19 2020-06-19 中国科学院重庆绿色智能技术研究院 一种基于改进的长短期记忆网络的车辆异常行为识别方法
CN111325437A (zh) * 2019-02-18 2020-06-23 北京嘀嘀无限科技发展有限公司 异常行驶行为的识别方法、装置以及电子设备

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11769159B2 (en) * 2017-11-13 2023-09-26 Aloke Chaudhuri System and method for human emotion and identity detection
CN110533912B (zh) * 2019-09-16 2022-05-20 腾讯科技(深圳)有限公司 基于区块链的驾驶行为检测方法及装置
CN111694917B (zh) * 2020-06-10 2021-03-02 北京嘀嘀无限科技发展有限公司 一种车辆异常轨迹检测、模型训练方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109214107A (zh) * 2018-09-26 2019-01-15 大连海事大学 一种船舶航行行为在线预测方法
CN111325437A (zh) * 2019-02-18 2020-06-23 北京嘀嘀无限科技发展有限公司 异常行驶行为的识别方法、装置以及电子设备
CN110631594A (zh) * 2019-10-24 2019-12-31 成都大成均图科技有限公司 基于复杂轨迹网络划分模型的离线地图匹配方法和系统
CN111310583A (zh) * 2020-01-19 2020-06-19 中国科学院重庆绿色智能技术研究院 一种基于改进的长短期记忆网络的车辆异常行为识别方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHANG GUOYIN; TAN FENG; WU YANXIA: "Ship Motion Attitude Prediction Based on an Adaptive Dynamic Particle Swarm Optimization Algorithm and Bidirectional LSTM Neural Network", IEEE ACCESS, IEEE, USA, vol. 8, 11 May 2020 (2020-05-11), USA, pages 90087 - 90098, XP011790042, DOI: 10.1109/ACCESS.2020.2993909 *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113269179A (zh) * 2021-06-24 2021-08-17 中国平安人寿保险股份有限公司 数据处理方法、装置、设备及存储介质
CN113269179B (zh) * 2021-06-24 2024-04-05 中国平安人寿保险股份有限公司 数据处理方法、装置、设备及存储介质
CN113968234A (zh) * 2021-11-29 2022-01-25 蔡定海 一种车辆辅助驾驶控制方法、装置及车载控制器
CN113968234B (zh) * 2021-11-29 2023-05-02 深圳市科莱德电子有限公司 一种车辆辅助驾驶控制方法、装置及车载控制器
CN114529871A (zh) * 2022-02-21 2022-05-24 创新奇智(上海)科技有限公司 一种酒驾识别方法、装置,电子设备及存储介质
CN114463984A (zh) * 2022-03-02 2022-05-10 智道网联科技(北京)有限公司 车辆轨迹显示方法及相关设备
CN114463984B (zh) * 2022-03-02 2024-02-27 智道网联科技(北京)有限公司 车辆轨迹显示方法及相关设备
CN114757304A (zh) * 2022-06-10 2022-07-15 北京芯盾时代科技有限公司 一种数据识别方法、装置、设备及存储介质
CN116524723A (zh) * 2023-06-27 2023-08-01 天津大学 一种货车轨迹异常识别方法及系统
CN116524723B (zh) * 2023-06-27 2023-09-12 天津大学 一种货车轨迹异常识别方法及系统

Also Published As

Publication number Publication date
CN112200046A (zh) 2021-01-08
CN112200046B (zh) 2023-09-12

Similar Documents

Publication Publication Date Title
WO2021115133A1 (zh) 驾驶行为识别方法、装置、电子设备及存储介质
Rasouli et al. Understanding pedestrian behavior in complex traffic scenes
WO2021114736A1 (zh) 医疗问诊辅助方法、装置、电子设备及介质
US11042619B2 (en) Vehicle occupant tracking and trust
CN110572354B (zh) 用于实时车辆事故管理的区块链和加密货币
US20150161464A1 (en) Detecting and reporting improper activity involving a vehicle
EP4036886A2 (en) Method and apparatus for monitoring vehicle, cloud control platform and system for vehicle-road collaboration
US9323993B2 (en) On-street parking management methods and systems for identifying a vehicle via a camera and mobile communications devices
WO2022134760A1 (zh) 数据处理方法、装置、电子设备及介质
WO2021217931A1 (zh) 基于分类模型的字段抽取方法、装置、电子设备及介质
US10733541B2 (en) System, method, and recording medium for geolocation data discovery in streaming texts
CN112016520B (zh) 基于ai的交通违法凭证生成方法、装置、终端及存储介质
CN112132032A (zh) 交通标志牌检测方法、装置、电子设备及存储介质
CN111950621A (zh) 基于人工智能的目标数据检测方法、装置、设备及介质
US20220161830A1 (en) Dynamic Scene Representation
Yao et al. Accident responsibility identification model for Internet of Vehicles based on lightweight blockchain
Gowri et al. Evaluation of right-turn lanes at signalized intersection in non-lane-based heterogeneous traffic using microscopic simulation model
WO2022073513A1 (zh) 信息录入辅助方法、装置、电子设备及存储介质
CN112115890B (zh) 基于人工智能的酒驾识别方法、装置、设备及介质
CN114201482A (zh) 人口动态分布统计方法、装置、电子设备及可读存储介质
CN111882873B (zh) 轨迹异常检测方法、装置、设备及介质
US11371854B2 (en) Method and apparatus for processing transit time of navigation path, device and computer storage medium
CN111724502B (zh) 车辆行驶数据处理方法、装置、设备和存储介质
CN112132037B (zh) 基于人工智能的人行道检测方法、装置、设备及介质
CN111933241B (zh) 医疗数据解析方法、装置、电子设备及存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20900601

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20900601

Country of ref document: EP

Kind code of ref document: A1