CN114445466A

CN114445466A - Processing method, device and equipment of vehicle information recognition model

Info

Publication number: CN114445466A
Application number: CN202011112819.9A
Authority: CN
Inventors: 杨小平
Original assignee: SF Technology Co Ltd
Current assignee: SF Technology Co Ltd
Priority date: 2020-10-16
Filing date: 2020-10-16
Publication date: 2022-05-06

Abstract

The application provides a processing method, a processing device and processing equipment of a vehicle information identification model, which are used for improving the identification precision of the vehicle information identification model. The method comprises the following steps: acquiring a vehicle image acquired by a camera at a preset place and identifying vehicle information, wherein the preset place is a place in a target place where vehicle management equipment is configured; when an abnormal vehicle image with abnormal vehicle information identification exists, the abnormal vehicle image is uploaded to a server, so that the server trains an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image is endowed with a label of target vehicle information added by a user, and the trained model is used as a vehicle information identification model which is used for identifying vehicle information of a vehicle corresponding to a target field; and receiving the vehicle information identification model issued by the server.

Description

Processing method, device and equipment of vehicle information recognition model

Technical Field

The application relates to the field of vehicles, in particular to a method, a device and equipment for processing a vehicle information identification model.

Background

For places such as parking lots and parks where a large number of vehicles are parked, managers often introduce an automatic system nowadays to realize automatic management such as parking management and charging management.

The system comprises a vehicle, a vehicle body, a vehicle front-end device, a vehicle rear-end device and a vehicle front-end device.

In the existing research process of the related art, the inventor finds that when the existing vehicle management system is applied to a specific field, the vehicle information identification precision may be unstable, for example, a normal vehicle information identification effect can be achieved in a parking lot a, and a normal vehicle information identification effect cannot be achieved in a parking lot B; for example, the normal vehicle information recognition effect cannot be achieved in the parking lot C, and the normal vehicle information recognition effect can be achieved in the parking lot D.

Disclosure of Invention

The application provides a processing method, a processing device and a processing device of a vehicle information identification model, which are used for improving the vehicle state identification precision of the vehicle information identification model, so that the normal work of vehicle management equipment can be guaranteed.

In a first aspect, the present application provides a method for processing a vehicle information recognition model, the method including:

acquiring a vehicle image acquired by a camera at a preset place and carrying out vehicle information identification on the vehicle image, wherein the preset place is a place in a target place where vehicle management equipment is configured;

when an abnormal vehicle image with abnormal vehicle information identification exists, the abnormal vehicle image is uploaded to a server, so that the server trains an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image is endowed with a label of target vehicle information added by a user, and the trained model is used as a vehicle information identification model, wherein the vehicle information identification model is used for identifying vehicle information of a vehicle corresponding to a target field;

and receiving the vehicle information identification model issued by the server.

In a second aspect, the present application provides a method for processing a vehicle information recognition model, the method including:

receiving an abnormal vehicle image uploaded by a vehicle management device, wherein the vehicle management device is used for acquiring the vehicle image acquired by a camera at a preset place and carrying out vehicle information identification on the vehicle image, the preset place is a place in a target place where the vehicle management device is configured, and the abnormal vehicle image is the vehicle image with abnormal vehicle information identification;

after the abnormal vehicle image is endowed with target vehicle information added by a user, training an initial neural network model according to the abnormal vehicle image, and taking the trained model as a vehicle information identification model, wherein the vehicle information identification model is used for identifying vehicle information of a vehicle corresponding to a target site;

and issuing the vehicle information identification model to the vehicle management equipment.

In a third aspect, the present application provides a processing apparatus for a vehicle information recognition model, the apparatus comprising:

the vehicle management device comprises an acquisition unit, a processing unit and a display unit, wherein the acquisition unit is used for acquiring a vehicle image acquired by a camera at a preset place, and the preset place is a place in a target place where the vehicle management device is configured;

the identification unit is used for identifying the vehicle information of the vehicle image;

the uploading unit is used for uploading the abnormal vehicle image to the server when the abnormal vehicle image with the abnormal vehicle information identification exists, so that the server trains an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image gives a label of target vehicle information added by a user, and takes the trained model as a vehicle information identification model, wherein the vehicle information identification model is used for identifying vehicle information of a vehicle corresponding to a target field;

and the receiving unit is used for receiving the vehicle information identification model issued by the server.

In a fourth aspect, the present application provides still another processing apparatus for a vehicle information recognition model, the apparatus including:

the vehicle management device is used for acquiring a vehicle image acquired by a camera at a preset place and identifying vehicle information of the vehicle image, the preset place is a place in a target place where the vehicle management device is configured, and the abnormal vehicle image is a vehicle image with abnormal vehicle information identification;

the training unit is used for training an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image gives target vehicle information added by a user, and taking the trained model as a vehicle information identification model, wherein the vehicle information identification model is used for identifying vehicle information of a vehicle corresponding to a target field;

and the issuing unit is used for issuing the vehicle information identification model to the vehicle management equipment.

In a fifth aspect, the present application further provides a processing device for a vehicle information identification model, which includes a processor and a memory, where the memory stores a computer program, and the processor executes the method provided in any one of the first aspect, the first implementation manner, the second aspect, or the second implementation manner when calling the computer program in the memory.

In a sixth aspect, the present application further provides a computer-readable storage medium storing a plurality of instructions, where the instructions are suitable for being loaded by a processor to execute the method provided by any one of the first aspect, the first implementation manner, the second aspect, or the second implementation manner of the present application.

From the above, the present application has the following advantageous effects:

the application configures a return mechanism on site, if the vehicle management equipment deployed on the site detects that the vehicle image with abnormal vehicle information identification occurs, the abnormal vehicle image is transmitted back to the cloud server, the cloud server trains the vehicle information recognition model for the transmitted abnormal vehicle image, and then the trained vehicle information recognition model is transmitted to the vehicle management equipment deployed in the parking lot, since fixed field access vehicles and their vehicle information generally have a relatively high degree of similarity, the current vehicle information identification model is obtained by training abnormal vehicle images of the vehicle information identification abnormality in the field, so that the vehicle information identification model has strong pertinence to the vehicle and the vehicle information thereof in the field, the vehicle information identification accuracy of the vehicle information identification model in the field can be improved to a certain extent.

Drawings

FIG. 1 is a schematic view of a scenario of a vehicle information recognition model processing method according to the present application;

FIG. 2 is a schematic flow chart of a method for processing a vehicle information recognition model according to the present application;

FIG. 3 is a schematic flow chart of another exemplary method for processing a vehicle information recognition model of the present application;

FIG. 4 is a schematic flow chart illustrating vehicle information recognition of a vehicle image according to the present application;

FIG. 5 is a schematic view of another embodiment of the present disclosure for identifying vehicle information from a vehicle image;

FIG. 6 is a schematic view of another embodiment of the present disclosure for identifying vehicle information from a vehicle image;

FIG. 7 is a schematic view of another embodiment of the present disclosure for identifying vehicle information from a vehicle image;

FIG. 8 is a schematic view of another embodiment of the present disclosure for identifying vehicle information from a vehicle image;

FIG. 9 is a schematic view of another embodiment of the present disclosure for identifying vehicle information from a vehicle image;

FIG. 10 is a schematic view of another embodiment of the present disclosure for identifying vehicle information from a vehicle image;

FIG. 11 is a schematic diagram of a processing device of a vehicle information recognition model according to the present application;

FIG. 12 is a schematic view of another configuration of a processing device of the present application for identifying a vehicle information model;

fig. 13 is a schematic structural diagram of a processing device of the vehicle information recognition model according to the present application.

Detailed Description

First, before the present application is introduced, the relevant contents of the present application with respect to the application background will be described.

The processing method, the processing device and the computer-readable storage medium of the vehicle information identification model can be applied to processing equipment of the vehicle information identification model, such as vehicle management equipment deployed in a specific parking lot or a server, and are used for improving the vehicle information identification precision of the vehicle information identification model, so that the normal work of the vehicle management equipment can be guaranteed.

The execution subject of the processing method of the vehicle information identification model can be a processing device of the vehicle information identification model, or a vehicle management device or a server device integrated with the device, wherein the device can be realized in a hardware or software mode.

The vehicle management device can be particularly applied to different types of parking places such as commercial parking lots (toll parking lots), park parking lots and company parking lots, and can be used for automatic management such as vehicle access gate control, parking management, charging management and vehicle monitoring on the parking places.

Taking a scene schematic diagram of the processing method of the vehicle information identification model of the present application shown in fig. 1 as an example, a vehicle management device 101 may be deployed in a park parking lot of a logistics company for operating a vehicle management system, and the vehicle management device may collect vehicle images of parking lots in and out of the park parking lot by configuring cameras at various places, such as entrances and exits, of the park parking lot, and when an abnormal vehicle image with abnormal vehicle information identification is identified, the processing method of the vehicle information identification model provided by the present application may be triggered, the abnormal vehicle image may be uploaded to a cloud server 103 through a network 102, the vehicle information identification model may be trained by the server 103 according to the abnormal vehicle image uploaded by the vehicle management device 101, and then the vehicle information identification model may be configured to the vehicle management device 101, so as to have strong pertinence to vehicles in the park and their vehicle information, the vehicle information identification accuracy of the vehicle information identification model in the park parking lot can be improved to a certain extent.

The vehicle image used for identifying the vehicle information is generally a multi-frame image, such as a moving image or a video, so as to determine the vehicle information from a moving perspective, but of course, for an individual vehicle information identification method, only a single frame image or a single picture may be used.

In the present application, the vehicle management device may be understood as a local workstation configured for a parking lot, and typically may be a desktop computer, where the desktop computer may provide a human-computer interaction function while providing a data processing function, so that a worker may view a vehicle image (monitoring image) acquired by a camera or input a related operation instruction. After the device such as a local desktop computer and a notebook computer is configured with the application program of the vehicle management device related to the processing method of the vehicle information identification model, the device can be used as the vehicle management device in the present application (one of the processing devices of the vehicle information identification model provided by the present application), the vehicle management device itself can be further extended and configured with a camera for collecting vehicle images, or configured with other devices, such as a card reader, a barrier gate, an indicator light, etc., so as to implement automatic management of vehicle access gate control, parking management, charging management, vehicle monitoring, etc. for the parking lot.

Next, a method for processing the vehicle information recognition model provided by the present application will be described.

Referring to fig. 2 first, fig. 2 is a schematic flow chart of a processing method of a vehicle information identification model according to the present application, which is shown from a vehicle management device side on a local side, and the processing method of the vehicle information identification model according to the present application may specifically include the following steps:

step S201, acquiring a vehicle image acquired by a camera at a preset place and carrying out vehicle information identification on the vehicle image, wherein the preset place is a place in a target place where vehicle management equipment is configured;

step S202, when an abnormal vehicle image with abnormal vehicle information identification exists, the abnormal vehicle image is uploaded to a server, so that the server trains an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image gives a label of target vehicle information added by a user, and the trained model is used as a vehicle information identification model, wherein the vehicle information identification model is used for identifying vehicle information of a vehicle corresponding to a target field;

and step S203, receiving the vehicle information identification model issued by the server.

Next, referring to fig. 3, fig. 3 is a schematic flow chart of a processing method of the vehicle information recognition model according to the present application, which is shown from a server side on a cloud end side, and the processing method of the vehicle information recognition model according to the present application may specifically include the following steps:

step S301, receiving an abnormal vehicle image uploaded by a vehicle management device, wherein the vehicle management device is used for acquiring a vehicle image acquired by a camera at a preset place and identifying vehicle information of the vehicle image, the preset place is a place in the target field where the vehicle management device is configured, and the abnormal vehicle image is a vehicle image with abnormal vehicle information identification;

step S302, after the abnormal vehicle image endows target vehicle information added by a user, training an initial neural network model according to the abnormal vehicle image, and taking the trained model as a vehicle information identification model, wherein the vehicle information identification model is used for identifying vehicle information of a vehicle corresponding to a target site;

step S303, issuing the vehicle information identification model to the vehicle management apparatus.

As can be seen from the above-mentioned embodiments shown in fig. 2 and fig. 3, respectively, the present application configures a backhaul mechanism on the parking lot, and if the vehicle management device deployed on the parking lot detects that the vehicle information identifies an abnormal vehicle image, the abnormal vehicle image is transmitted back to the cloud server, the cloud server trains the vehicle information recognition model for the transmitted abnormal vehicle image, and then the trained vehicle information recognition model is transmitted to the vehicle management equipment deployed in the parking lot, since fixed field access vehicles and their vehicle information generally have a relatively high degree of similarity, the current vehicle information model is obtained by training abnormal vehicle images of the vehicle information recognition abnormality of the field, so that the vehicle information model has strong pertinence to the vehicle and the vehicle information of the field, the vehicle information identification accuracy of the vehicle information identification model in the field can be improved to a certain extent.

The steps of the processing method of the vehicle information identification model and the possible implementation manner in practical application are further described.

In the present application, the target site configured by the vehicle management device may specifically be the above-mentioned parking site, such as different types of parking sites including a commercial parking lot (toll parking lot), a park parking lot, a company parking lot, and the like.

Cameras are deployed at a plurality of places in a parking lot and are used for collecting vehicle images, and typically, the cameras can be deployed at the entrance and the exit of the parking lot, or the cameras can be deployed at places meeting place selection requirements, such as places where vehicles need to pass through in the parking lot or places with wide visual fields.

The camera not only can be understood as an independent camera, but also can be understood as a camera contained in the assembly of some equipment, and the vehicle image can be acquired at a preset place of a parking lot.

The method for acquiring the vehicle image comprises the steps of acquiring the vehicle image acquired by the camera at a preset place, acquiring the vehicle image at the preset place through the camera in real time, and acquiring the vehicle image from a storage device storing the vehicle image after the vehicle image is acquired by the deployed camera at the preset place.

After the vehicle image is obtained, vehicle state recognition can be performed in the image.

For example, the vehicle information in the image can be recognized through Artificial Intelligence (AI), i.e., through a neural network model, the neural network model is trained from a large number of different vehicle images, and particularly, can collect a large number of vehicle images, and labeling the images with vehicle information in the images as a training set, sequentially inputting the images in the training set into the activated neural network model, so that the model can be propagated forward, the vehicle information in the image can be identified, the loss function can be calculated according to the output identification result, the backward propagation can be carried out, the parameter of the model can be optimized, when the training requirements of identification precision, training times or training time, etc. are met, training of the model, which may be referred to as a vehicle information recognition model, may be completed and deployed at various parking lots to serve a vehicle management apparatus vehicle management system of the parking lot.

In order to distinguish from the vehicle information recognition model mentioned below, in the present application, the vehicle information recognition model used for recognizing the vehicle information of the vehicle image in step S201 may be referred to as an initial vehicle information recognition model, which is generally a general model used in each parking lot, or may be a vehicle information recognition model obtained by updating the general model into a specific parking lot in a one-to-one manner through subsequent data processing in the present application after the general model is configured into the specific parking lot.

In practical applications, it can be understood that the vehicle information recognition model cannot accurately recognize the vehicle information all the time, and since in practical applications, for example, ambient light, shadows, lens stains, aging of equipment components, signal interference, and the like, application conditions in terms of ambient conditions, hardware conditions, or software conditions may affect the image, and further affect the vehicle information recognition accuracy of the vehicle information recognition model, a part of the vehicle images may not recognize the vehicle information or have a poor vehicle information recognition effect, and the development process of the vehicle information recognition model may be understood as improving the vehicle information recognition accuracy as much as possible.

In the application, the abnormal vehicle images can be identified by directly considering the vehicle information, and the vehicle information identification model is updated or optimized by taking the images as the abnormal vehicle images, so that the vehicle information identification scene local to the parking lot is updated one-to-one.

Specifically, when a certain vehicle management device detects that an abnormal vehicle image with abnormal vehicle information identification exists, the image can be uploaded to a server through a network, and after the server receives the abnormal vehicle image uploaded by the vehicle management device, a vehicle information identification model can be retrained for a parking lot corresponding to the vehicle management device according to the image.

During the training process, the abnormal vehicle images can be labeled by workers. Secondly, the server can obtain the vehicle information recognition model applied by the vehicle management equipment at present, and is used for carrying out model training and updating again according to the abnormal vehicle images on the basis of the original vehicle information recognition model; or the server can also obtain an initialized neural network model and train the model only according to the abnormal vehicle images; alternatively, the server may acquire an arbitrary neural network model and train the model based on the abnormal vehicle images.

The server can also perform homogenization processing on different types of abnormal vehicle images, and can be understood. In an actual environment, the abnormal vehicle image has more data of no vehicle, rest and work, but the data of the vehicle arriving and leaving behind is less, so that the situation that the data is not uniform can occur, and the recognition effect of the final model is not ideal when the training is carried out by using the data which is not uniformly distributed. Therefore, the data can be homogenized to obtain the least data of one class in the data, then the data of each class is equalized according to the data distribution condition of other classes, and the classified data with more data quantity is deleted and not brought into the training process.

In the training process, the server can also automatically distribute a Graphic Processing Unit (GPU), analyze the GPU resource of different abnormal vehicle images by setting different data partition sizes (Batch Size) during training for different training tasks of abnormal vehicle imagesThe occupation situation is analyzed to obtain a linear GPU use situation G_all＝G_base+batch_size*G_batchWhen a plurality of tasks occur, the GPU resources are dynamically distributed to a new training task according to the known Batch Size and the tasks, so that training failure caused by memory overflow (out of memory) is avoided, and after the distributed GPU resources are determined, the training task can be performed in a mode of issuing a webpage.

After the server finishes the model training and obtains a new vehicle information identification model, the server can issue the model to the vehicle management equipment of a specific parking lot, namely, the vehicle management equipment which uploads an abnormal vehicle image for the training model originally, and after receiving the issued new vehicle information identification model, the vehicle management equipment can configure the model so as to identify the new vehicle image through the new vehicle information identification model.

For example, when the vehicle management device X uploads the abnormal vehicle image, the device identifier X may be configured for the image, and the image may be uploaded from the vehicle management device X through the device identifier X, and then, when the server performs model training, the model may also be distinguished according to the device identifier X, and the model obtained through training may also be configured with the device identifier X to identify a "one-to-one" correspondence relationship between the vehicle information identification model and the vehicle management device X, which is more suitable for vehicle information identification of a vehicle in a parking lot where the vehicle management device X is located.

As can be seen from the above, the configuration scenario of the vehicle information model provided in the present application may configure a one-to-one vehicle information recognition model for each parking lot, and since the vehicles entering and exiting from a fixed parking lot and the vehicle information thereof generally have relatively high similarity, and the vehicle information recognition model is obtained by training abnormal vehicle images in which the vehicle information of the parking lot is recognized abnormally, the vehicle information recognition model has a strong pertinence to the vehicles and the vehicle information thereof in the parking lot, and the vehicle information recognition accuracy of the vehicle information recognition model in the parking lot may be improved to a certain extent.

Secondly, in practical application, the configuration scene of the vehicle information recognition model can have further advantages in other aspects.

For example, as mentioned above, when the server performs training of the vehicle information identification model, the worker can label the vehicle information of the abnormal vehicle image uploaded by the vehicle management device, and in practical application, the labeling process does not need or rarely involve algorithm content, so that the labeling operation requirement for the worker is very low, the worker can even be a non-algorithm person, a non-programmer or even a non-technician, and only needs to label the vehicle information of the abnormal vehicle image for the labeling task distributed by the server under the office scene (e.g. webpage side operation) of remote office through a labeling tool, obviously, the training mode not only avoids a large amount of labor cost required for configuring the vehicle information identification model for the parking lot, but also can distribute the training workload of the model into different parts, for example, the training workload of the model can be further distributed to the non-technician for labeling operation of the vehicle information, this is very beneficial for the maintenance of the vehicle information identification model, and can significantly reduce the labor cost.

For another example, the configuration scenario of the vehicle information recognition model not only lies in the cloud to train the vehicle information recognition model, but also proposes a new business model of the vehicle information recognition model, after a vehicle information recognition model is configured on a specific parking lot, a highly customized model updating maintenance service can be completed subsequently through a remote mode on the specific parking lot in a one-to-one mode, and a service mode of automatically and iteratively updating and deploying the model is realized, compared with a service mode that a traditional vehicle information recognition model is configured on the specific parking lot and workers need to be dispatched to the site to perform the model updating maintenance service, the new service mode provided by the application, the method has higher practicability and application value due to the obviously reduced labor cost and the obviously improved updating and maintaining efficiency.

Next, the following describes how the abnormal vehicle image can be identified through which possible implementation manners in practical application of the present application.

Specifically, in the present application, the vehicle information identification may specifically include vehicle state information identification, license plate information identification, and compartment loading information identification, and the corresponding contents are as follows:

vehicle state information identification

And identifying the vehicle state information, namely comparing the identified vehicle state information with preset standard vehicle state information, such as arriving at post, loading, unloading, resting, leaving post and the like, and if the vehicle state information does not meet the preset standard vehicle state information, determining that the current vehicle image is an abnormal vehicle image. Further, whether the vehicle state information identification result is normal or not can be judged by combining other information.

In an exemplary implementation manner, referring to a flow diagram of the present application for identifying vehicle information of a vehicle image shown in fig. 4, in the present application, identifying vehicle information of a vehicle image may specifically include the following steps:

step S401, obtaining vehicle state information sent by a vehicle, a user terminal or vehicle sensing equipment, wherein the vehicle state information is used for indicating the current motion state or the preset motion state of the vehicle;

it is understood that in practical applications, the vehicle or a user on the vehicle may actively inform the vehicle management device of the vehicle status of the vehicle, for example, the vehicle management device may configure a receiving device at a part of a parking lot, for example, at a charging parking space, a discharge gate, a loading gate, etc., and when the user drives the vehicle, the user may operate the vehicle on-board terminal of the vehicle, so that the vehicle on-board terminal searches for the receiving device of the surrounding vehicle management device and sends the vehicle status information triggered by the user, where the vehicle status information indicates the current vehicle status of the vehicle or the preset vehicle status about to enter.

Alternatively, the vehicle state information of the vehicle may be passively collected, for example, a vehicle sensing device, such as a geomagnetic coil, which is capable of sensing an approach of the vehicle, may be disposed at a location, such as a toll parking space, a discharge gate, or a loading gate, and when the vehicle enters the location, a vehicle state corresponding to a preset function of the location may be determined, for example, when the vehicle is in the toll parking space, the vehicle sensing information collected at the toll parking space may be used to indicate that the vehicle is in a parking state, and when the vehicle is in the loading gate, the vehicle sensing information collected at the loading gate may be used to indicate that the vehicle is in a loading state.

The vehicle induction device, in particular a sensor of the type of a geomagnetic coil, an infrared sensor, an ultrasonic sensor, etc., may be adapted in particular to the actual needs.

Step S402, tracking the vehicle in the continuous frame images of the vehicle image, and identifying the motion track of the vehicle;

on the other hand, the vehicle state detection method and the vehicle state detection device can also dynamically identify the vehicle through image processing from an image processing level, and detect the vehicle state of the vehicle.

For example, a vehicle may be identified in the image, and the identified vehicle may be tracked to track the motion trajectory of the vehicle, it may be understood that, during tracking, a plurality of dense tracking points may be identified and the motion trajectories of the tracking points may be tracked, and the transportation trajectories of the tracking points may be analyzed to obtain the motion trajectory of the vehicle.

For example, the corner Detection of the vehicle of the image can be performed on the N frames of pictures by a preset corner Detection algorithm, Comer Detection, and based on a preset Detection range of the pictures, so as to obtain a large number of corners. The angular point is a specific image feature point, such as an intersection point of two or more edges, a point where an image gradient rate or a gradient change rate reaches a threshold, a point where an object edge is discontinuous, or a pixel point corresponding to a local maximum of a first-order derivative (i.e., a gradient of a gray level), and the like.

And then, a preset Optical Flow algorithm Optical Flow is called, the Optical flows of the corner points are calculated and used as the motion trail of the vehicle, and the Optical Flow algorithm is suitable for motion tracking and has higher detection precision.

Taking the motion track of one corner point in two pictures as an example, the coordinate of the corner point in the current picture is (x)_i，y_i) The angular point displacement is (Deltax)_i，Δy_i) The motion locus of the corner point is ((x)_i，y_i)，(x_i+Δx_i，y_i+Δy_i) The motion track of the angular point in the N frames of pictures is calculated in turn in this way and is recorded as (P)_i)，i∈(0，N)。

In addition, in the above, it has been mentioned that the vehicle information can be labeled on the abnormal vehicle images uploaded to the server by the staff, and in practical application, if the images are continuous frame images, the continuous frame images can be labeled in combination with the optical flow diagram, the optical flow diagram is an image composed of optical flow information, a motion track of each feature point can be shown, the optical flow diagram of the feature point of the image area can be superimposed on the image area corresponding to the vehicle state information to be labeled, and the corresponding semantic information is configured to complete the labeling, and of course, the subsequently mentioned license plate information and the carriage loading information can also be labeled in this way.

Step S403, detecting whether the motion track accords with the current motion state or the preset motion state, if not, triggering step S404;

it can be understood that, for different vehicle states, the present application may configure a corresponding motion trajectory range, for example, if the vehicle is in a parking state, the motion trajectory has a feature of being stationary for a long time; if the loading state is the loading state, the motion track characteristics of stopping the unloading bayonet, opening the vehicle door and unloading are provided.

Therefore, whether the motion trail of the vehicle identified in the vehicle image is matched with the motion state indicated in the vehicle state information or not can be compared, if the motion trail is matched with the motion state indicated in the vehicle state information, the vehicle state identification is normal, and if the motion trail is not matched with the motion state indicated in the vehicle state information, the image identification is possibly wrong, and an abnormal condition occurs.

In practical applications, there are several typical situations where the vehicle state identification is abnormal, such as:

1. the state of the vehicle is originally in a state of continuing to the post, but is in a state of no vehicle at present;

2. when the front gate is in an idle (no-vehicle) state, a loading state, an unloading state or a resting state of the vehicle appears;

3. when the vehicle at the front gate is at the gate (continuously on duty, continuously in rest state, continuously in loading and unloading state), the vehicle-free state appears.

In step S404, the vehicle image is confirmed as an abnormal vehicle image in which the vehicle information is abnormal.

After the vehicle image with the motion trail not conforming to the corresponding vehicle state information is detected to exist, the image can be identified as an abnormal vehicle image.

Secondly, license plate information recognition

And identifying license plate information, namely comparing the identified license plate information with preset standard license plate information, such as the size of a license plate and the number of characters (letters) of the license plate, and if the license plate information does not meet the preset standard license plate information, determining that the current vehicle image is an abnormal vehicle image. Furthermore, other information can be combined to judge whether the license plate information recognition result is normal.

In another exemplary implementation manner, referring to another flowchart of the present application for identifying vehicle information of a vehicle image shown in fig. 5, in the present application, identifying vehicle information of a vehicle image may specifically include the following steps:

step S501, when the preset location of the vehicle image is the exit location of the target site and the license plate recognition result of the vehicle image is normal, detecting whether the same license plate recognition result exists in the license plate recognition results recognized by the historical vehicle images within a preset time period, and if not, triggering step S502;

for example, a driving route or a driving direction is planned on a driving road in a target site, and correspondingly, one or more exit points may be set along the driving route or the driving direction, that is, in the present application, an exit point may be an exit point set between the whole parking site and the site, or an exit point set in the parking site.

In practical applications, it can be considered that when a certain vehicle passes through an exit location, the vehicle inevitably passes through an entrance location corresponding to the exit location in a previous period of time, and therefore, the vehicle inevitably leaves a vehicle image and a license plate recognition result thereof at the entrance location.

Therefore, if the vehicle management device detects that the vehicle identification result of the current vehicle image does not exist in the license plate identification result of the historical vehicle image within the preset time period, it is obvious that the vehicle identification result of the current vehicle image exists in the previous historical vehicle image, and therefore, although the current vehicle image can normally identify the license plate identification result, the license plate can be considered to have a special condition due to the fact that the corresponding historical license plate identification result does not exist, and the pass-back mechanism of the application can be triggered, that is, the following step S602 is triggered, and the current vehicle image is identified as the abnormal vehicle image.

In step S502, the vehicle image is identified as an abnormal vehicle image in which the vehicle information identifies abnormality.

After step S502 is triggered, the current vehicle image may be identified as an abnormal vehicle image.

In another exemplary implementation manner, referring to a flowchart of the present application for identifying vehicle information of a vehicle image shown in fig. 6, in the present application, identifying vehicle information of a vehicle image may specifically include the following steps:

step S601, detecting whether the license plate in the vehicle image is complete, if the license plate in the vehicle image is missing, triggering step S602;

it can be understood that in practical applications, driving habits or camera angles of some vehicles may affect positions or display effects of license plates in an image, and if the number of the license plates is fixed and the license plates identified by the model are missing, data of the license plates are certain to be data that is difficult to determine by the model, so that the image can also be used as an abnormal vehicle image.

Specifically, it is possible to detect whether the shape of the frame of the license plate in the image is a rectangle or a rectangle-like figure, and it is obvious that there is a license plate missing situation if the frame is not a rectangle or a rectangle-like figure.

In step S602, the vehicle image is identified as an abnormal vehicle image in which the vehicle information identifies an abnormality.

After the vehicle image with the missing license plate is detected, the image can be identified as an abnormal vehicle image.

Third, carriage loading information identification

And the carriage loading information identification can compare the identified carriage loading information with preset standard carriage loading information, such as the size of the carriage, the change of the carriage loading rate, the carriage loading articles and the like, and if the carriage loading information does not meet the preset standard carriage loading information, the current vehicle image can be confirmed to be an abnormal vehicle image. Furthermore, other information can be combined to judge whether the carriage loading information identification result is normal.

In another exemplary implementation manner, referring to another flowchart of the present application for identifying vehicle information of a vehicle image shown in fig. 7, in the present application, identifying vehicle information of a vehicle image may specifically include the following steps:

step S701, recognizing carriage loading information in the vehicle image to obtain a carriage loading information recognition result;

in the present application, a car loading information recognition result of a vehicle image is recognized through an initial vehicle information recognition model, and the car loading information recognition result is used for indicating the loading condition of a car of the vehicle, such as a typical car loading rate, and for a logistics vehicle which completes loading, it is obvious that the car of the logistics vehicle should be loaded with a large number of logistics pieces so as to have a high loading rate, and for a logistics vehicle which completes unloading, the car of the logistics vehicle should be not loaded with logistics pieces or loaded with few logistics pieces so as to have a low loading rate.

Step S702, judging whether the size of the detection frame indicated in the vehicle loading information is smaller than a preset detection frame size, if so, triggering step S703;

it is understood that the recognized car loading information recognition result may only describe whether the recognition is normal, the car loading information recognized normally (for example, the vehicle loading rate, etc.), the recognition abnormality cause (for example, the image cannot be opened, the car is not included in the image, the recognition time is too long, etc.), and some situations in the recognition process.

For example, in the present application, whether the car loading information recognition result is normal or not may be further determined by combining the car detection box size in the recognition process.

It can be understood that, in the application scenario of the present application, the captured normal vehicle image should not only include the car of the car, but also clearly reflect the situation of the car, so that if the car detection frame obtained by detecting the car in the recognition process is small or no car detection frame at all, obviously, the current vehicle image is not in accordance with the actual application scenario for the initial vehicle information recognition model, and belongs to a sample difficult to recognize, resulting in the occurrence of a recognition error or an unrecognizable situation.

The area surrounded by the car detection frame is a car image recognized in the image, and the detection frame is generally configured in a rectangular shape and indicated by coordinate values (x, y).

In step S703, the vehicle image is confirmed as an abnormal vehicle image in which the vehicle information is abnormal.

Therefore, when it is detected that the compartment detection frame size indicated in the recognition result is smaller than the preset detection frame size, the corresponding vehicle image may be recognized as an abnormal vehicle image.

In another exemplary implementation manner, referring to another flowchart of the present application for identifying vehicle information of a vehicle image shown in fig. 8, in the present application, identifying vehicle information of a vehicle image may specifically include the following steps:

step S801, recognizing license plates, vehicle states and compartment loading information in the vehicle images to obtain license plate recognition results, vehicle state recognition results and compartment loading information recognition results;

it can be understood that, the vehicle information can be judged whether to be normally identified according to the license plate identification result, the vehicle state identification result and the compartment loading information identification result.

The vehicle information recognition model mentioned in the present application may include an initial vehicle information recognition model, and may have a recognition function of a vehicle license plate and a vehicle state in addition to a recognition function of a vehicle loading information recognition result.

Or, the vehicle information recognition model mentioned in the present application may specifically be configured with corresponding sub-recognition models individually for car loading information recognition, license plate recognition, and vehicle state recognition.

Step S802, judging whether the license plate recognition result, the vehicle state recognition result and the carriage loading information recognition result are matched, if so, triggering step S803;

after the license plate recognition result, the vehicle state recognition result and the carriage loading information recognition result are obtained, the abnormal vehicle image can be judged according to whether the characteristics of the license plate recognition result, the vehicle state recognition result and the carriage loading information recognition result are matched.

For example, only when the vehicle has the information of the car in the processes of arriving and leaving behind, and only when the vehicle has the information of the car in the processes of arriving and leaving behind, namely, in the vehicle state from entering to leaving behind, the time sequence change characteristics of the license plate recognition result, the car loading information recognition result and the license plate recognition result can be sequentially appeared, and if the time sequence characteristics are not met, the current vehicle image can be used as an abnormal vehicle image for recognizing abnormity;

for another example, in the vehicle state of the vehicle on duty and off duty, the license plate recognition result and the compartment detection result do not exist, and the current vehicle image is an abnormal vehicle image with abnormal recognition;

for another example, in the vehicle state of the vehicle on duty and off duty, if the current license plate recognition result does not have a matching license plate recognition result in the vehicle images shot at other places in the preset event, the current vehicle image can also be used as an abnormal vehicle image for recognizing abnormality;

for example, when the vehicle is loaded in a state where the vehicle is on duty or off duty but the vehicle loading rate is not changed, the current vehicle image may be used as the abnormal vehicle image for recognizing the abnormality.

For another example, in a state where the loaded vehicle is loaded for a long period of time, the current vehicle image may be used as an abnormal vehicle image for recognizing an abnormality while maintaining a low load factor in the vehicle.

For another example, when the size of the license plate is opposite to that of the carriage, it is obvious that the situation is not met, and the current vehicle image can be used as an abnormal vehicle image for recognizing the abnormality.

In step S803, the vehicle image is identified as an abnormal vehicle image in which the vehicle information identifies an abnormality.

Therefore, when the time sequence change characteristics of the detection license plate recognition result, the vehicle state recognition result and the carriage loading information recognition result do not accord with the preset time sequence change characteristics, the corresponding vehicle image can be recognized as the abnormal vehicle image.

Second, in addition to determining whether the vehicle information identifies an abnormality in combination with the above-mentioned vehicle state information, license plate information, and compartment loading information, it may also be determined based on other reference factors, such as the degree of credibility of the vehicle information identification result and the degree of blur of the vehicle image itself.

In another exemplary implementation manner, referring to another flow diagram of the present application for identifying vehicle information of a vehicle image shown in fig. 9, in the present application, identifying vehicle information of a vehicle image may specifically include the following steps:

step S901, obtaining the confidence coefficient of the vehicle state recognition result, wherein the confidence coefficient is used for indicating the credibility of the vehicle state recognition result, and if the confidence coefficient is lower than a preset confidence coefficient threshold value, triggering step S1002;

it can be understood that, when the vehicle information is identified on the input vehicle image through the neural network model, the identification result output by the model may include, in addition to the specific content of the vehicle information, a confidence for the specific content of the current vehicle information, where the confidence is used to indicate a confidence level, and a higher confidence level obviously means a higher confidence level; the lower the confidence level, the unstable or inaccurate recognition of the image by the model is indicated, and the image can be used as an abnormal vehicle image.

In step S902, the vehicle image is identified as an abnormal vehicle image in which the vehicle information identifies an abnormality.

In this way, a confidence threshold may be set for the confidence, and when the confidence is lower than the threshold, the vehicle image may be recognized as an abnormal vehicle image regardless of whether the vehicle state is recognized by the vehicle image.

In another exemplary implementation manner, referring to another flow diagram of the present application for identifying vehicle information of a vehicle image shown in fig. 10, in the present application, identifying vehicle information of a vehicle image may specifically include the following steps:

step S1001, acquiring a fuzzy detection value of a vehicle image, wherein the fuzzy detection value is used for indicating the fuzzy degree of the vehicle image, and if the fuzzy detection value is higher than a preset fuzzy threshold value, triggering step S1002;

it can be understood that the abnormal vehicle image can also be judged from the blurring degree of the image.

Specifically, the blur detection value may be used to indicate the blur degree of the image content in the vehicle image, for example, the object in the normal vehicle image has an obvious contour and has a high contrast, the object in the vehicle image with a high blur degree does not have an obvious contrast, and a large change of pixel values is absent between pixel points.

In step S1002, the vehicle image is identified as an abnormal vehicle image in which the vehicle information identifies an abnormality.

When the vehicle image with the fuzzy detection value higher than the fuzzy threshold value is detected, the vehicle image can be identified as an abnormal vehicle image, and it can be understood that although the vehicle information of the vehicle image may also be identified in the fuzzy vehicle image, most of the fuzzy vehicle images still have the condition of high identification difficulty, so that the vehicle image can be used as the abnormal vehicle image to train a new vehicle information identification model with higher pertinence, and secondly, the vehicle image identification model can also play a role in improving the diversity of data samples, and further improve the training effect of the vehicle information identification model.

In order to better implement the processing method of the vehicle information identification model provided by the application, the application also provides a processing device of the vehicle information identification model.

Referring to fig. 11, fig. 11 is a schematic structural diagram of a processing device of a vehicle information recognition model according to the present application, which is shown from a vehicle management device side on a local side, in the present application, a processing device 1100 of a vehicle information recognition model may specifically include the following structure:

an obtaining unit 1101, configured to obtain a vehicle image acquired by a camera at a preset location, where the preset location is a location in a target site where a vehicle management device is configured;

an identifying unit 1102 for performing vehicle information identification on the vehicle image;

an uploading unit 1103, configured to upload, when there is an abnormal vehicle image with vehicle information recognition abnormality, the abnormal vehicle image to a server, so that the server trains an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image gives a label to target vehicle information added by a user, and takes the trained model as a vehicle information recognition model, where the vehicle information recognition model is used to recognize vehicle information of a vehicle corresponding to a target site;

and a receiving unit 1104, configured to receive the vehicle information identification model sent by the server. .

In an exemplary implementation manner, the identifying unit 1102 is specifically configured to:

acquiring vehicle state information sent by a vehicle, a user terminal or vehicle sensing equipment, wherein the vehicle state information is used for indicating the current motion state or the preset motion state of the vehicle;

tracking the vehicle in continuous frame images contained in the vehicle image, and identifying the motion track of the vehicle;

detecting whether the motion track accords with the current motion state or the preset motion state;

if not, the vehicle image is confirmed as an abnormal vehicle image with abnormal vehicle information identification.

In another exemplary implementation manner, the identifying unit 1102 is specifically configured to:

when the preset place of the vehicle image is an exit place of the target place and the license plate recognition result of the vehicle image is normal, detecting whether the same license plate recognition result exists in the license plate recognition results recognized by the historical vehicle images within a preset time period;

detecting whether the license plate in the vehicle image is complete;

and if the license plate in the vehicle image is missing, determining the vehicle image as an abnormal vehicle image with abnormal vehicle information identification.

identifying carriage loading information in the vehicle image to obtain a carriage loading information identification result;

judging whether the size of the detection frame indicated in the vehicle loading information is smaller than the size of a preset detection frame or not;

if yes, the vehicle image is confirmed as an abnormal vehicle image with abnormal vehicle information identification.

recognizing license plates, vehicle states and carriage loading information in the vehicle images to obtain license plate recognition results, vehicle state recognition results and carriage loading information recognition results;

judging whether the license plate recognition result, the vehicle state recognition result and the carriage loading information recognition result are consistent or not;

acquiring the confidence coefficient of the vehicle information identification result, wherein the confidence coefficient is used for indicating the credibility of the vehicle information identification result;

and if the confidence coefficient is lower than a preset confidence threshold value, determining the vehicle image as an abnormal vehicle image with abnormal vehicle information identification.

acquiring a fuzzy detection value of the vehicle image, wherein the fuzzy detection value is used for indicating the fuzzy degree of the vehicle image;

and if the fuzzy detection value is higher than a preset fuzzy threshold value, determining the vehicle image as an abnormal vehicle image with abnormal vehicle information identification.

Referring to fig. 12, fig. 12 is a schematic view of another structure of the processing device of the vehicle information recognition model according to the present application, shown from the server side of the cloud side, in the present application, the processing device 1200 of the vehicle information recognition model may specifically include the following structure:

the receiving unit 1201 is used for receiving an abnormal vehicle image uploaded by a vehicle management device, wherein the vehicle management device is used for acquiring the vehicle image acquired by a camera at a preset place and identifying vehicle information of the vehicle image, the preset place is a place in a target place where the vehicle management device is configured, and the abnormal vehicle image is a vehicle image with abnormal vehicle information identification;

the training unit 1202 is configured to train an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image gives the target vehicle information added by the user, and use the trained model as a vehicle information recognition model, where the vehicle information recognition model is used to recognize vehicle information of a vehicle corresponding to a target site;

the issuing unit 1203 is configured to issue the vehicle information identification model to the vehicle management device.

In an exemplary implementation manner, the abnormal vehicle image is a vehicle image in which a motion trajectory of the vehicle does not conform to a current motion state or a preset motion state, the current motion state or the preset motion state is obtained from vehicle state information sent by the acquired vehicle, the user terminal, or the vehicle sensing device, and the motion trajectory of the vehicle is obtained by performing vehicle tracking and recognition in consecutive frame images of the vehicle image.

In yet another exemplary implementation manner, the abnormal vehicle image is a vehicle image in which the same license plate recognition result exists in the license plate recognition results that are not detected in the historical vehicle images within the preset time period when the preset location of the image is the exit location of the target site and the license plate recognition result of the image is normal.

In yet another exemplary implementation, the abnormal vehicle image is a vehicle image in which a license plate in the image is missing.

In still another exemplary implementation, the abnormal vehicle information is a vehicle image in which a detection frame size indicated in the car loading information recognition result is smaller than a preset detection frame size, wherein the car loading information recognition result is obtained by recognizing the car loading information in the vehicle image recognition image.

In yet another exemplary implementation manner, the abnormal vehicle image is a vehicle image in which a license plate recognition result, a vehicle state recognition result, and a vehicle loading information recognition result are not matched, wherein the license plate recognition result, the vehicle state recognition result, and the vehicle loading information recognition result are obtained by recognizing a license plate, a vehicle state, and vehicle loading information in the vehicle image.

In still another exemplary implementation manner, the abnormal vehicle image is a vehicle image in which a confidence level of the vehicle information recognition result is lower than a preset confidence threshold, wherein the confidence level is used for indicating a confidence level of the vehicle information recognition result.

In still another exemplary implementation, the abnormal vehicle image is a vehicle image in which a blur detection value of the image is higher than a preset blur threshold value, wherein the blur detection value is used to indicate a degree of blur of the vehicle image.

The present application further provides a processing device of a vehicle information identification model, referring to fig. 13, fig. 13 shows a schematic structural diagram of the processing device of the vehicle information identification model of the present application, which may be the above-mentioned vehicle management device on the local side or the server on the cloud end side in practical application, specifically, the processing device of the vehicle information identification model of the present application includes a processor 1301, a memory 1302, and an input/output device 1303, where the processor 1301 is configured to implement each step of the processing method of the vehicle information identification model in any embodiment corresponding to fig. 1 to fig. 10 when executing the computer program stored in the memory 1302; alternatively, the processor 1301 is configured to implement the functions of the units in the embodiments corresponding to fig. 11 or fig. 12 when executing the computer program stored in the memory 1302, and the memory 1302 is configured to store the computer program required by the processor 1301 to execute the processing method of the vehicle information identification model in any of the embodiments corresponding to fig. 1 to fig. 10.

Illustratively, a computer program may be partitioned into one or more modules/units, which are stored in the memory 1302 and executed by the processor 1301 to accomplish the present application. One or more modules/units may be a series of computer program instruction segments capable of performing certain functions, the instruction segments being used to describe the execution of a computer program in a computer device.

The processing devices of the vehicle information identification model may include, but are not limited to, the processor 1301, the memory 1302, and the input-output device 1303. Those skilled in the art will understand that the illustration is only an example of a processing device of the vehicle information identification model, and does not constitute a limitation of the processing device of the vehicle information identification model, and may include more or less components than those shown in the drawings, or combine some components, or different components, for example, the device may further include a network access device, a bus, etc., and the processor 1301, the memory 1302, the input-output device 1303, the network access device, etc., are connected through the bus.

The Processor 1301 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, the processor being a control center of a processing device of the vehicle information recognition model, and various interfaces and lines are used to connect various parts of the entire device.

The memory 1302 may be used to store computer programs and/or modules, and the processor 1301 may implement various functions of the computer apparatus by running or executing the computer programs and/or modules stored in the memory 1302 and calling data stored in the memory 1302. The memory 1302 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created from use of a processing device of the vehicle information recognition model, and the like. In addition, the memory may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other volatile solid state storage device.

The processor 1301, when being used to execute the computer program stored in the memory 1302, may specifically implement the following functions:

Alternatively, when the processor 1301 is configured to execute the computer program stored in the memory 1302, the following functions may be specifically implemented:

It can be clearly understood by those skilled in the art that, for convenience and simplicity of description, for a specific working process of the processing apparatus and the device for a vehicle information identification model and the corresponding units thereof described above, reference may be made to the description of the processing method for a vehicle information identification model in any embodiment corresponding to fig. 1 to fig. 10, which is not repeated herein.

It will be understood by those skilled in the art that all or part of the steps of the methods of the above embodiments may be performed by instructions or by associated hardware controlled by the instructions, which may be stored in a computer readable storage medium and loaded and executed by a processor.

For this reason, the present application provides a computer-readable storage medium, where a plurality of instructions are stored, where the instructions can be loaded by a processor to execute steps in the processing method for a vehicle information identification model in any embodiment of the present application, as shown in fig. 1 to fig. 10, and specific operations may refer to descriptions of the processing method for a vehicle information identification model in any embodiment of fig. 1 to fig. 10, which are not repeated herein.

Wherein the computer-readable storage medium may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like.

Since the instructions stored in the computer-readable storage medium can execute the steps in the processing method of the vehicle information identification model in any embodiment of the present application, such as those shown in fig. 1 to fig. 10, the beneficial effects that can be achieved by the processing method of the vehicle information identification model in any embodiment of the present application, such as those shown in fig. 1 to fig. 10, can be achieved, and are described in detail in the foregoing description, and are not repeated herein.

The method, the device, the equipment and the computer-readable storage medium for processing the vehicle information identification model provided by the present application are described in detail above, and a specific example is applied in the present application to explain the principle and the implementation of the present application, and the description of the above embodiment is only used to help understanding the method and the core idea of the present application; meanwhile, for those skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.

Claims

1. A method for processing a vehicle information recognition model, the method comprising:

acquiring a vehicle image acquired by a camera at a preset place and carrying out vehicle information identification on the vehicle image, wherein the preset place is a place in a target place where the vehicle management equipment is configured;

when an abnormal vehicle image with abnormal vehicle information identification exists, uploading the abnormal vehicle image to a server, so that the server trains an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image gives a label of target vehicle information added by a user, and taking the trained model as a vehicle information identification model, wherein the vehicle information identification model is used for identifying the vehicle information of a vehicle corresponding to the target field;

2. The method of claim 1, wherein the vehicle information identifying the vehicle image comprises:

the method comprises the steps of obtaining vehicle state information sent by a vehicle, a user terminal or vehicle sensing equipment, wherein the vehicle state information is used for indicating the current motion state or the preset motion state of the vehicle;

and if not, confirming the vehicle image as an abnormal vehicle image with abnormal vehicle information identification.

3. The method of claim 1, wherein the vehicle information identifying the vehicle image comprises:

when the preset place of the vehicle image is the exit place of the target place and the license plate recognition result of the vehicle image is normal, detecting whether the same license plate recognition result exists in the license plate recognition results recognized by the historical vehicle images within a preset time period;

4. The method of claim 1, wherein the vehicle information identifying the vehicle image comprises:

detecting whether the license plate in the vehicle image is complete;

5. The method of claim 1, wherein the vehicle information identifying the vehicle image comprises:

judging whether the size of a detection frame indicated in the vehicle loading information is smaller than the size of a preset detection frame or not;

and if so, confirming the vehicle image as an abnormal vehicle image with abnormal vehicle information identification.

6. The method of claim 1, wherein the vehicle information identifying the vehicle image comprises:

judging whether the license plate recognition result, the vehicle state recognition result and the carriage loading information recognition result are matched or not;

7. The method of claim 1, wherein the vehicle information identifying the vehicle image comprises:

acquiring a confidence degree of a vehicle information identification result, wherein the confidence degree is used for indicating the credibility of the vehicle information identification result;

and if the confidence degree is lower than a preset confidence threshold value, confirming the vehicle image as an abnormal vehicle image with abnormal vehicle information identification.

8. The method of claim 1, wherein the vehicle information identifying the vehicle image comprises:

and if the fuzzy detection value is higher than a preset fuzzy threshold value, confirming the vehicle image as an abnormal vehicle image with abnormal vehicle information identification.

9. A method for processing a vehicle information recognition model, the method comprising:

receiving an abnormal vehicle image uploaded by a vehicle management device, wherein the vehicle management device is used for acquiring a vehicle image acquired by a camera at a preset place and identifying vehicle information of the vehicle image, the preset place is a place in a target place where the vehicle management device is configured, and the abnormal vehicle image is a vehicle image with abnormal vehicle information identification;

after the abnormal vehicle image is endowed with target vehicle information added by a user, training an initial neural network model according to the abnormal vehicle image, and taking the trained model as a vehicle information identification model, wherein the vehicle information identification model is used for identifying the vehicle information of a vehicle corresponding to the target site;

10. A processing apparatus of a vehicle information recognition model, characterized by comprising:

the uploading unit is used for uploading the abnormal vehicle image to a server when the abnormal vehicle image with abnormal vehicle information identification exists, so that the server trains an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image is endowed with a label of target vehicle information added by a user, and the trained model is used as a vehicle information identification model, wherein the vehicle information identification model is used for identifying the vehicle information of a vehicle corresponding to the target field;

11. A processing apparatus of a vehicle information recognition model, characterized by comprising:

the vehicle management device is used for acquiring a vehicle image acquired by a camera at a preset place and identifying vehicle information of the vehicle image, wherein the preset place is a place in a target place where the vehicle management device is configured, and the abnormal vehicle image is a vehicle image with abnormal vehicle information identification;

the training unit is used for training an initial neural network model according to the abnormal vehicle image after the abnormal vehicle image gives target vehicle information added by a user, and taking the trained model as a vehicle information recognition model, wherein the vehicle information recognition model is used for recognizing vehicle information of a vehicle corresponding to the target site;

12. A processing apparatus of a vehicle information recognition model, characterized by comprising a processor and a memory, the memory having stored therein a computer program, the processor executing the processing method of the vehicle information recognition model according to any one of claims 1 to 10 when calling the computer program in the memory.

13. A computer readable storage medium storing a plurality of instructions adapted to be loaded by a processor to perform the processing method of the vehicle information recognition model according to any one of claims 1 to 10.