A method and device for identifying vehicle overcrowding

Abstract

The present application relates to the field of traffic safety technology, and in particular to a method and device for identifying vehicle overcrowding. The present application determines the number of people in each frame of an image corresponding to each driving surveillance video, and intercepts a first driving surveillance video in which an overcrowded scene appears and a second driving surveillance video in which a non-overcrowded scene appears from a driving surveillance video in which the number of people is greater than a first preset number, and uses the image in the first driving surveillance video as an overcrowded sample and the image in the second driving surveillance video as a non-overcrowded sample. By adopting the above method, the samples for training the overcrowded recognition model to be trained can be from the images corresponding to the overcrowded scene and the non-overcrowded scene in the same driving surveillance video corresponding to the same vehicle, that is, the overcrowded recognition model to be trained can learn the scene of the same vehicle more comprehensively, and the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

Description

Technical Field

[0001] The present application relates to the field of traffic safety technology, and in particular to a method and device for identifying an overloaded vehicle. Background Art

[0002] Overloading a vehicle is a serious traffic violation that poses a great threat to the safety of the people in the vehicle, but some people still ignore their own safety and overload or overload the vehicle. In order to better ensure the safety and interests of drivers and passengers, monitoring equipment is usually installed in the vehicle to monitor the people in the vehicle and the driving process in real time, so as to restrain the behavior of drivers and passengers.

[0003] However, manually reviewing surveillance videos and identifying whether the corresponding vehicles in the surveillance videos are overloaded requires a lot of manpower and time costs, and the recognition efficiency is low. In addition, due to the limitation of subjective factors, the accuracy cannot be guaranteed. Therefore, how to mine the data set samples for machine learning to achieve machine recognition of whether the vehicle is overloaded is a technical problem that needs to be solved urgently. Summary of the invention

[0004] In view of this, the purpose of the embodiments of the present application is to provide a method and device for identifying overcrowded vehicles, so that the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained can be derived from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving monitoring video corresponding to the same vehicle, so that the trained overcrowded recognition model can more accurately identify overcrowded vehicles.

[0005] This application mainly includes the following aspects:

[0006] In a first aspect, an embodiment of the present application provides a method for identifying an overloaded vehicle, the method comprising:

[0007] Acquire at least one driving surveillance video, and determine the number of persons in each frame of an image corresponding to each driving surveillance video in the at least one driving surveillance video;

[0008] Acquire a first image in which the number of people is greater than a first preset number, and record a first time period in which an overcrowded scene occurs and a second time period in which a non-overcrowded scene occurs in the driving monitoring video in which the first image is located;

[0009] Intercepting a first vehicle monitoring video within the first time period and a second vehicle monitoring video within the second time period from the vehicle monitoring video where the first image is located;

[0010] The images in the first vehicle monitoring video are used as overcrowded samples, and the images in the second vehicle monitoring video are used as non-overcrowded samples. The overcrowded recognition model to be trained is trained, and the trained overcrowded recognition model is used to identify overcrowded vehicles.

[0011] In a possible implementation, after acquiring the first image in which the number of people is greater than the first preset number, and before recording the first time period in which an overcrowded scene appears and the second time period in which a non-overcrowded scene appears in the driving monitoring video in which the first image is located, the identification method further includes:

[0012] Counting the first number of the first images in each driving monitoring video;

[0013] Determining whether the first number is greater than or equal to a first preset threshold;

[0014] If the first number is greater than or equal to the first preset threshold, the first image of the first preset threshold is selected from the driving monitoring video corresponding to the first number.

[0015] In a possible implementation, after acquiring at least one driving surveillance video and determining the number of persons in each frame of an image corresponding to each driving surveillance video in the at least one driving surveillance video, the recognition method further includes:

[0016] Acquire a second image in which the number of personnel is less than or equal to the first preset number and the number of personnel is greater than or equal to a second preset number;

[0017] The non-overcrowded sample is determined based on the second image.

[0018] In a possible implementation manner, determining the non-overcrowded sample based on the second image includes:

[0019] Counting the second number of the second images in each driving monitoring video;

[0020] Determining whether the second number is greater than or equal to a second preset threshold;

[0021] If the second number is greater than or equal to the second preset threshold, selecting a second image of the second preset threshold from the traffic monitoring video corresponding to the second number as the non-overcrowded sample;

[0022] If the second number is less than the second preset threshold, it is determined that all the second images in the driving monitoring video corresponding to the second number are the non-overcrowded samples.

[0023] In a possible implementation, the number of people in each frame of image is determined according to the following steps:

[0024] Inputting the at least one driving surveillance video into the trained personnel quantity detection model to determine the number of personnel in each frame image corresponding to each driving surveillance video in the at least one driving surveillance video;

[0025] Wherein, the personnel quantity detection model includes a face detection model or a body detection model.

[0026] In a possible implementation manner, the overcrowding recognition model to be trained is trained according to the following steps:

[0027] The overcrowded recognition model to be trained is trained according to the overcrowded samples and the type labels of the overcrowded samples, and the non-overcrowded samples and the type labels of the non-overcrowded samples.

[0028] In a possible implementation, the identification method further includes:

[0029] Obtain the target surveillance video corresponding to the target travel order;

[0030] Based on the target monitoring video and the trained overcrowded recognition model, it is determined whether the vehicle corresponding to the target travel order is overcrowded.

[0031] In a possible implementation manner, obtaining a target monitoring video corresponding to a target travel order includes:

[0032] Obtaining order information of the target travel order;

[0033] After the travel time in the order information, the target monitoring video is acquired in real time.

[0034] In a possible implementation manner, determining whether the vehicle corresponding to the target travel order is overcrowded based on the target monitoring video and the trained overcrowded identification model includes:

[0035] Determining, according to the target surveillance video, a trained overcrowding recognition model that matches the vehicle model in the target surveillance video;

[0036] Divide the target surveillance video into frames to obtain multiple frame images;

[0037] The multiple frames of images are input into the trained overcrowded recognition model in sequence to determine whether the vehicle is overcrowded.

[0038] In a possible implementation manner, after determining that the vehicle is overloaded, the identification method further includes:

[0039] An overcrowding warning message is sent to the service provider and service requester corresponding to the target travel order.

[0040] In a second aspect, an embodiment of the present application further provides a vehicle overload identification device, the identification device comprising:

[0041] A first determination module is used to obtain at least one driving monitoring video and determine the number of people in each frame image corresponding to each driving monitoring video in the at least one driving monitoring video;

[0042] A recording module, used for acquiring a first image in which the number of people is greater than a first preset number, and recording a first time period in which an overcrowded scene occurs and a second time period in which a non-overcrowded scene occurs in the driving monitoring video in which the first image is located;

[0043] An interception module, used to intercept a first vehicle monitoring video within the first time period and a second vehicle monitoring video within the second time period from the vehicle monitoring video where the first image is located;

[0044] The training module is used to use the images in the first vehicle monitoring video as overcrowded samples and the images in the second vehicle monitoring video as non-overcrowded samples to train the overcrowded recognition model to be trained. The trained overcrowded recognition model is used to identify overcrowded vehicles in motion.

[0045] In a possible implementation, the recording module includes:

[0046] A counting unit, used for counting a first number of the first images in each driving monitoring video;

[0047] A judging unit, configured to judge whether the first number is greater than or equal to a first preset threshold;

[0048] A selection unit is used to select the first image of the first preset threshold from the driving monitoring video corresponding to the first number if the first number is greater than or equal to the first preset threshold.

[0049] In a possible implementation manner, the identification device further includes:

[0050] A first acquisition module, used for acquiring a second image in which the number of personnel is less than or equal to the first preset number and the number of personnel is greater than or equal to a second preset number;

[0051] The second determination module is used to determine the non-overcrowded sample based on the second image.

[0052] In a possible implementation manner, the second determination module is used to determine the non-overcrowded sample according to the following steps:

[0053] Counting the second number of the second images in each driving monitoring video;

[0054] Determining whether the second number is greater than or equal to a second preset threshold;

[0055] If the second number is greater than or equal to the second preset threshold, selecting a second image of the second preset threshold from the traffic monitoring video corresponding to the second number as the non-overcrowded sample;

[0056] If the second number is less than the second preset threshold, it is determined that all the second images in the driving monitoring video corresponding to the second number are the non-overcrowded samples.

[0057] In a possible implementation manner, the first determining module is used to determine the number of people in each frame of image according to the following steps:

[0058] Inputting the at least one driving surveillance video into the trained personnel quantity detection model to determine the number of personnel in each frame image corresponding to each driving surveillance video in the at least one driving surveillance video;

[0059] Wherein, the personnel quantity detection model includes a face detection model or a body detection model.

[0060] In a possible implementation manner, the training module is used to train the overcrowding recognition model to be trained according to the following steps:

[0061] The overcrowded recognition model to be trained is trained according to the overcrowded samples and the type labels of the overcrowded samples, and the non-overcrowded samples and the type labels of the non-overcrowded samples.

[0062] In a possible implementation manner, the identification device further includes:

[0063] The second acquisition module is used to acquire the target monitoring video corresponding to the target travel order;

[0064] The third determination module is used to determine whether the vehicle corresponding to the target travel order is overcrowded based on the target monitoring video and the trained overcrowding recognition model.

[0065] In a possible implementation manner, the second acquisition module is used to acquire the target monitoring video corresponding to the target travel order according to the following steps:

[0066] Obtaining order information of the target travel order;

[0067] After the travel time in the order information, the target monitoring video is acquired in real time.

[0068] In a possible implementation manner, the third determination module is used to determine whether the vehicle corresponding to the target travel order is overcrowded according to the following steps:

[0069] Determining, according to the target surveillance video, a trained overcrowding recognition model that matches the vehicle model in the target surveillance video;

[0070] Divide the target surveillance video into frames to obtain multiple frame images;

[0071] The multiple frames of images are input into the trained overcrowded recognition model in sequence to determine whether the vehicle is overcrowded.

[0072] In a possible implementation manner, the identification device further includes:

[0073] The sending module is used to send overcrowding warning information to the service provider and service requester corresponding to the target travel order.

[0074] In a third aspect, an embodiment of the present application further provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the method for identifying an overloaded vehicle as described in the first aspect or any possible implementation manner of the first aspect.

[0075] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for identifying an overloaded vehicle as described in the first aspect or any possible implementation manner of the first aspect are executed.

[0076] In an embodiment of the present application, by determining the number of persons in each frame image corresponding to each acquired driving surveillance video, and intercepting the first driving surveillance video within a first time period in which an overcrowded scene occurs, and the second driving surveillance video within a second time period in which a non-overcrowded scene occurs from the driving surveillance video where the first image in which the number of persons is greater than a first preset number is located, further, using the image in the first driving surveillance video as an overcrowded sample, and using the image in the second driving surveillance video as a non-overcrowded sample, the above-mentioned method of obtaining overcrowded samples and non-overcrowded samples can make the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained come from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving surveillance video corresponding to the same vehicle, that is, the overcrowded recognition model to be trained learns the scenes of the same vehicle more comprehensively, and thus, the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

[0077] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0079] Figure 1 A flow chart of a method for identifying an overloaded vehicle provided in the first embodiment of the present application is shown;

[0080] Figure 2 A flow chart showing another method for identifying an overloaded vehicle provided in the second embodiment of the present application is shown;

[0081] Figure 3 One of the functional module diagrams of a vehicle overcrowding identification device provided in the third embodiment of the present application is shown;

[0082] Figure 4 The second functional module diagram of a vehicle overload identification device provided in the third embodiment of the present application is shown;

[0083] Figure 5 A schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present application is shown. DETAILED DESCRIPTION

[0084] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art, under the guidance of the content of the present application, can add one or more other operations to the flowchart, or remove one or more operations from the flowchart.

[0085] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0086] In order to enable those skilled in the art to use the contents of this application, the following implementation methods are given in combination with the specific application scenario of "identification of vehicle overcrowding". For those skilled in the art, the general principles defined here can be applied to other embodiments and application scenarios without departing from the spirit and scope of this application.

[0087] It should be noted that the terms "service provider" and "driver" in this application can be used interchangeably, "service requester" and "passenger" can be used interchangeably, "service providing end" and "driver terminal" can be used interchangeably, and "service requesting end" and "passenger terminal" can be used interchangeably.

[0088] The following methods, devices, electronic devices or computer-readable storage media of the embodiments of the present application can be applied to any scenario where vehicle overcrowding identification is required. The embodiments of the present application are not limited to specific application scenarios. Any scheme using the vehicle overcrowding identification method and device provided by the embodiments of the present application is within the protection scope of the present application.

[0089] It is worth noting that before the present application was submitted, the existing solutions usually required manual review of surveillance videos and identification of whether the vehicle corresponding to the surveillance video was overcrowded, which required huge manpower and time costs, had low recognition efficiency, and due to the limitations of subjective factors, accuracy could not be guaranteed.

[0090] Therefore, how to mine data set samples for machine learning and train an overcrowded identification model through the data set samples so as to identify whether the vehicle is overcrowded through the overcrowded identification model is a technical problem that needs to be solved urgently.

[0091] In response to the above problems, an embodiment of the present application determines the number of people in each frame image corresponding to each acquired driving surveillance video, and intercepts the first driving surveillance video within a first time period where an overcrowded scene occurs, and the second driving surveillance video within a second time period where a non-overcrowded scene occurs, from the driving surveillance video where the first image in which the number of people is greater than a first preset number. Furthermore, the image in the first driving surveillance video is used as an overcrowded sample, and the image in the second driving surveillance video is used as a non-overcrowded sample. By adopting the above-mentioned method of obtaining overcrowded samples and non-overcrowded samples, the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained can be from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving surveillance video corresponding to the same vehicle, that is, the overcrowded recognition model to be trained learns the scenes of the same vehicle more comprehensively. In this way, the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

[0092] To facilitate the understanding of the present application, the technical solution provided by the present application is described in detail below in conjunction with specific embodiments.

[0093] Embodiment 1

[0094] Figure 1 This is a flow chart of a method for identifying an overloaded vehicle provided in the first embodiment of the present application. The device for executing the method for identifying an overloaded vehicle may be a cloud platform or a server. The following describes the method for identifying an overloaded vehicle provided in the first embodiment of the present application from the perspective of the server as the execution subject. Figure 1 As shown, the vehicle overload identification method provided in the first embodiment of the present application includes the following steps:

[0095] S101: Acquire at least one driving surveillance video, and determine the number of persons in each frame of an image corresponding to each driving surveillance video in the at least one driving surveillance video.

[0096] In a specific implementation, in the process of obtaining overcrowded samples and non-overcrowded samples, first, at least one driving surveillance video is obtained. Here, the driving surveillance video can be obtained from historical travel orders. Usually, one historical travel order corresponds to a driving surveillance video. After obtaining at least one driving surveillance video, the number of people in each frame image corresponding to each driving surveillance video in the at least one driving surveillance video is determined respectively. Specifically, each driving surveillance video can be framed to obtain multiple frames of images corresponding to each driving surveillance video, and the number of people in each frame image can be determined respectively. Here, each frame image can be input into a trained personnel number detection model to determine the number of people in the frame image, or the number of people in the frame image can be determined by other recognition technologies. The personnel number detection model can be a face detection model or other model that can detect the number of people in an image.

[0097] It should be noted that overcrowded samples refer to images with overcrowded features in driving surveillance videos, and non-overcrowded samples refer to images with non-overcrowded features in driving surveillance videos. Here, images with overcrowded features may be images in which the number of people in the image is greater than the number of people that the vehicle corresponding to the driving surveillance video where the image is located can normally carry, wherein the number of people that a vehicle can normally carry is related to the model of the vehicle, for example, the normal number of people that a 5-seater car can carry is 5.

[0098] S102: Acquire a first image in which the number of people is greater than a first preset number, and record a first time period in which an overcrowded scene occurs and a second time period in which a non-overcrowded scene occurs in the driving monitoring video in which the first image is located.

[0099] In a specific implementation, after respectively determining the number of people in each frame image corresponding to each driving surveillance video, respectively compare the number of people in each frame image with a first preset number, where the first preset number is the number of people normally carried by the vehicle, and obtain a first image in which the number of people is greater than the first preset number, that is, obtain an image with an overcrowded feature. After obtaining the first image with an overcrowded feature, determine the driving surveillance video where the first image is located, obtain the driving surveillance video where the first image is located, and record a first time period in which an overcrowded scene appears and a second time period in which a non-overcrowded scene appears in the driving surveillance video where the first image is located. Here, the first time period and the second time period can be determined by inputting the driving surveillance video where the first image is located into a trained personnel number detection model.

[0100] It should be noted that each driving surveillance video may correspond to one or more first images, or there may be no first image, that is, each driving surveillance video may or may not contain an overcrowded scene. Here, the driving surveillance video where the first image is located may contain videos of time periods with non-overcrowded scenes, or there may be no videos of time periods with non-overcrowded scenes, that is, the driving surveillance video where the first image is located may only contain videos of overcrowded scenes.

[0101] Here, the first preset number is the number of people that the vehicle normally carries, and the first preset number can be set in advance according to the vehicle model, such as a 5-seater model, a 6-seater model, a 7-seater model, and so on.

[0102] S103: intercepting a first driving surveillance video within the first time period and a second driving surveillance video within the second time period from the driving surveillance video where the first image is located.

[0103] In a specific implementation, after obtaining a first image in which the number of people is greater than a first preset number from at least one driving surveillance video, the first driving surveillance video in the first time period where an overcrowded scene occurs is captured from the driving surveillance video where the first image is located, based on the start and end times of the first time period and the start and end times of the second time period, and the second driving surveillance video in the second time period where a non-overcrowded scene occurs is captured from the driving surveillance video where the first image is located. In this way, videos of the time periods corresponding to the overcrowded scenes and the non-overcrowded scenes in the same driving surveillance video are obtained.

[0104] S104: Using the images in the first vehicle traffic monitoring video as overcrowded samples and using the images in the second vehicle traffic monitoring video as non-overcrowded samples, training the overcrowded recognition model to be trained, and using the trained overcrowded recognition model to recognize overcrowded vehicles.

[0105] In a specific implementation, the first driving monitoring video and the second driving monitoring video are videos corresponding to different time periods in the driving monitoring video where the same first image is located, and the image in the first driving monitoring video corresponding to the overcrowded scene is used as the overcrowded sample, and the image in the second driving monitoring video corresponding to the non-overcrowded scene is used as the non-overcrowded sample. In this way, the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained can be from the images corresponding to the overcrowded scene and the non-overcrowded scene in the same driving monitoring video corresponding to the same vehicle. In the process of training the overcrowded recognition model to be trained, the overcrowded recognition model to be trained can simultaneously learn the images of the same vehicle in the overcrowded scene and the non-overcrowded scene, that is, the overcrowded recognition model to be trained learns the scenes of the same vehicle more comprehensively, and when identifying whether the vehicle is overcrowded, it can accurately and quickly identify whether the vehicle is overcrowded. By adopting the above-mentioned method of obtaining overcrowded samples and non-overcrowded samples, the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

[0106] It should be noted that a small number of images can be screened out in the first traffic surveillance video as overcrowded samples, and a small number of images can be screened out in the second traffic surveillance video as non-overcrowded samples, instead of taking all the images in the first traffic surveillance video and the second traffic surveillance video as samples. In this way, the number of unnecessary repeated training processes in the training process of the overcrowded recognition model to be trained can be reduced, and the efficiency of the overcrowded recognition model to be trained can be improved while ensuring the demand for overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained.

[0107] In an embodiment of the present application, by determining the number of persons in each frame image corresponding to each acquired driving surveillance video, and intercepting the first driving surveillance video within a first time period in which an overcrowded scene occurs, and the second driving surveillance video within a second time period in which a non-overcrowded scene occurs from the driving surveillance video where the first image in which the number of persons is greater than a first preset number is located, further, using the image in the first driving surveillance video as an overcrowded sample, and using the image in the second driving surveillance video as a non-overcrowded sample, the above-mentioned method of obtaining overcrowded samples and non-overcrowded samples can make the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained come from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving surveillance video corresponding to the same vehicle, that is, the overcrowded recognition model to be trained learns the scenes of the same vehicle more comprehensively, and thus, the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

[0108] In a possible implementation, after obtaining the first image in which the number of people is greater than the first preset number in step 102, and before recording in step 102 a first time period in which an overcrowded scene occurs and a second time period in which a non-overcrowded scene occurs in the driving monitoring video in which the first image is located, the method for identifying an overcrowded vehicle further includes the following steps:

[0109] Step 1021: Count the first number of the first images in each driving monitoring video.

[0110] In a specific implementation, after obtaining the first image in which the number of people in each frame image corresponding to each driving surveillance video is greater than the first preset number, the first number of first images in which the number of people in each driving surveillance video is greater than the first preset number is counted respectively, that is, the number of images in the overcrowded scene in each driving surveillance video is counted.

[0111] Step 1022: Determine whether the first number is greater than or equal to a first preset threshold.

[0112] In a specific implementation, after counting the first number of first images in each driving surveillance video in which the number of people is greater than the first preset number, that is, after counting the number of images in overcrowded scenes in each driving surveillance video, it is determined whether the number of images with the first image in each driving surveillance video is greater than or equal to the first preset threshold.

[0113] Step 1023: If the first number is greater than or equal to the first preset threshold, select the first image of the first preset threshold from the driving monitoring video corresponding to the first number.

[0114] In a specific implementation, if the first number of first images in which the number of people in a certain driving surveillance video is greater than a first preset number is greater than or equal to a first preset threshold, that is, the number of first images with overcrowding characteristics in the driving surveillance video is relatively large, only first images with a first preset threshold can be selected from the first images with the first number as the screened first images, that is, only images with overcrowding characteristics of the first preset threshold are obtained in each driving surveillance video. In this way, the number of repeated training processes that occur during the training of the overcrowding recognition model to be trained due to the existence of too many first images of the same type from the same driving surveillance video can be reduced, and the efficiency of the overcrowding recognition model to be trained can be improved while ensuring the demand for overcrowding samples and non-overcrowding samples for training the overcrowding recognition model to be trained.

[0115] Here, the first preset threshold can be set according to actual needs. The first preset threshold is used to limit the number of first images screened out from each driving monitoring video. The first preset threshold can usually be set to 3.

[0116] In a possible implementation, after obtaining at least one driving surveillance video in step 101 and determining the number of persons in each frame of an image corresponding to each driving surveillance video in the at least one driving surveillance video, the method for identifying an overloaded vehicle further includes the following steps:

[0117] Step 1011: Acquire a second image in which the number of personnel is less than or equal to the first preset number and the number of personnel is greater than or equal to a second preset number.

[0118] In a specific implementation, after determining the number of people in each frame image corresponding to each driving surveillance video in at least one acquired driving surveillance video, a second image in which the number of people is less than a first preset number and greater than or equal to a second preset number can be acquired, that is, an image with a non-overcrowded feature.

[0119] Here, the second preset number is smaller than the first preset number. The second preset number can be set in advance according to the vehicle model. For example, if the vehicle model is a 5-seater vehicle, the second preset number can be set to 4. By setting the second preset number, some images that are obviously not overcrowded scenes can be filtered out, such as images with 2 people. For these images that are obviously not overcrowded scenes, the overcrowded recognition model to be trained can easily identify them. In this way, by only training the second images that are less than or equal to the first preset number and greater than or equal to the second preset number, that is, only training the model close to the number of people carried by the vehicle, the learning ability of the overcrowded recognition model training can be improved while improving the training efficiency of the overcrowded recognition model to be trained.

[0120] Step 1012: Determine the non-overcrowded sample based on the second image.

[0121] In a specific implementation, a non-overcrowded sample may be determined based on the acquired second image having a non-overcrowded feature.

[0122] It should be noted that the non-overcrowded samples in the present application can be obtained in two ways. One is to select directly from the corresponding second image in the driving surveillance video, and the other is to select from the corresponding image in the second driving surveillance video within the second time period where the non-overcrowded scene appears in the driving surveillance video where the first image is located. In this way, non-overcrowded samples can be obtained through multiple methods, which can improve the diversity of non-overcrowded samples and thereby improve the learning ability of the overcrowded model to be trained.

[0123] In a possible implementation manner, determining the non-overcrowded sample based on the second image in step 1012 includes the following steps:

[0124] Step a: Count the second number of the second images in each driving monitoring video.

[0125] In a specific implementation, after obtaining the second image in which the number of people in each frame image corresponding to each driving surveillance video is less than or equal to the first preset number and greater than or equal to the second preset number, that is, after determining the second image with non-overcrowded features in each driving surveillance video, the second number of second images corresponding to each driving surveillance video is counted respectively, that is, the number of second images with non-overcrowded features in each driving surveillance video is counted.

[0126] Step b: Determine whether the second number is greater than or equal to a second preset threshold.

[0127] In a specific implementation, after counting the second number of second images in each driving monitoring video, it is determined whether the number of images having the second image in each driving monitoring video is greater than or equal to a second preset threshold.

[0128] Step c: If the second number is greater than or equal to the second preset threshold, a second image of the second preset threshold is selected from the driving monitoring video corresponding to the second number as the non-overcrowded sample.

[0129] In a specific implementation, if the second number of second images of a certain driving surveillance video is greater than or equal to a second preset threshold, that is, the number of second images with non-overcrowded features in the driving surveillance video is relatively large, only second images with a second preset threshold can be selected from the second images with the second number as the screened non-overcrowded samples. In this way, the number of repeated training processes occurring during the training of the overcrowded recognition model to be trained due to the excessive number of second images of the same type from the same driving surveillance video can be reduced. This can improve the efficiency of the overcrowded recognition model to be trained while ensuring the demand for non-overcrowded samples for training the overcrowded recognition model to be trained.

[0130] It should be noted that the second preset threshold has no direct relationship with the first preset threshold, and the second preset threshold can be set according to actual needs. The second preset threshold is used to limit the number of second images screened from each driving monitoring video.

[0131] Step d: If the second number is less than the second preset threshold, determine that all the second images in the driving monitoring video corresponding to the second number are the non-overcrowded samples. The second preset threshold can usually be set to 3.

[0132] In a specific implementation, if the second number of second images in a certain driving surveillance video is less than a second preset threshold, that is, the number of second images with non-overcrowding features in the driving surveillance video is small, all second images with the second number can be used as non-overcrowding samples.

[0133] In a possible implementation manner, in step 101, the number of people in each frame of image is determined according to the following steps:

[0134] Input the at least one driving surveillance video into the trained personnel quantity detection model to determine the number of people in each frame image corresponding to each driving surveillance video in the at least one driving surveillance video; wherein the personnel quantity detection model includes a face detection model or a body detection model.

[0135] In a specific implementation, each of the at least one driving monitoring video can be input into the trained personnel quantity detection model, and the number of personnel corresponding to each frame image in each driving monitoring video can be output respectively.

[0136] Here, the face detection model can be a face recognition model, which can determine the number of people corresponding to each frame by identifying the number of faces in each frame; the human body detection model can determine the number of people corresponding to each frame by identifying the number of human bodies in each frame.

[0137] In a possible implementation, the overcrowded recognition model to be trained is trained according to the following steps: the overcrowded recognition model to be trained is trained according to the overcrowded samples and the type labels of the overcrowded samples, and the non-overcrowded samples and the type labels of the non-overcrowded samples.

[0138] In a specific implementation, overcrowded samples (images with overcrowded features) and non-overcrowded samples (images with non-overcrowded features) can be input into the overcrowded recognition model to be trained respectively, or the overcrowded samples and non-overcrowded samples can be input into the overcrowded recognition model to be trained together, and the recognition results of whether the vehicles are overcrowded corresponding to the overcrowded samples and the non-overcrowded samples can be output. The parameters of the overcrowded recognition model to be trained are adjusted according to the recognition results, the type labels of the overcrowded samples and the type labels of the non-overcrowded samples, so as to achieve the purpose of training the overcrowded recognition model to be trained.

[0139] In an embodiment of the present application, by determining the number of persons in each frame image corresponding to each acquired driving surveillance video, and intercepting the first driving surveillance video within a first time period in which an overcrowded scene occurs, and the second driving surveillance video within a second time period in which a non-overcrowded scene occurs from the driving surveillance video where the first image in which the number of persons is greater than a first preset number is located, further, using the image in the first driving surveillance video as an overcrowded sample, and using the image in the second driving surveillance video as a non-overcrowded sample, the above-mentioned method of obtaining overcrowded samples and non-overcrowded samples can make the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained come from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving surveillance video corresponding to the same vehicle, that is, the overcrowded recognition model to be trained learns the scenes of the same vehicle more comprehensively, and thus, the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

[0140] Embodiment 2

[0141] Figure 2 This is a flowchart of another method for identifying an overloaded vehicle provided in the second embodiment of the present application. The device for executing the method for identifying an overloaded vehicle may be a cloud platform or a server. The following describes the method for identifying an overloaded vehicle provided in the second embodiment of the present application from the perspective of the server as the execution subject. Figure 2 As shown, the vehicle overload identification method provided in the second embodiment of the present application includes the following steps:

[0142] S201: Acquire at least one driving surveillance video, and determine the number of persons in each frame of an image corresponding to each driving surveillance video in the at least one driving surveillance video.

[0143] S202: Acquire a first image in which the number of people is greater than a first preset number, and record a first time period in which an overcrowded scene occurs and a second time period in which a non-overcrowded scene occurs in the driving monitoring video in which the first image is located.

[0144] S203: intercepting a first driving surveillance video within the first time period and a second driving surveillance video within the second time period from the driving surveillance video where the first image is located.

[0145] S204: Using the images in the first traffic monitoring video as overcrowded samples and the images in the second traffic monitoring video as non-overcrowded samples, training the overcrowded recognition model to be trained, and using the trained overcrowded recognition model to recognize overcrowded vehicles.

[0146] S205: Obtain the target surveillance video corresponding to the target travel order.

[0147] In a specific implementation, after the overcrowding recognition model is trained, the target monitoring video corresponding to the target travel order can be obtained first. Here, each travel order corresponds to a monitoring video.

[0148] It should be noted that monitoring equipment can be configured in the vehicle to monitor the people in the vehicle and the driving process in real time, and monitoring videos corresponding to the travel order can also be obtained through the terminals corresponding to the driver and / or passengers.

[0149] S206: Determine whether the vehicle corresponding to the target travel order is overcrowded based on the target monitoring video and the trained overcrowding recognition model.

[0150] In a specific implementation, the target surveillance video is input into the trained overcrowded identification model, which can directly output the identification result of whether the vehicle corresponding to the target travel order is overcrowded.

[0151] Among them, the description of S201 to S204 can refer to the description of S101 to S104, and can achieve the same technical effect, which is not repeated here.

[0152] In an embodiment of the present application, by acquiring the target monitoring video corresponding to the target travel order and the trained overcrowding identification model, the identification result of whether the vehicle corresponding to the target travel order is overcrowded can be determined, that is, whether the vehicle is overcrowded can be identified through the machine model without consuming manpower costs, and since there are no human subjective factors, the accuracy and efficiency of the trained overcrowding identification model in identifying whether there is overcrowding can be guaranteed.

[0153] In a possible implementation, obtaining a target monitoring video corresponding to a target travel order in step S205 includes the following steps:

[0154] The order information of the target travel order is obtained; after the travel time in the order information, the target monitoring video is obtained in real time.

[0155] In a specific implementation, after detecting that a passenger has initiated a target travel order, the order information of the target travel order can be obtained first. Here, the order information may include travel time, travel starting point, travel destination and other information. Since the passenger has not boarded the vehicle before the travel time, the target monitoring video is only obtained in real time after the travel time. In this way, unnecessary monitoring videos can be avoided from being input into the trained overcrowded time identification model, thereby improving the efficiency of identifying whether the vehicle is overcrowded.

[0156] In a possible implementation, determining whether the vehicle corresponding to the target travel order is overcrowded based on the target monitoring video and the trained overcrowded identification model in step 206 includes the following steps:

[0157] Step 2061: Determine, based on the target surveillance video, a trained overcrowding recognition model that matches the vehicle model in the target surveillance video.

[0158] In a specific implementation, after obtaining the target monitoring video corresponding to the target travel order, the model of the vehicle in the target monitoring video is determined according to the target monitoring video, and further, the trained overcrowding recognition model is matched according to the model of the vehicle.

[0159] It should be noted that the trained overcrowding recognition models are different for different vehicle models, and the number of people in the images corresponding to the overcrowded samples and non-overcrowded samples corresponding to the overcrowded recognition models is also different. For example, the first preset number set in the overcrowded recognition model corresponding to a 5-seater car is 5.

[0160] Step 2062: Divide the target surveillance video into frames to obtain multiple frame images.

[0161] In a specific implementation, the target surveillance video is processed by dividing the frames into frames, and multiple frames of images corresponding to the target surveillance video can be obtained.

[0162] Step 2063: Input the multiple frames of images into the trained overcrowded recognition model in sequence to determine whether the vehicle is overcrowded.

[0163] In a specific implementation, each frame of the obtained multiple frames of images is input into the trained overcrowded recognition model in turn, and the recognition result corresponding to each frame of the image can be output. When the recognition results corresponding to more than a preset number of images are all overcrowded recognition results, it is determined that the vehicle corresponding to the target travel order is overcrowded. When the recognition results corresponding to less than a preset number of images are overcrowded recognition results, it may be that there is a possibility of incorrect recognition of individual images, and therefore it is determined that the vehicle corresponding to the target travel order is not overcrowded.

[0164] In a possible implementation manner, after determining that the vehicle is overloaded, the method for identifying an overloaded vehicle further includes the following steps:

[0165] An overcrowding warning message is sent to the service provider and service requester corresponding to the target travel order.

[0166] In a specific implementation, if it is identified that the vehicle corresponding to the target travel order is overcrowded, an overcrowding warning message can be sent to the service provider and service requester corresponding to the target travel order to constrain the behavior of the service provider and service requester.

[0167] In an embodiment of the present application, by acquiring the target monitoring video corresponding to the target travel order and the trained overcrowding identification model, the identification result of whether the vehicle corresponding to the target travel order is overcrowded can be determined, that is, whether the vehicle is overcrowded can be identified through the machine model without consuming manpower costs, and since there are no human subjective factors, the accuracy and efficiency of the trained overcrowding identification model in identifying whether there is overcrowding can be guaranteed.

[0168] Embodiment 3

[0169] Based on the same application concept, the embodiments of the present application also provide a vehicle overcrowding identification device corresponding to the vehicle overcrowding identification method provided in Embodiment 1 and Embodiment 2. Since the principle of solving the problem by the device in the embodiments of the present application is similar to the vehicle overcrowding identification method in the above-mentioned Embodiment 1 and Embodiment 2 of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0170] See also Figure 3 FIG. 1 is a functional module diagram of a vehicle overload identification device 300 provided in Embodiment 3 of the present application, see Figure 4 As shown in FIG. 2 , it is a functional module diagram of a vehicle overload identification device 300 provided in the third embodiment of the present application. Figure 3 and Figure 4 As shown, the vehicle overload identification device 300 includes:

[0171] A first determination module 310 is used to obtain at least one driving monitoring video and determine the number of people in each frame of an image corresponding to each driving monitoring video in the at least one driving monitoring video;

[0172] The recording module 320 is used to obtain a first image in which the number of people is greater than a first preset number, and record a first time period in which an overcrowded scene appears and a second time period in which a non-overcrowded scene appears in the driving monitoring video in which the first image is located;

[0173] The interception module 330 is used to intercept the first driving monitoring video within the first time period and the second driving monitoring video within the second time period from the driving monitoring video where the first image is located;

[0174] The training module 340 is used to use the images in the first vehicle monitoring video as overcrowded samples and the images in the second vehicle monitoring video as non-overcrowded samples to train the overcrowded recognition model to be trained. The trained overcrowded recognition model is used to identify overcrowded vehicles.

[0175] In an embodiment of the present application, the number of persons in each frame image corresponding to each acquired driving surveillance video is determined by the first determination module 310, and the first driving surveillance video within a first time period in which an overcrowded scene occurs and the second driving surveillance video within a second time period in which a non-overcrowded scene occurs are intercepted from the driving surveillance video where the first image in which the number of persons is greater than the first preset number by the interception module 330. Furthermore, the image in the first driving surveillance video is used as an overcrowded sample, and the image in the second driving surveillance video is used as a non-overcrowded sample. By adopting the above-mentioned method of obtaining overcrowded samples and non-overcrowded samples, the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained can be made from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving surveillance video corresponding to the same vehicle, that is, the overcrowded recognition model to be trained learns the scenes of the same vehicle more comprehensively. In this way, the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

[0176] In a possible implementation, Figure 4 As shown, the recording module 320 includes:

[0177] A counting unit 322, configured to count a first number of the first images of each driving monitoring video;

[0178] A judging unit 324, configured to judge whether the first number is greater than or equal to a first preset threshold;

[0179] The selection unit 326 is used to select the first image of the first preset threshold from the driving monitoring video corresponding to the first number if the first number is greater than or equal to the first preset threshold.

[0180] In a possible implementation, Figure 4 As shown, the vehicle overcrowding identification device 300 further includes:

[0181] A first acquisition module 350 is used to acquire a second image in which the number of personnel is less than or equal to the first preset number and the number of personnel is greater than or equal to a second preset number;

[0182] The second determination module 360 ​​is configured to determine the non-overcrowded sample based on the second image.

[0183] In a possible implementation, Figure 4 As shown, the second determination module 360 ​​is used to determine the non-overcrowded sample according to the following steps:

[0184] Counting the second number of the second images in each driving monitoring video;

[0185] Determining whether the second number is greater than or equal to a second preset threshold;

[0186] If the second number is greater than or equal to the second preset threshold, selecting a second image of the second preset threshold from the traffic monitoring video corresponding to the second number as the non-overcrowded sample;

[0187] If the second number is less than the second preset threshold, it is determined that all the second images in the driving monitoring video corresponding to the second number are the non-overcrowded samples.

[0188] In a possible implementation, Figure 3 and Figure 4 As shown, the first determination module 310 is used to determine the number of people in each frame of image according to the following steps:

[0189] Inputting the at least one driving surveillance video into the trained personnel quantity detection model to determine the number of personnel in each frame image corresponding to each driving surveillance video in the at least one driving surveillance video;

[0190] Wherein, the personnel quantity detection model includes a face detection model or a body detection model.

[0191] In a possible implementation, Figure 3 and Figure 4 As shown, the training module 340 is used to train the overcrowding recognition model to be trained according to the following steps:

[0192] The overcrowded recognition model to be trained is trained according to the overcrowded samples and the type labels of the overcrowded samples, and the non-overcrowded samples and the type labels of the non-overcrowded samples.

[0193] In a possible implementation, Figure 4 As shown, the vehicle overcrowding identification device 300 further includes:

[0194] The second acquisition module 370 is used to acquire the target monitoring video corresponding to the target travel order;

[0195] The third determination module 380 is used to determine whether the vehicle corresponding to the target travel order is overcrowded based on the target monitoring video and the trained overcrowded recognition model.

[0196] In a possible implementation, Figure 4 As shown, the second acquisition module 370 is used to acquire the target monitoring video corresponding to the target travel order according to the following steps:

[0197] Obtaining order information of the target travel order;

[0198] After the travel time in the order information, the target monitoring video is acquired in real time.

[0199] In a possible implementation, Figure 4 As shown, the third determination module 380 is used to determine whether the vehicle corresponding to the target travel order is overloaded according to the following steps:

[0200] Determining, according to the target surveillance video, a trained overcrowding recognition model that matches the vehicle model in the target surveillance video;

[0201] Divide the target surveillance video into frames to obtain multiple frame images;

[0202] The multiple frames of images are input into the trained overcrowded recognition model in sequence to determine whether the vehicle is overcrowded.

[0203] In a possible implementation, Figure 4 As shown, the vehicle overcrowding identification device 300 further includes:

[0204] The sending module 390 is used to send overcrowding warning information to the service provider and service requester corresponding to the target travel order.

[0205] In an embodiment of the present application, the number of persons in each frame image corresponding to each acquired driving surveillance video is determined by the first determination module 310, and the first driving surveillance video within a first time period in which an overcrowded scene occurs and the second driving surveillance video within a second time period in which a non-overcrowded scene occurs are intercepted from the driving surveillance video where the first image in which the number of persons is greater than the first preset number by the interception module 330. Furthermore, the image in the first driving surveillance video is used as an overcrowded sample, and the image in the second driving surveillance video is used as a non-overcrowded sample. By adopting the above-mentioned method of obtaining overcrowded samples and non-overcrowded samples, the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained can be made from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving surveillance video corresponding to the same vehicle, that is, the overcrowded recognition model to be trained learns the scenes of the same vehicle more comprehensively. In this way, the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

[0206] Embodiment 4

[0207] Based on the same application concept, see Figure 5 As shown, it is a schematic diagram of the structure of an electronic device 500 provided in Embodiment 4 of the present application, including: a processor 510, a memory 520 and a bus 530, wherein the memory 520 stores machine-readable instructions executable by the processor 510, and when the electronic device 4500 is running, the processor 510 communicates with the memory 520 through the bus 530, and the machine-readable instructions are executed by the processor 510 when running to perform the steps of the method for identifying an overloaded vehicle as described in any one of Embodiments 1 and / or 2.

[0208] In an embodiment of the present application, by determining the number of persons in each frame image corresponding to each acquired driving surveillance video, and intercepting the first driving surveillance video within a first time period in which an overcrowded scene occurs, and the second driving surveillance video within a second time period in which a non-overcrowded scene occurs from the driving surveillance video where the first image in which the number of persons is greater than a first preset number is located, further, using the image in the first driving surveillance video as an overcrowded sample, and using the image in the second driving surveillance video as a non-overcrowded sample, the above-mentioned method of obtaining overcrowded samples and non-overcrowded samples can make the overcrowded samples and non-overcrowded samples for training the overcrowded recognition model to be trained come from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving surveillance video corresponding to the same vehicle, that is, the overcrowded recognition model to be trained learns the scenes of the same vehicle more comprehensively, and thus, the accuracy of the trained overcrowded recognition model in identifying overcrowded vehicles can be improved.

[0209] Embodiment 5

[0210] Based on the same application concept, embodiment 5 of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the vehicle overloading identification method provided in the above-mentioned embodiment 1 and / or embodiment 2 are executed.

[0211] Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, etc. When the computer program on the storage medium is run, the above-mentioned vehicle overcrowding identification method can be executed. By making the overcrowded samples and non-overcrowded samples for training the overcrowded identification model to be trained come from the images corresponding to the overcrowded scenes and non-overcrowded scenes in the same driving monitoring video corresponding to the same vehicle, the trained overcrowded identification model can more accurately identify the overcrowding of the moving vehicle.

[0212] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or can be integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0213] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0214] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit.

[0215] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0216] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for identifying an overloaded vehicle, characterized in that: The identification method comprises: Acquire at least one driving surveillance video, and determine the number of persons in each frame of an image corresponding to each driving surveillance video in the at least one driving surveillance video; Acquire a first image in which the number of people is greater than a first preset number, and record a first time period in which an overcrowded scene appears and a second time period in which a non-overcrowded scene appears in the driving surveillance video in which the first image is located, wherein the first time period and the second time period are determined by inputting the driving surveillance video in which the first image is located into a trained model for detecting the number of people; intercepting a first driving surveillance video within the first time period and a second driving surveillance video within the second time period from the driving surveillance video where the first image is located, including: intercepting the first driving surveillance video where an overcrowded scene appears and intercepting the second driving surveillance video where a non-overcrowded scene appears from the driving surveillance video where the first image is located according to the start and end times of the first time period and the start and end times of the second time period, so as to obtain videos from the same driving surveillance video corresponding to the time periods of the overcrowded scene and the non-overcrowded scene; The images in the first vehicle monitoring video are used as overcrowded samples, and the images in the second vehicle monitoring video are used as non-overcrowded samples. The overcrowded recognition model to be trained is trained, and the trained overcrowded recognition model is used to identify overcrowded vehicles.

2. The identification method according to claim 1, characterized in that: After acquiring the first image in which the number of people is greater than the first preset number, before recording the first time period in which an overcrowded scene appears and the second time period in which a non-overcrowded scene appears in the driving monitoring video where the first image is located, the recognition method further includes: Counting the first number of the first images in each driving monitoring video; Determining whether the first number is greater than or equal to a first preset threshold; If the first number is greater than or equal to the first preset threshold, the first image of the first preset threshold is selected from the driving monitoring video corresponding to the first number.

3. The identification method according to claim 1, characterized in that: After acquiring at least one driving surveillance video and determining the number of persons in each frame of an image corresponding to each driving surveillance video in the at least one driving surveillance video, the recognition method further includes: Acquire a second image in which the number of personnel is less than or equal to the first preset number and the number of personnel is greater than or equal to a second preset number; The non-overcrowded sample is determined based on the second image.

4. The identification method according to claim 3, characterized in that: The determining the non-overcrowded sample based on the second image includes: Counting the second number of the second images in each driving monitoring video; Determining whether the second number is greater than or equal to a second preset threshold; If the second number is greater than or equal to the second preset threshold, selecting a second image of the second preset threshold from the traffic monitoring video corresponding to the second number as the non-overcrowded sample; If the second number is less than the second preset threshold, it is determined that all the second images in the driving monitoring video corresponding to the second number are the non-overcrowded samples.

5. The identification method according to claim 1, characterized in that: Determine the number of people in each frame using the following steps: Inputting the at least one driving surveillance video into the trained personnel quantity detection model to determine the number of personnel in each frame image corresponding to each driving surveillance video in the at least one driving surveillance video; Wherein, the personnel quantity detection model includes a face detection model or a body detection model.

6. The identification method according to claim 1, characterized in that: Train the overcrowding recognition model to be trained according to the following steps: The overcrowded recognition model to be trained is trained according to the overcrowded samples and the type labels of the overcrowded samples, and the non-overcrowded samples and the type labels of the non-overcrowded samples.

7. The identification method according to claim 1, characterized in that: The identification method further comprises: Obtain the target surveillance video corresponding to the target travel order; Based on the target monitoring video and the trained overcrowded recognition model, it is determined whether the vehicle corresponding to the target travel order is overcrowded.

8. The identification method according to claim 7, characterized in that: The step of obtaining the target surveillance video corresponding to the target travel order includes: Obtaining order information of the target travel order; After the travel time in the order information, the target monitoring video is acquired in real time.

9. The identification method according to claim 7, characterized in that: The determining whether the vehicle corresponding to the target travel order is overcrowded based on the target monitoring video and the trained overcrowded identification model includes: Determining, according to the target surveillance video, a trained overcrowding recognition model that matches the vehicle model in the target surveillance video; Divide the target surveillance video into frames to obtain multiple frame images; The multiple frames of images are input into the trained overcrowded recognition model in sequence to determine whether the vehicle is overcrowded.

10. The identification method according to claim 7, characterized in that: After determining that the vehicle is overloaded, the identification method further includes: An overcrowding warning message is sent to the service provider and service requester corresponding to the target travel order.

11. A vehicle overload identification device, characterized in that: The identification device comprises: A first determination module is used to obtain at least one driving monitoring video and determine the number of people in each frame image corresponding to each driving monitoring video in the at least one driving monitoring video; a recording module, used for acquiring a first image in which the number of people is greater than a first preset number, and recording a first time period in which an overcrowded scene appears and a second time period in which a non-overcrowded scene appears in a driving surveillance video in which the first image is located, wherein the first time period and the second time period are determined by inputting the driving surveillance video in which the first image is located into a trained model for detecting the number of people; an interception module, for intercepting a first driving surveillance video within the first time period and a second driving surveillance video within the second time period from the driving surveillance video where the first image is located, including: according to the start and end times of the first time period and the start and end times of the second time period, intercepting the first driving surveillance video where an overcrowded scene appears and intercepting the second driving surveillance video where a non-overcrowded scene appears from the driving surveillance video where the first image is located, so as to obtain videos from the time periods corresponding to the overcrowded scene and the non-overcrowded scene in the same driving surveillance video; The training module is used to use the images in the first vehicle monitoring video as overcrowded samples and the images in the second vehicle monitoring video as non-overcrowded samples to train the overcrowded recognition model to be trained. The trained overcrowded recognition model is used to identify overcrowded vehicles in motion.

12. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the vehicle overload identification method according to any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for identifying an overloaded vehicle as claimed in any one of claims 1 to 10 are executed.

Images