AI-based in-vehicle safety monitoring system and method

Abstract

The invention relates to the technical field of computer systems based on specific models, and discloses an AI-based in-vehicle safety monitoring system and method, and the system comprises a vehicle safety monitoring authorization identification module, a vehicle safety monitoring analysis module, and a vehicle safety monitoring result feedback processing module. In-vehicle real-time state image parameters are dynamically obtained through a shooting lens, and real data support is provided for accurately monitoring potential dangerous goods in a vehicle. Image parameters of potential dangerous articles in the vehicle are scientifically preset, and image feature matching is carried out by combining an artificial intelligence algorithm with the real-time state image parameters in the vehicle, so that the potential dangerous articles in the vehicle are intelligently and accurately analyzed, and the intelligence and the safety of the vehicle-mounted monitoring system are improved; the information of potential dangerous goods in the vehicle is efficiently pushed and fed back to the vehicle owner based on the vehicle owner contact characteristic parameters, and the safety monitoring service quality and efficiency of the vehicle-mounted monitoring system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of computer systems based on specific models, and specifically to an AI-based in-vehicle safety monitoring system and method. Background Art

[0002] The vehicle-mounted monitoring architecture is a three-layer networked integrated monitoring system consisting of a vehicle-mounted terminal, a transmission network and a monitoring center. It provides functions such as vehicle anti-theft, anti-robbery, driving route monitoring, real-time wireless transmission of video images inside and outside the vehicle, rapid response to accidents, and call command to solve the dynamic management problems of existing vehicles and improve driving safety. However, the existing vehicle-mounted monitoring system cannot intelligently identify items with safety hazards in the vehicle, nor can it efficiently and intelligently feedback safety hazard information, which reduces the safety and intelligence of the vehicle.

[0003] The Chinese invention patent application with publication number CN118246571A discloses a model training method based on driver status monitoring and related equipment, which collects first training sample data and second training sample data; inputs the second training sample data into the teacher model for predicting the driver status to obtain the third training sample data with pseudo labels; trains the student model according to the first training sample data, the second training sample data and the third training sample data to obtain the weight of the student model after training; updates the weight of the teacher model according to the weight of the student model after training to obtain a new teacher model; repeats the steps of using the new teacher model for data prediction and training the student model to obtain the target teacher model. The accuracy of the model is improved and the time and labor costs of model training are reduced; however, the above technical solution can only predict the driver status and cannot monitor the safety hazards of items in the vehicle online, which reduces the safety of the vehicle. Summary of the invention

[0004] 1. Technical issues to be resolved In order to solve the problem that the above-mentioned existing vehicle-mounted monitoring systems cannot intelligently identify items with safety hazards in the car, and cannot efficiently and intelligently feedback safety hazard information, thereby reducing the safety and intelligence of the vehicle, the above purposes are achieved: accurately identifying the vehicle monitoring driver's authorization information, intelligently analyzing potential dangerous items in the car, efficiently pushing information on potential dangerous items in the car, collecting the owner's reply information within a limited time, accurately evaluating the owner's reception status of potential dangerous items information, and scientifically handling potential dangerous items information in an emergency.

[0005] (II) Technical solution The present invention is implemented by the following technical solution: an AI-based in-vehicle safety monitoring method, the method comprising the following steps: S1. Collecting the vehicle driver's authorization text data; S2, performing vehicle safety monitoring authorization identification processing according to the vehicle driver authorization text data and the vehicle safety monitoring authorization keyword text data, and generating driver vehicle safety monitoring authorization identification data; if it is not passed, directly ending the in-vehicle safety monitoring operation; S3, when the vehicle passes, collect real-time status image data inside the vehicle; S4, analyzing and processing the potential dangerous items in the vehicle according to the real-time status image data in the vehicle and the image data of the potential dangerous items in the vehicle, generating analysis data of the potential dangerous items in the vehicle, and directly ending the vehicle safety monitoring operation when the potential dangerous items do not exist; S5. When it is present, push and feedback the information of the potential dangerous items in the vehicle according to the vehicle owner contact feature data and the analysis data of the potential dangerous items in the vehicle; S6. Performing timing collection and processing of the vehicle owner's reply information based on the time limit data of the reply to the feedback of the potential dangerous items in the vehicle, and generating the vehicle owner's reply feature text data; S7. Analyze and process the vehicle owner's potentially dangerous goods information reception status on the vehicle owner's reply feature text data to generate evaluation data on the vehicle owner's in-vehicle potentially dangerous goods information reception status; when the vehicle owner is in a receiving status, directly terminate the in-vehicle safety monitoring operation; when the vehicle owner is in a non-receiving status, push and feed back the evaluation data on the vehicle owner's in-vehicle potentially dangerous goods information reception status to the vehicle safety monitoring result emergency platform.

[0006] Preferably, the operation steps of collecting the vehicle driver's authorization text data are as follows: S11. Collecting the driver's authorization text information for performing in-vehicle safety monitoring operations online through the vehicle terminal and generating vehicle driver authorization text data .

[0007] Preferably, the vehicle safety monitoring authorization identification processing is performed based on the vehicle driver authorization text data and the vehicle safety monitoring authorization keyword text data to generate the driver vehicle safety monitoring authorization identification data; when it is not passed, the operation steps of directly ending the in-vehicle safety monitoring operation are as follows: S21. Establishing a vehicle safety monitoring authorization keyword text data set , ;in Indicates Vehicle safety monitoring authorization keyword text data, Indicates the maximum value of the vehicle safety monitoring authorization keyword text data, wherein the vehicle safety monitoring authorization keyword text data includes pass, yes, agree, allow and confirm; S22, using the KD tree nearest neighbor search algorithm to search the vehicle driver authorization text data A text data set of keywords related to the vehicle safety monitoring authorization Vehicle safety monitoring authorization keyword text data Perform authorization text character matching and generate driver vehicle safety monitoring authorization identification data based on the authorization text character matching result ; when and If the authorization text character match is successful, it means that the vehicle driver agrees to authorize the execution of the in-vehicle safety monitoring operation, and the driver's vehicle safety monitoring authorization identification data is output. for passing; when and If the authorization text character match is successful, it means that the vehicle driver does not agree to authorize the execution of the in-vehicle safety monitoring operation, then the driver's vehicle safety monitoring authorization identification data is output. If the test result fails, the in-vehicle safety monitoring operation is terminated directly.

[0008] Preferably, when the vehicle passes, the operation steps for collecting the real-time status image data in the vehicle are as follows: S31, when the driver's vehicle safety monitoring authorization identification data When the vehicle passes through, the real-time status image of the vehicle interior is collected online through the camera lens and the real-time status image data of the vehicle interior is generated. .

[0009] Preferably, the operation steps of analyzing and processing the potential dangerous items in the vehicle are performed according to the real-time in-vehicle state image data and the in-vehicle potential dangerous items image data to generate in-vehicle potential dangerous items analysis data, and directly terminating the in-vehicle safety monitoring operation when the potential dangerous items do not exist are as follows: S41. Establishing a data set of images of potentially dangerous objects in the vehicle , ;in Indicates image data of potentially dangerous objects in the car, Indicates the maximum number of types of potentially dangerous items in the car, including bottled water, canned spray, mobile power, lighters, perfume and glasses in direct sunlight; S42: the real-time status image data in the vehicle and a data set of images of potentially dangerous objects in the vehicle Image data of potentially dangerous items in the vehicle Perform image feature matching and generate analysis data of potential dangerous items in the vehicle based on the image feature matching results , execute and generate analysis data of potential dangerous goods in the car The specific steps are as follows: S421. Initialize parameters, update the number of crow populations for dangerous object identification, maximum number of iterations, and flight distance ; S422, initializing the dangerous object recognition crow individual potential dangerous object image data set in the car The initial position and memory in the search space, The dangerous object recognition crows are randomly distributed in a multidimensional search space, namely The random distribution space dimension of dangerous object recognition crows is A collection of potentially dangerous objects image data in vehicles In the search space; in the first iteration, it is assumed that the dangerous object recognition crow individual hides the food in the initial position; that is, in the first iteration, it is assumed that the dangerous object recognition crow individual hides the real-time state image data in the vehicle Image data collection of potentially dangerous objects in the car Image data of potentially dangerous objects in the car at the initial position in the search space Make a match; S423, calculating the individual fitness value of each dangerous object recognition crow, that is, calculating the real-time status image data of the vehicle Image data collection of potentially dangerous objects in the car Image data of potentially dangerous objects in the car in the search space The fitness value of S424, update the dangerous object identification crow individual potential dangerous object image data set in the car The position in the search space, the formula for updating the position of the individual crow for dangerous object identification is as follows: ,in Indicates Iterations Dangerous Object Identification Crow Individual A collection of image data of potentially dangerous objects in vehicles The new position in the search space, Indicates Iteration 1 Dangerous Object Identification Crow Individual A collection of image data of potentially dangerous objects in vehicles The position in the search space, represents a random number uniformly distributed between

[01] , Indicates Iterations Dangerous Object Identification Crow Individual A collection of image data of potentially dangerous objects in vehicles The flight distance in the search space, Indicates Iteration 1 Dangerous Object Identification Crow Individual A collection of image data of potentially dangerous objects in vehicles The food hiding location in the search space, i.e. Iteration 1 Dangerous Object Identification Crow Individual Image dataset of potentially dangerous objects in vehicles Search the search space for the image data related to the real-time status of the vehicle and the image data of the potentially dangerous objects in the vehicle Matching positions; S425, determine the feasibility of the new position, determine the feasibility of the new position of each dangerous object identification crow individual; if the new position of the dangerous object identification crow individual is feasible, the dangerous object identification crow individual will update its position, that is, in the potential dangerous object image data set in the vehicle Search for the real-time status image data in the vehicle Matching image data of potentially dangerous items in the car , the dangerous object recognition crow individual is updated to the successfully matched image data of potential dangerous objects in the car Otherwise, the dangerous object recognition crow individual stays at the current location and does not move to the new location; S426, evaluate the fitness value of the new position, calculate the fitness value of each dangerous object recognition crow's individual new position, that is, calculate the potential dangerous object image data set in the car The real-time status image data in the vehicle described in the search space Image data of potentially dangerous objects in the vehicle at new locations The fitness value of S427, update memory, if the fitness value of the new position of the dangerous object recognition crow is greater than the fitness value of the initial position in the memory, the dangerous object recognition crow updates its memory through the new position, otherwise it does not update its memory; that is, in the potential dangerous object image data set in the car Search for the real-time status image data in the vehicle Image data of potentially dangerous items in the car with the largest fitness value ; S428: When the maximum number of iterations is met, output the real-time status image data in the vehicle Image data of potentially dangerous objects in the vehicle Image feature matching results; S429, according to the real-time status image data in the vehicle in step S428 Image data of potentially dangerous objects in the vehicle The image feature matching results generate analysis data of potential dangerous objects in the car ; when and If the image feature matching is successful, it means there are potentially dangerous items inside the vehicle, and the analysis data of potentially dangerous items inside the vehicle will be output. If it exists, the image data of the potential dangerous objects in the vehicle is output simultaneously. Corresponding data on the types of potentially dangerous items in the vehicle; when and If the image feature matching is successful, it means that there are no potentially dangerous items in the vehicle, and the analysis data of potentially dangerous items in the vehicle will be output. does not exist; at this time, the in-vehicle safety monitoring operation is terminated directly.

[0010] Preferably, when the condition exists, the operation steps of pushing and feedback the information of the potentially dangerous items in the vehicle according to the vehicle owner contact characteristic data and the potentially dangerous items in the vehicle analysis data are as follows: S51. Establishing vehicle owner contact characteristic data , the vehicle owner contact characteristic data includes any one of a mobile phone number, a WeChat account, and a QQ account; S52: When the analysis data of the potentially dangerous items in the vehicle is When the potential dangerous object image data in the vehicle is present, The corresponding data on the types of potentially dangerous items in the vehicle is obtained through the vehicle owner contact feature data The corresponding vehicle owner's contact information is combined with the on-board terminal to push and feedback information on potential dangerous items in the vehicle in real time online.

[0011] Preferably, the operation steps of collecting and processing the owner's reply information timing based on the reply limit time data of the potential dangerous items in the vehicle and generating the owner's reply feature text data are as follows: S61. Establishing the time limit for responding to feedback on potentially dangerous items in the vehicle The data on the time limit for responding to feedback on potentially dangerous items in the vehicle indicates the preset time limit for the vehicle owner to respond to the information after the operation of pushing and responding to information on potentially dangerous items in the vehicle is completed. The unit is minutes; S62: When the operation of pushing and feeding back the information of the potentially dangerous items in the vehicle is completed, the time limit data of the feedback reply based on the potentially dangerous items in the vehicle is Process the time value of the owner's reply information, and at the same time, feedback the time limit data of the potentially dangerous items in the car through the vehicle terminal Collect the owner's reply information online within the corresponding time length, and generate the owner's reply feature text data .

[0012] Preferably, the vehicle owner's potentially dangerous item information reception status analysis and processing are performed on the vehicle owner's reply characteristic text data to generate vehicle owner's in-vehicle potentially dangerous item information reception status evaluation data; when it is in the reception status, the in-vehicle safety monitoring operation is directly terminated; when it is in the non-reception status, the vehicle owner's in-vehicle potentially dangerous item information reception status evaluation data is pushed and fed back to the vehicle safety monitoring result emergency platform in the following operating steps: S71: Replying to the vehicle owner with characteristic text data Perform character analysis on the car owner's reply information, and generate evaluation data on the receiving status of the car owner's potentially dangerous items information based on the character analysis results of the car owner's reply information ; When the car owner replies with characteristic text data When it is empty, it indicates the time limit for responding to potential dangerous items in the vehicle. If the owner's reply is not received within the corresponding time, the evaluation data of the receiving status of the potentially dangerous items in the owner's car will be output. In this case, the vehicle owner's vehicle information receiving status evaluation data of the potential dangerous goods in the vehicle is received. Push feedback to the vehicle safety monitoring results emergency platform; When the car owner replies with characteristic text data When there are characters, it indicates the time limit for responding to potential dangerous items in the vehicle If the owner's reply is received within the corresponding time, the evaluation data of the receiving status of the potentially dangerous items in the owner's car will be output. It is in receiving state; at this time, the in-vehicle safety monitoring operation is directly ended.

[0013] An AI-based in-vehicle safety monitoring system, used to implement the AI-based in-vehicle safety monitoring method, the system includes a vehicle safety monitoring authorization identification module, a vehicle safety monitoring analysis module, and a vehicle safety monitoring result feedback processing module; The vehicle safety monitoring authorization identification module includes a vehicle driver authorization information collection unit, a vehicle safety monitoring authorization keyword storage unit, and a vehicle safety monitoring authorization identification unit; The vehicle driver authorization information collection unit collects vehicle driver authorization text data through the vehicle terminal; the vehicle safety monitoring authorization keyword storage unit is used to store vehicle safety monitoring authorization keyword text data; the vehicle safety monitoring authorization identification unit performs vehicle safety monitoring authorization identification processing based on the vehicle driver authorization text data and the vehicle safety monitoring authorization keyword text data to generate driver vehicle safety monitoring authorization identification data; The vehicle safety monitoring and analysis module includes a real-time status image acquisition unit in the vehicle, a potential dangerous item image storage unit in the vehicle, a potential dangerous item analysis unit in the vehicle, a vehicle owner contact information storage unit, and a potential dangerous item analysis result feedback unit in the vehicle; The in-vehicle real-time state image acquisition unit is used to collect in-vehicle real-time state image data through a shooting lens; the in-vehicle potential dangerous goods image storage unit is used to store in-vehicle potential dangerous goods image data; the in-vehicle potential dangerous goods analysis unit is used to perform in-vehicle potential dangerous goods analysis and processing based on the in-vehicle real-time state image data and the in-vehicle potential dangerous goods image data based on artificial intelligence to generate in-vehicle potential dangerous goods analysis data; the owner contact information storage unit is used to store owner contact feature data; the in-vehicle potential dangerous goods analysis result feedback unit is used to push and feedback information on in-vehicle potential dangerous goods based on the owner contact feature data and the in-vehicle potential dangerous goods analysis data; The vehicle safety monitoring result feedback processing module includes a vehicle potential dangerous goods feedback result timing unit, a vehicle owner reply information collection unit, a vehicle owner potential dangerous goods information receiving status analysis unit, and a vehicle safety monitoring result emergency feedback unit; The timing unit for feedback results of potential dangerous items in the vehicle performs numerical processing on the timing of the owner's reply information based on the time limit data for feedback responses of potential dangerous items in the vehicle; the owner's reply information collection unit collects the owner's reply feature text data through the vehicle-mounted terminal; the owner's potential dangerous items information reception status analysis unit performs analysis and processing on the owner's reply feature text data to generate evaluation data on the owner's reception status of potential dangerous items information in the vehicle; the vehicle safety monitoring result emergency feedback unit pushes and feeds back the owner's potential dangerous items information reception status evaluation data to the vehicle safety monitoring result emergency platform.

[0014] (III) Beneficial effects The present invention provides an AI-based in-vehicle safety monitoring system and method, which has the following beneficial effects: 1. Accurately collect the authorization information of vehicle drivers through the use of vehicle-mounted terminal timing, and provide data support for the subsequent efficient identification of vehicle monitoring driver authorization information; standard preset vehicle safety monitoring authorization keyword text parameters are combined with intelligent search algorithms to perform character matching with vehicle driver authorization information, so as to achieve scientific and efficient identification of vehicle monitoring driver authorization information, and realize the intelligent and humanized in-vehicle safety monitoring operations.

[0015] 2. The real-time status image parameters of the vehicle are dynamically acquired through the shooting lens to provide real data support for the accurate monitoring of potential dangerous items in the vehicle; the image parameters of potential dangerous items in the vehicle are scientifically preset and combined with artificial intelligence algorithms to match image features with the real-time status image parameters in the vehicle, so as to realize intelligent and accurate analysis of potential dangerous items in the vehicle, and improve the intelligence and safety of the vehicle-mounted monitoring system; based on the vehicle owner's contact feature parameters, the information of potential dangerous items in the vehicle is efficiently pushed and fed back to the vehicle owner, which improves the safety monitoring service quality and efficiency of the vehicle-mounted monitoring system.

[0016] 3. Through scientifically setting the time limit parameters for feedback and reply of potential dangerous items in the car, the owner's reply information is safely timed and counted, and the on-board terminal is used to accurately receive the owner's reply information, so as to realize the time-limited collection of the owner's reply information; the reception status of the owner's potential dangerous items information is analyzed and processed based on the owner's reply information, so as to realize the accurate assessment of the owner's reception status of potential dangerous items information and realize the functional diversity of the on-board monitoring system; the evaluation parameters of the owner's reception status of potential dangerous items information in the car are pushed and fed back to the vehicle safety monitoring result emergency platform, so as to realize the scientific emergency handling of potential dangerous items information when the owner has not received the information of potential dangerous items in the car, and improve the monitoring quality of the on-board monitoring system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a module of an AI-based in-vehicle safety monitoring system provided by the present invention; Figure 2 A flow chart of an AI-based in-vehicle safety monitoring method provided by the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] The embodiments of the AI-based in-vehicle safety monitoring system and method are as follows: Embodiment 1: See also Figure 1 - Figure 2 , an AI-based in-vehicle safety monitoring method, the method comprising the following steps: S1. Collecting the vehicle driver's authorization text data; S2. Perform vehicle safety monitoring authorization identification processing based on the vehicle driver authorization text data and the vehicle safety monitoring authorization keyword text data to generate the driver vehicle safety monitoring authorization identification data; if it fails, directly end the in-vehicle safety monitoring operation; S3, when the vehicle passes, collect real-time status image data inside the vehicle; S4, analyzing and processing the potential dangerous items in the vehicle according to the real-time status image data in the vehicle and the image data of the potential dangerous items in the vehicle, generating analysis data of the potential dangerous items in the vehicle, and directly ending the in-vehicle safety monitoring operation when the potential dangerous items do not exist; S5. When it exists, push and feedback the information of the potential dangerous items in the vehicle based on the vehicle owner contact feature data and the analysis data of the potential dangerous items in the vehicle; S6. Performing timing collection and processing of the vehicle owner's reply information based on the time limit data of the reply to the feedback of the potential dangerous items in the vehicle, and generating the vehicle owner's reply feature text data; S7. Analyze and process the vehicle owner's potentially dangerous items information reception status based on the vehicle owner's reply feature text data to generate vehicle owner's vehicle potential dangerous items information reception status evaluation data; when it is in the reception status, directly terminate the vehicle safety monitoring operation; when it is in the non-reception status, push the vehicle owner's vehicle potential dangerous items information reception status evaluation data to the vehicle safety monitoring result emergency platform.

[0020] For further information, see Figure 1 - Figure 2 , the steps for collecting the vehicle driver's authorization text data are as follows: S11. Collecting the driver's authorization text information for performing in-vehicle safety monitoring operations online through the vehicle terminal and generating vehicle driver authorization text data .

[0021] The vehicle safety monitoring authorization identification processing is performed based on the vehicle driver authorization text data and the vehicle safety monitoring authorization keyword text data to generate the driver vehicle safety monitoring authorization identification data; if it fails, the operation steps of directly ending the in-vehicle safety monitoring operation are as follows: S21. Establishing a vehicle safety monitoring authorization keyword text data set , ;in Indicates Vehicle safety monitoring authorization keyword text data, Indicates the maximum value of the vehicle safety monitoring authorization keyword text data quantity, the vehicle safety monitoring authorization keyword text data includes pass, yes, agree, allow and confirm; S22, using the KD tree nearest neighbor search algorithm to obtain the vehicle driver authorization text data Vehicle safety monitoring authorization keyword text data collection Vehicle safety monitoring authorization keyword text data Perform authorization text character matching and generate driver vehicle safety monitoring authorization identification data based on the authorization text character matching results ; when and If the authorization text character match is successful, it means that the vehicle driver agrees to authorize the execution of the in-vehicle safety monitoring operation, and the driver's vehicle safety monitoring authorization identification data is output. for passing; when and If the authorization text character match is successful, it means that the vehicle driver agrees to authorize the execution of the in-vehicle safety monitoring operation, and the driver's vehicle safety monitoring authorization identification data is output. If the test result fails, the in-vehicle safety monitoring operation is terminated directly.

[0022] Through the vehicle driver authorization information collection unit, the vehicle-mounted terminal timing is used to accurately collect the vehicle driver authorization information, providing data support for the subsequent efficient identification of the vehicle monitoring driver authorization information; the vehicle safety monitoring authorization keyword storage unit and the vehicle safety monitoring authorization identification unit cooperate with each other, and the standard preset vehicle safety monitoring authorization keyword text parameters are combined with the intelligent search algorithm to match the vehicle driver authorization information with characters, so as to realize scientific and efficient identification of the vehicle monitoring driver authorization information and realize the intelligence and humanization of the in-vehicle safety monitoring operation.

[0023] For further information, see Figure 1 - Figure 2 When it is passing, the operation steps for collecting real-time status image data inside the vehicle are as follows: S31, when the driver's vehicle safety monitoring authorization identification data When the vehicle passes through, the real-time status image of the vehicle interior is collected online through the camera lens and the real-time status image data of the vehicle interior is generated. .

[0024] According to the real-time status image data and the image data of the potential dangerous items in the vehicle, the potential dangerous items in the vehicle are analyzed and processed to generate the analysis data of the potential dangerous items in the vehicle. If the potential dangerous items in the vehicle do not exist, the operation steps of directly ending the vehicle safety monitoring operation are as follows: S41. Establishing a data set of images of potentially dangerous objects in the vehicle , ;in Indicates image data of potentially dangerous objects in the car, Indicates the maximum number of types of potentially dangerous items in the car, including bottled water, canned spray, mobile power, lighters, perfume and glasses in direct sunlight; S42, real-time status image data in the vehicle Image data collection of potentially dangerous objects in the car Image data of potentially dangerous items in vehicles Perform image feature matching and generate analysis data of potential dangerous items in the vehicle based on the image feature matching results , execute and generate analysis data of potential dangerous goods in the car The specific steps are as follows: S421. Initialize parameters, update the number of crow populations for dangerous object identification, maximum number of iterations, and flight distance ; S422, initializing the dangerous object recognition crow individual potential dangerous object image data set in the car The initial position and memory in the search space, The dangerous object recognition crows are randomly distributed in a multidimensional search space, namely The random distribution space dimension of dangerous object recognition crows is A collection of potentially dangerous objects image data in vehicles In the search space; in the first iteration, it is assumed that the dangerous object recognition crow individual hides the food in the initial position; that is, in the first iteration, it is assumed that the dangerous object recognition crow individual hides the real-time status image data in the car Image data collection of potentially dangerous objects in the car Image data of potentially dangerous objects in the car at the initial position in the search space Make a match; S423, calculate the individual fitness value of each dangerous object recognition crow, that is, calculate the real-time status image data of the vehicle Image data collection of potentially dangerous objects in the car Image data of potentially dangerous objects in the car in the search space The fitness value of S424, update the dangerous object identification crow individual potential dangerous object image data set in the car The position in the search space, the formula for updating the position of the individual crow for dangerous object identification is as follows: ,in Indicates Iteration 1 Dangerous Object Identification Crow Individual A collection of image data of potentially dangerous objects in vehicles The new position in the search space, Indicates Iteration 1 Dangerous Object Identification Crow Individual A collection of image data of potentially dangerous objects in vehicles The position in the search space, represents a random number uniformly distributed between

[01] , Indicates Iteration 1 Dangerous Object Identification Crow Individual A collection of image data of potentially dangerous objects in vehicles The flight distance in the search space, Indicates Iteration 1 Dangerous Object Identification Crow Individual A collection of image data of potentially dangerous objects in vehicles The food hiding location in the search space, i.e. Iteration 1 Dangerous Object Identification Crow Individual Image dataset of potentially dangerous objects in vehicles Search the search space to find the image data related to the real-time status of the vehicle and image data of potentially dangerous items in the car Matching positions; S425, determine the feasibility of the new position, determine the feasibility of the new position of each dangerous object identification crow individual; if the new position of the dangerous object identification crow individual is feasible, the dangerous object identification crow individual will update its position, that is, in the potential dangerous object image data set in the vehicle Search for real-time status image data in the car Matching image data of potentially dangerous items in the car , the dangerous object recognition crow individual is updated to the successfully matched image data of potential dangerous objects in the car Otherwise, the dangerous object recognition crow individual stays at the current location and does not move to the new location; S426, evaluate the fitness value of the new position, calculate the fitness value of each dangerous object recognition crow's individual new position, that is, calculate the potential dangerous object image data set in the car Real-time status image data in the car in the search space Image data of potentially dangerous objects in the vehicle at new locations The fitness value of S427, update memory, if the fitness value of the new position of the dangerous object recognition crow is greater than the fitness value of the initial position in the memory, the dangerous object recognition crow updates its memory through the new position, otherwise it does not update its memory; that is, in the potential dangerous object image data set in the car Search for real-time status image data in the car Image data of potentially dangerous items in the car with the largest fitness value ; S428: When the maximum number of iterations is met, output the real-time status image data inside the vehicle Image data of potentially dangerous objects in the car Image feature matching results; S429: Based on the real-time status image data in the vehicle in step S428 Image data of potentially dangerous objects in the car The image feature matching results generate analysis data of potential dangerous objects in the car ; when and If the image feature matching is successful, it means there are potentially dangerous items inside the vehicle, and the analysis data of potentially dangerous items inside the vehicle will be output. At the same time, it outputs the image data of potential dangerous objects in the car. Corresponding data on the types of potentially dangerous items in the vehicle; when and If the image feature matching is successful, it means that there are no potentially dangerous items in the vehicle, and the analysis data of potentially dangerous items in the vehicle will be output. does not exist; at this time, the in-vehicle safety monitoring operation is terminated directly.

[0025] When it exists, the operation steps of pushing and feedback of information on potential dangerous items in the vehicle based on the vehicle owner contact feature data and the analysis data of potential dangerous items in the vehicle are as follows: S51. Establishing vehicle owner contact characteristic data , the car owner’s contact characteristic data includes any one of a mobile phone number, a WeChat account, or a QQ account; S52, when the potential dangerous goods in the car are analyzed When there is a potential dangerous object in the car, the image data The corresponding data on the types of potentially dangerous items in the car is obtained through the car owner's contact feature data The corresponding vehicle owner's contact information is combined with the on-board terminal to push and feedback information on potential dangerous items in the vehicle in real time online.

[0026] Through the in-vehicle real-time status image acquisition unit, the real-time status image parameters of the in-vehicle are dynamically acquired using a shooting lens, providing real data support for the accurate monitoring of potential dangerous items in the vehicle; the in-vehicle potential dangerous items image storage unit and the in-vehicle potential dangerous items analysis unit cooperate with each other, scientifically preset the in-vehicle potential dangerous items image parameters and combine the artificial intelligence algorithm with the in-vehicle real-time status image parameters to perform image feature matching, so as to realize intelligent and accurate analysis of potential dangerous items in the vehicle, thereby improving the intelligence and safety of the on-board monitoring system; the owner contact information storage unit and the in-vehicle potential dangerous items analysis result feedback unit cooperate with each other, and based on the owner contact feature parameters, the information of potential dangerous items in the vehicle is efficiently pushed and fed back to the owner, thereby improving the safety monitoring service quality and efficiency of the on-board monitoring system.

[0027] For further information, see Figure 1 - Figure 2 , the operation steps for collecting and processing the owner's reply information timing according to the reply time limit data of the potential dangerous items in the car, and generating the owner's reply feature text data are as follows: S61. Establishing the time limit for responding to feedback on potentially dangerous items in the vehicle The data on the time limit for responding to feedback on potentially dangerous items in the vehicle indicates the preset time limit for the vehicle owner to respond to the information after the feedback operation on the potentially dangerous items in the vehicle is completed. The unit is minutes; S62: When the push feedback operation of the information of the potentially dangerous items in the vehicle is completed, the time limit data of the feedback reply based on the potentially dangerous items in the vehicle Process the time value of the owner's reply information, and at the same time, use the vehicle terminal to feedback the time limit data of the potential dangerous items in the car Collect the owner's reply information online within the corresponding time length, and generate the owner's reply feature text data .

[0028] The vehicle owner's reply feature text data is analyzed and processed for the vehicle owner's potential dangerous goods information reception status, and the vehicle owner's in-vehicle potential dangerous goods information reception status evaluation data is generated; when it is in the reception status, the in-vehicle safety monitoring operation is directly terminated; when it is in the non-reception status, the vehicle owner's in-vehicle potential dangerous goods information reception status evaluation data is pushed and fed back to the vehicle safety monitoring result emergency platform. The operation steps are as follows: S71. Replying to the vehicle owner with characteristic text data Perform character analysis on the car owner's reply information, and generate evaluation data on the receiving status of the car owner's potentially dangerous items information based on the character analysis results of the car owner's reply information ; When the owner replies with feature text data When it is empty, it indicates the time limit for responding to potential dangerous items in the vehicle. If the owner's reply is not received within the corresponding time, the evaluation data of the receiving status of the potentially dangerous items in the owner's car will be output. In this case, the owner will receive the information of potential dangerous goods in the car. Push feedback to the vehicle safety monitoring results emergency platform; When the owner replies with feature text data When there are characters, it indicates the time limit for responding to potential dangerous items in the vehicle. If the owner's reply is received within the corresponding time, the evaluation data of the receiving status of the potentially dangerous items in the owner's car will be output. It is in receiving state; at this time, the in-vehicle safety monitoring operation is directly ended.

[0029] Through the cooperation between the timing unit for the feedback result of potential dangerous goods in the vehicle and the vehicle owner's reply information collection unit, the time limit parameters for the feedback reply of potential dangerous goods in the vehicle are scientifically set to conduct the safety timing statistics of the vehicle owner's reply information, and the vehicle-mounted terminal is used to accurately receive the vehicle owner's reply information to realize the time-limited collection of the vehicle owner's reply information; the vehicle owner's potential dangerous goods information reception status analysis unit analyzes and processes the vehicle owner's potential dangerous goods information reception status based on the vehicle owner's reply information, realizes accurate assessment of the vehicle owner's potential dangerous goods information reception status, and realizes the functional diversity of the vehicle-mounted monitoring system; the vehicle safety monitoring result emergency feedback unit pushes and feeds back the vehicle owner's potential dangerous goods information reception status assessment parameters to the vehicle safety monitoring result emergency platform, realizes scientific emergency processing of potential dangerous goods information when the vehicle owner has not received the potential dangerous goods information in the vehicle, and improves the monitoring quality of the vehicle-mounted monitoring system.

[0030] Embodiment 2: See also Figure 1 - Figure 2 , an AI-based in-vehicle safety monitoring system, used to implement an AI-based in-vehicle safety monitoring method, the system includes a vehicle safety monitoring authorization identification module, a vehicle safety monitoring analysis module, and a vehicle safety monitoring result feedback processing module; The vehicle safety monitoring authorization identification module includes a vehicle driver authorization information collection unit, a vehicle safety monitoring authorization keyword storage unit, and a vehicle safety monitoring authorization identification unit; The vehicle driver authorization information collection unit collects the vehicle driver authorization text data through the vehicle terminal; the vehicle safety monitoring authorization keyword storage unit is used to store the vehicle safety monitoring authorization keyword text data; the vehicle safety monitoring authorization identification unit performs vehicle safety monitoring authorization identification processing based on the vehicle driver authorization text data and the vehicle safety monitoring authorization keyword text data to generate the driver vehicle safety monitoring authorization identification data; The vehicle safety monitoring and analysis module includes a real-time status image acquisition unit in the vehicle, a potential dangerous item image storage unit in the vehicle, a potential dangerous item analysis unit in the vehicle, a vehicle owner contact information storage unit, and a potential dangerous item analysis result feedback unit in the vehicle; The in-vehicle real-time state image acquisition unit is used to collect the in-vehicle real-time state image data through the shooting lens; the in-vehicle potential dangerous goods image storage unit is used to store the in-vehicle potential dangerous goods image data; the in-vehicle potential dangerous goods analysis unit is used to analyze and process the in-vehicle potential dangerous goods based on the in-vehicle real-time state image data and the in-vehicle potential dangerous goods image data based on artificial intelligence, and generate in-vehicle potential dangerous goods analysis data; the owner contact information storage unit is used to store the owner contact feature data; the in-vehicle potential dangerous goods analysis result feedback unit is used to push and feedback the in-vehicle potential dangerous goods information based on the owner contact feature data and the in-vehicle potential dangerous goods analysis data; The vehicle safety monitoring result feedback processing module includes a vehicle potential dangerous goods feedback result timing unit, a vehicle owner reply information collection unit, a vehicle owner potential dangerous goods information receiving status analysis unit, and a vehicle safety monitoring result emergency feedback unit; The timing unit for feedback results of potential dangerous items in the vehicle processes the timing of the owner's reply information based on the time limit data of the feedback reply of potential dangerous items in the vehicle; the owner's reply information collection unit collects the owner's reply feature text data through the vehicle terminal; the owner's potential dangerous items information reception status analysis unit analyzes the owner's reply feature text data and generates the owner's potential dangerous items information reception status evaluation data; the vehicle safety monitoring result emergency feedback unit pushes the owner's potential dangerous items information reception status evaluation data to the vehicle safety monitoring result emergency platform.

[0031] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An AI-based in-vehicle safety monitoring method, characterized in that: The method comprises the following steps: S1. Collecting the vehicle driver's authorization text data; S2, performing vehicle safety monitoring authorization identification processing according to the vehicle driver authorization text data and the vehicle safety monitoring authorization keyword text data, and generating driver vehicle safety monitoring authorization identification data; if it is not passed, directly ending the in-vehicle safety monitoring operation; S3, when the vehicle passes, collect real-time status image data inside the vehicle; S4, analyzing and processing the potential dangerous items in the vehicle according to the real-time status image data in the vehicle and the image data of the potential dangerous items in the vehicle, generating analysis data of the potential dangerous items in the vehicle, and directly ending the vehicle safety monitoring operation when the potential dangerous items do not exist; S5. When it is present, push and feedback the information of the potential dangerous items in the vehicle according to the vehicle owner contact feature data and the analysis data of the potential dangerous items in the vehicle; S6. Performing timing collection and processing of the vehicle owner's reply information based on the time limit data of the reply to the feedback of the potential dangerous items in the vehicle, and generating the vehicle owner's reply feature text data; S7, analyzing and processing the vehicle owner's potentially dangerous goods information reception status on the vehicle owner's reply feature text data, generating vehicle owner's vehicle potential dangerous goods information reception status evaluation data; when the vehicle owner is in the reception status, directly ending the vehicle safety monitoring operation; When it is in the unreceived state, the evaluation data of the receiving state of the information of the potentially dangerous items in the owner's vehicle is pushed and fed back to the vehicle safety monitoring result emergency platform.

2. The AI-based in-vehicle safety monitoring method according to claim 1, characterized in that: The S1 comprises the following steps: S11. Collecting the driver's authorization text information for performing in-vehicle safety monitoring operations online through the vehicle terminal and generating vehicle driver authorization text data .

3. The AI-based in-vehicle safety monitoring method according to claim 2, characterized in that: The S2 comprises the following steps: S21. Establishing a vehicle safety monitoring authorization keyword text data set , ;in Indicates Vehicle safety monitoring authorization keyword text data, Indicates the maximum number of vehicle safety monitoring authorization keyword text data; S22, using the KD tree nearest neighbor search algorithm to search the vehicle driver authorization text data A text data set of keywords related to the vehicle safety monitoring authorization Vehicle safety monitoring authorization keyword text data Perform authorization text character matching and generate driver vehicle safety monitoring authorization identification data based on the authorization text character matching result ; when and If the authorization text characters are matched successfully, the driver's vehicle safety monitoring authorization identification data will be output for passing; when and If the authorization text characters are not matched successfully, the driver vehicle safety monitoring authorization identification data is output If the test result fails, the in-vehicle safety monitoring operation is terminated directly.

4. The AI-based in-vehicle safety monitoring method according to claim 3 is characterized in that: The S3 comprises the following steps: S31, when the driver's vehicle safety monitoring authorization identification data When the vehicle passes through, the real-time status image of the vehicle interior is collected online through the camera lens and the real-time status image data of the vehicle interior is generated. .

5. The AI-based in-vehicle safety monitoring method according to claim 4 is characterized in that: The S4 comprises the following steps: S41. Establishing a data set of images of potentially dangerous objects in the vehicle , ;in Indicates image data of potentially dangerous objects in the car, Indicates the maximum number of types of potentially dangerous items in the vehicle; S42: the real-time status image data in the vehicle and a data set of images of potentially dangerous objects in the vehicle Image data of potentially dangerous items in the vehicle Perform image feature matching and generate analysis data of potential dangerous items in the vehicle based on the image feature matching results , execute and generate analysis data of potential dangerous goods in the car The specific steps are as follows: S421. Initialize parameters, update the number of crow populations for dangerous object identification, maximum number of iterations, and flight distance ; S422, initializing the dangerous object recognition crow individual potential dangerous object image data set in the car The initial position and memory in the search space, The dangerous object recognition crows are randomly distributed in a multidimensional search space, namely The random distribution space dimension of dangerous object recognition crows is A collection of potentially dangerous objects image data in vehicles In the search space; in the first iteration, it is assumed that the dangerous object recognition crow individual hides the food in the initial position; that is, in the first iteration, it is assumed that the dangerous object recognition crow individual hides the real-time state image data in the vehicle Image data collection of potentially dangerous objects in the car Image data of potentially dangerous objects in the car at the initial position in the search space Make a match; S423, calculating the individual fitness value of each dangerous object recognition crow, that is, calculating the real-time status image data of the vehicle Image data collection of potentially dangerous objects in the car Image data of potentially dangerous objects in the car in the search space The fitness value of S424, update the dangerous object identification crow individual potential dangerous object image data set in the car Position in the search space; S425, determine the feasibility of the new position, determine the feasibility of the new position of each dangerous object identification crow individual; if the new position of the dangerous object identification crow individual is feasible, the dangerous object identification crow individual will update its position, that is, in the potential dangerous object image data set in the vehicle Search for the real-time status image data in the vehicle Matching image data of potentially dangerous items in the car , the dangerous object recognition crow individual is updated to the successfully matched image data of potential dangerous objects in the car Otherwise, the dangerous object recognition crow individual stays at the current location and does not move to the new location; S426, evaluate the fitness value of the new position, calculate the fitness value of each dangerous object recognition crow's individual new position, that is, calculate the potential dangerous object image data set in the car The real-time status image data in the vehicle described in the search space Image data of potentially dangerous objects in the vehicle at new locations The fitness value of S427, update memory, if the fitness value of the new position of the dangerous object recognition crow is greater than the fitness value of the initial position in the memory, the dangerous object recognition crow updates its memory through the new position, otherwise it does not update its memory; that is, in the potential dangerous object image data set in the car Search for the real-time status image data in the vehicle Image data of potentially dangerous items in the car with the largest fitness value ; S428: When the maximum number of iterations is met, output the real-time status image data in the vehicle Image data of potentially dangerous objects in the vehicle Image feature matching results; S429: Based on the real-time status image data in the vehicle in step S428 Image data of potentially dangerous objects in the vehicle The image feature matching results generate analysis data of potential dangerous objects in the car ; when and If the image feature matching is successful, the analysis data of potential dangerous items in the vehicle will be output If it exists, the image data of the potential dangerous objects in the vehicle is output simultaneously. Corresponding data on the types of potentially dangerous items in the vehicle; when and If the image features are not matched successfully, the analysis data of potential dangerous items in the vehicle will be output. does not exist; at this time, the in-vehicle safety monitoring operation is terminated directly.

6. The AI-based in-vehicle safety monitoring method according to claim 5, characterized in that: The S5 comprises the following steps: S51. Establishing vehicle owner contact characteristic data ; S52: When the analysis data of the potentially dangerous items in the vehicle is When the potential dangerous object image data in the vehicle is present, The corresponding data on the types of potentially dangerous items in the vehicle is obtained through the vehicle owner contact feature data The corresponding vehicle owner's contact information is combined with the on-board terminal to push and feedback information on potential dangerous items in the vehicle in real time online.

7. The AI-based in-vehicle safety monitoring method according to claim 6, characterized in that: The S6 comprises the following steps: S61. Establishing the time limit for responding to feedback on potentially dangerous items in the vehicle , The unit is minutes; S62: When the operation of pushing and feeding back the information of the potentially dangerous items in the vehicle is completed, the time limit data of the feedback reply based on the potentially dangerous items in the vehicle is Process the time value of the owner's reply information, and at the same time, feedback the time limit data of the potentially dangerous items in the car through the vehicle terminal Collect the owner's reply information online within the corresponding time length, and generate the owner's reply feature text data .

8. The AI-based in-vehicle safety monitoring method according to claim 7, characterized in that: The S7 comprises the following steps: S71: Replying to the vehicle owner with characteristic text data Perform character analysis on the car owner's reply information, and generate evaluation data on the receiving status of the car owner's potentially dangerous items information based on the character analysis results of the car owner's reply information ; When the car owner replies with characteristic text data When it is an empty character, the owner's car's potential dangerous goods information reception status assessment data is output In this case, the vehicle owner's vehicle information receiving status evaluation data of the potential dangerous goods in the vehicle is received. Push feedback to the vehicle safety monitoring results emergency platform; When the car owner replies with characteristic text data If there are characters, the owner's car's potential dangerous goods reception status assessment data is output It is in receiving state; at this time, the in-vehicle safety monitoring operation is directly ended.

9. An AI-based in-vehicle safety monitoring system, used to implement an AI-based in-vehicle safety monitoring method according to any one of claims 1 to 8, characterized in that: The system includes a vehicle safety monitoring authorization identification module, a vehicle safety monitoring analysis module, and a vehicle safety monitoring result feedback processing module.

Images