Urban temporary parking vehicle charging method and device and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and provides a city temporary parking vehicle charging method, and the method comprises the steps: obtaining the vehicle feature data of a vehicle in a parking region in real time, and the vehicle feature data at least comprises the license plate image information of the vehicle in the parking region; performing enhanced feature extraction on the license plate image information by adopting a lightweight backbone network comprising asymmetric convolution blocks, and dynamically selecting an identification path through a classification module to decode to obtain vehicle identification information; a temporary charging account is established according to the vehicle identification information, and the temporary charging account is associated with a vehicle entrance timestamp and a preset charging rule; when a vehicle departure trigger event is detected, performing staged fee deduction operation based on the stay duration in the temporary charging account; and after the fee deduction verification is passed, generating an electronic release instruction, and updating the parking space state database at the same time. According to the technical scheme, the vehicle identifier can be automatically and accurately recognized, charging is correct, and therefore the dispute probability is reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly to a method, device, and storage medium for charging urban temporarily parked vehicles. Background Art

[0002] With the rapid growth of urban parking demand, existing temporary parking vehicle charging systems have significant defects in terms of real-time performance, accuracy, and efficiency, specifically including:

[0003] 1) The license plate recognition accuracy is insufficient, resulting in frequent errors in entry / exit verification and relying on manual verification.

[0004] 2) The charging and payment processes are disjointed, causing congestion at the exit.

[0005] 3) The credit risk control mechanism is lacking, resulting in low efficiency in dealing with abnormal parking (such as over-time occupancy), and a high rate of manual intervention is required.

[0006] 4) The system response latency is high, making it difficult to meet the requirements of high-concurrency scenarios.

[0007] The above defects have significantly increased the average departure time of the existing system, leading to an increase in the dispute rate and seriously restricting the turnover efficiency of parking resources. Summary of the Invention

[0008] This application provides a method, device, and storage medium for charging urban temporarily parked vehicles, which can automatically and accurately identify vehicle identification and correctly charge to reduce the probability of disputes.

[0009] On the one hand, this application provides a method for charging urban temporarily parked vehicles, and the method includes:

[0010] Obtain in real time the vehicle feature data of the vehicle in the parking area, and the vehicle feature data at least includes the license plate image information of the vehicle in the parking area;

[0011] Use a lightweight backbone network containing an asymmetric convolution block to perform enhanced feature extraction on the license plate image information, and dynamically select an identification path through a classification module for decoding to obtain vehicle identification information;

[0012] Establish a temporary charging account according to the vehicle identification information, and the temporary charging account is associated with the vehicle entry timestamp and a preset charging rule;

[0013] When a vehicle departure trigger event is detected, perform a phased deduction operation based on the parking duration in the temporary charging account;

[0014] When the deduction verification passes, generate an electronic release instruction and update the parking space status database at the same time.

[0015] On the other hand, the present application provides an urban temporary parking vehicle charging device, and the device includes:

[0016] An acquisition module, configured to acquire in real time vehicle feature data of a vehicle in a parking area, where the vehicle feature data at least includes license plate image information of the vehicle in the parking area;

[0017] A processing module, configured to perform enhanced feature extraction on the license plate image information by using a lightweight backbone network including an asymmetric convolution block, and perform decoding through a classification module to dynamically select an identification path to obtain vehicle identification information;

[0018] An account establishment module, configured to establish a temporary charging account for the vehicle identification information, and the temporary charging account is associated with a vehicle entry timestamp and a preset charging rule;

[0019] A deduction module, configured to perform a staged deduction operation based on the parking duration in the temporary charging account when a vehicle departure trigger event is detected;

[0020] A release module, configured to generate an electronic release instruction after the deduction verification is passed, and update the parking space status database at the same time.

[0021] In a third aspect, the present application provides an electronic device, and the device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the technical solution of the above-mentioned urban temporary parking vehicle charging method are implemented.

[0022] In a fourth aspect, the present application provides a storage medium, and the storage medium stores a computer program. When the computer program is executed by a processor, the steps of the technical solution of the above-mentioned urban temporary parking vehicle charging method are implemented.

[0023] As can be seen from the technical solutions provided by the present application above, on the one hand, for the above-mentioned urban temporary parking vehicle charging method, the entire process from vehicle entry to departure is automated, reducing the manual intervention nodes; on the other hand, a lightweight backbone network including an asymmetric convolution block is used to perform enhanced feature extraction on the license plate image information, which can enhance the license plate positioning and character segmentation capabilities in complex scenarios (such as tilt, low light), significantly reducing the misrecognition rate, thereby reducing the probability of disputes; on the third hand, a staged deduction operation is performed based on the parking duration in the temporary charging account, realizing the dynamic matching of the rate rule and the spatio-temporal parameters, and avoiding the charging errors caused by rule lag in the traditional system. Description of the Drawings

[0024] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0025] Figure 1 is a flowchart of the urban on-street parking vehicle charging method provided by the embodiment of the present application;

[0026] Figure 2 is a structural schematic diagram of the urban on-street parking vehicle charging device provided by the embodiment of the present application;

[0027] Figure 3 is a structural schematic of the electronic device provided by the embodiment of the present application. Detailed implementation manners

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0029] In this specification, adjectives such as first and second can only be used to distinguish one element or action from another element or action, and do not necessarily require or imply any actual such relationship or order. Where circumstances permit, reference to an element or component or step (etc.) should not be construed as being limited to only one of the elements, components, or steps, but can be one or more of the elements, components, or steps, etc.

[0030] In this specification, for ease of description, the sizes of the various parts shown in the accompanying drawings are not drawn in actual proportional relationships.

[0031] With the rapid growth of urban parking demand, there are significant defects in the real-time performance, accuracy, and efficiency of existing on-street parking vehicle charging systems, specifically including: 1) insufficient license plate recognition accuracy, resulting in frequent errors in entry / exit verification and relying on manual verification; 2) the disconnection between the charging and payment processes, causing congestion at the exit; 3) the lack of a credit risk control mechanism, resulting in low efficiency in handling abnormal parking (such as over-time occupancy) and a high rate of manual intervention required; 4) high system response latency, making it difficult to meet the requirements of high-concurrency scenarios (such as 50+ vehicle passages per minute during the morning and evening rush hours in commercial areas). The above defects have significantly increased the average departure time of the existing system, leading to an increase in the dispute rate and severely restricting the turnover efficiency of parking resources.

[0032] In view of the above problems of the prior art, the present application proposes a method for charging temporary parking vehicles in a city. The flowchart is as shown in the appendix Figure 1 and mainly includes steps S101 to S105, which are described in detail as follows:

[0033] Step S101: Real-time obtain the vehicle feature data of the vehicles in the parking area. Among them, the vehicle feature data at least includes the license plate image information of the vehicles in the parking area.

[0034] In the embodiment of the present application, the vehicle feature data of the vehicles in the parking area can be real-time obtained through the image acquisition device deployed in the parking area. Different from the traditional solution that relies on a single camera to collect license plate images and is vulnerable to light and occlusion interference, the image acquisition device in the embodiment of the present application can be a multi-sensor fusion system (such as geomagnetic + camera) to ensure that vehicle features can still be stably obtained under extreme weather (heavy rain / fog).

[0035] For safety reasons, after the image acquisition device real-time obtains the vehicle feature data of the vehicles in the parking area, the license plate image information can be real-time desensitized to hide the sensitive information in the non-license plate area; write the entry timestamp, license plate number hash value, and deduction amount into the immutable log of the blockchain; in the payment verification stage, establish an end-to-end encryption channel through the national cipher SM2 / SM4 hybrid encryption algorithm. In the above embodiment, writing the entry timestamp, license plate number hash value, and deduction amount into the immutable log of the blockchain can specifically be: performing SHA-256 hash calculation on the entry timestamp, license plate number hash value, and deduction amount to generate the first-level Merkle tree node; verifying the following timing logic through a smart contract: a) the generation time of the entry event hash value is earlier than the deduction event; b) the deduction success status is earlier than the generation of the release instruction; constructing a data index according to the Merkle tree node and providing a time range filtering query interface.

[0036] Step S102: Use a lightweight backbone network containing asymmetric convolution blocks to perform enhanced feature extraction on the license plate image information, and decode through a classification module to dynamically select the recognition path to obtain vehicle identification information.

[0037] Traditional OCR requires manual setting of segmentation rules for double-line license plates, while the present application solves the problem of recognizing special format license plates by enabling adaptive switching between single / double-line license plate recognition modes through a dynamic path selection mechanism. On the other hand, the asymmetric convolution block obtains a larger receptive field by increasing a small amount of computational effort. It should be noted that the lightweight backbone network described here is a pre-trained lightweight backbone network. Specifically, the process of training the lightweight backbone network includes the following steps S1021 to step S1023:

[0038] Step S1021: Construct a composite loss function L.

[0039] L = L c + λ3 * L r = λ1 * L slice + λ2 * L class + λ3 * L r

[0040] Here, L c is the classification loss, L c = λ1 * L slice + λ2 * L class , L r is the recognition loss. L slice is the Softmax loss function of the branch with the Softmax activation function, L class is the cross-entropy error function of the branch with the Sigmoid activation function. L slice and L class are defined as follows:

[0041]

[0042] where N is the mini-batch size of training, and T is the number of classes. Considering that the final height of the license plate image information of the vehicle in the parking area is 36 (the size h×w of the license plate image initially obtained in real time is 36×96 and is reduced to 36×24 after the max-pooling operation), this application can set T to 18, i.e., T = 18, to regress a parameter that can evenly divide the height. y is the target label, i.e., the true license plate type, and p is the predicted log odds. The recognition loss L r also consists of two parts: the single-line loss and the double-line loss, and it can be expressed as:

[0043]

[0044] where, L ctc is the single-line loss, and are the connection temporal classification losses of the upper and lower row characters respectively, collectively referred to as the double-line loss. Since y is the true license plate type, either a single-line license plate or a double-line license plate, therefore, if y = 1 represents a double-line license plate, then means that the recognition loss at this time is the double-line loss; conversely, if y = 0 represents a single-line license plate, then L r = L ctc , meaning that the recognition loss at this time is the single-line loss. In the embodiments of this application, L ctc , and are collectively referred to as the connection temporal classification loss and can all be obtained based on the connection temporal classification algorithm, specifically given by the following formula:

[0045]

[0046] Among them, P(z|x) represents the conditional probability of outputting the target sequence z through the input x. S is defined as the training data set. To calculate the connectionist temporal classification loss, the time length needs to be greater than twice the character length. Considering that the number of characters on most license plates is between 6 and 9, in this embodiment, the final feature length of decoding is set to L = 24. Since this embodiment directly uses the CNN output for decoding, the length L is used to represent the time length. Define as the predicted probability of the sequence π t at time t, P(z|x) can be expressed as:

[0047]

[0048] Step S1022: Dynamically adjust the weight coefficients λ1, λ2, and λ3 according to the double-line license plate probability value.

[0049] Considering that L c is easier to converge than L r , and using y = 1 to represent the double-line license plate, the weight coefficient λ3 of L r can be set to 10, that is, λ3 = 10. From the perspective of the convergence speed, the weight coefficients of L slice and L class can be set to 1 and 2 respectively. Therefore, the weight coefficients λ1, λ2, and λ3 can be expressed as follows:

[0050] λ1:λ2:λ3 = 1:2:(10*y)

[0051] The above y is the target label, that is, the true license plate type.

[0052] Step S1023: Implement phased training.

[0053] Specifically, it can be: 1) Freeze the parameters of the recognition module and train the classification module until convergence; 2) Jointly fine-tune the parameters of the classification module and the recognition module.

[0054] As an embodiment of this application, a lightweight backbone network containing an asymmetric convolution block is used to enhance the feature extraction of the license plate image information, and the vehicle identification information is obtained by dynamically selecting the recognition path through the classification module, which can be realized through steps S'1021 to S'1023. The detailed description is as follows:

[0055] Step S'1021: Extract the enhanced feature map of the license plate image information through the lightweight backbone network, where the lightweight backbone network contains cascaded asymmetric convolution blocks, and each asymmetric convolution block is composed of a first convolution layer with a convolution kernel size of m×n and a second convolution layer with a convolution kernel size of j×k.

[0056] On the one hand, traditional OCR technology has errors in cutting the upper and lower row characters of double-line license plates (such as new energy vehicle license plates), while adaptive feature segmentation can be achieved through a dual-task branch, namely slice parameter prediction and double-line probability judgment, avoiding the lack of generalization caused by manually setting fixed segmentation rules. On the other hand, the combination of a lightweight backbone network and an asymmetric convolution block can reduce parameter redundancy while ensuring accuracy, meeting the deployment requirements of edge devices. Therefore, in the embodiments of the present application, an enhanced feature map of the license plate image information can be extracted through a lightweight backbone network, and the lightweight backbone network includes cascaded asymmetric convolution blocks, and each asymmetric convolution block is composed of a first convolutional layer with a convolution kernel size of m×n and a second convolutional layer with a convolution kernel size of j×k. The asymmetric convolution block obtains a larger receptive field by increasing a small amount of computational effort, and asymmetric convolutions with different sizes of convolution kernels can be used to obtain a larger receptive field. By using asymmetric convolutions, for example, a convolution kernel of size 5×1 followed by a convolution kernel of size 1×5, the computational cost can be significantly saved.

[0057] Specifically, the enhanced feature map of the license plate image information extracted through the lightweight backbone network can be: performing a first convolution and a second convolution operation on the initial feature map output by the lightweight backbone network by a first convolutional layer and a second convolutional layer respectively; performing channel dimension concatenation on the first feature map and the second feature map output by the first convolution operation and the second convolution operation to generate a concatenated feature map; using a global context enhancement module to perform spatial attention weighting on the concatenated feature map to generate a weighted feature map; adding the weighted feature map and the initial feature map through a residual connection to obtain an enhanced feature map.

[0058] In the above embodiments, performing spatial attention weighting on the concatenated feature map by using a global context enhancement module to generate a weighted feature map can be specifically implemented through the following steps S1 to S4, which are described as follows:

[0059] Step S1: Performing a global average pooling operation on the concatenated feature map to obtain a spatially compressed feature map.

[0060] In the embodiments of the present application, the reason for performing a global average pooling operation on the concatenated feature map is that average pooling can reduce the spatial dimension of the concatenated feature map while retaining the average information of the features. It reduces the size of the feature map by performing average calculations on local regions of the concatenated feature map, thereby reducing the computational effort and the number of parameters, and also helps to prevent overfitting.

[0061] Step S2: Performing L2 norm normalization processing on the spatially compressed feature map to generate a normalized feature map.

[0062] The L2 norm is a norm of a vector, which represents the square root of the sum of the squares of the vector elements. Batch normalization can be used to improve the training speed and stability of the network. By normalizing the features, it can accelerate convergence, reduce sensitivity to initialization, and help prevent overfitting.

[0063] Step S3: Perform spatial dimension concatenation on the normalized feature map and the cascaded feature map to generate a weighted feature map.

[0064] Step S4: Adjust the channel dimension of the weighted feature map through a 1×1 convolutional layer to make the channel dimension of the adjusted weighted feature map consistent with that of the initial feature map.

[0065] Step S’1022: Parallelly perform the following operations in the classification module: Predict the license plate slicing parameters through the Softmax activation branch; and calculate the probability value of a double-line license plate through the Sigmoid activation branch, where the slicing parameters are used for feature map segmentation, and the probability value characterizes whether the current license plate is of the double-line type.

[0066] In the embodiment of the present application, the branch with the Softmax activation function, i.e., the Softmax activation branch, is designed to regress the slicing parameters for feature map segmentation. This embodiment uses discrete classification to simulate the segmentation position parameters for the double-line type feature map slicing. It should be noted that this branch is optional. If the height ratio between the two rows of characters is almost constant, this branch can be replaced by a fixed parameter. The branch with the sigmoid activation function, i.e., the Sigmoid activation branch, is designed to regress the probability that the license plate is double-line. The output of this branch is used for the character decoder. Specifically, predicting the license plate slicing parameters through the Softmax activation branch can be: Divide the initial feature map height output by the lightweight backbone network evenly into T candidate slicing intervals; determine the optimal slicing position through the discrete classification result output by the Softmax activation branch.

[0067] Step S’1023: According to the double-line license plate probability value and the slicing parameters, dynamically activate the feature branch in the recognition module that matches the license plate type for decoding to obtain vehicle identification information.

[0068] In the embodiment of the present application, the recognition module is designed to extract the final features and decode characters from the final features, and it includes three branches. The first branch, namely the single-line feature branch, extracts the final features of the enhanced feature map of the entire license plate image information and only works when the license plate is single-line. The remaining two branches are the upper-row feature branch and the lower-row feature branch, which are used to extract the upper and lower features of the enhanced feature map of the entire license plate image information respectively, and they only work when the license plate is double-line. In other words, the recognition module has three outputs, and the final result is determined by the classification result. When the license plate is single-line, only the first output is used. On the contrary, the second and third outputs are used to decode the upper-row and lower-row characters. Specifically, according to the double-line license plate probability value and the slicing parameters, the feature branch in the recognition module that matches the license plate type is dynamically activated for decoding to obtain the vehicle identification information, which can be: when the double-line license plate probability value is less than the threshold, the single-line feature branch is activated for decoding to obtain the vehicle identification information; when the double-line license plate probability value is greater than or equal to the threshold, the upper-row feature branch and the lower-row feature branch are synchronously activated for decoding to obtain the vehicle identification information. Here, the threshold can be set to 0.5, and the vehicle identification information specifically includes the license plate number, license plate color, vehicle type, license plate type (single-line license plate or double-line license plate), license plate area coordinates, and license plate integrity flag, etc.

[0069] It should be noted that since the connectionist temporal classification training network is used when the single-line feature branch of the embodiment of the present application decodes or the upper-row feature branch and the lower-row feature branch decode, the decoding process is to retrieve the most likely label z * 。

[0070] z * =argmax P(z|x)

[0071] where z is the label path and x is the input sequence to be decoded. The present application selects the beam search method for sorting. The algorithm continues this process until a certain termination condition is reached, for example, reaching the maximum search depth or finding a solution that meets specific conditions), because there is a trade-off between search time and accuracy.

[0072] Since the license plate number is a special character sequence with specific specifications, after the vehicle identification information is decoded by dynamically selecting the recognition path through the classification module, the license plate number in the vehicle identification information can be further filtered. Specifically, the number of characters of the decoded license plate number is compared with the predetermined number; if the number of characters of the decoded license plate number does not meet the predetermined number, it is determined that the decoded license plate number is not a valid license plate number and is directly discarded; otherwise, the characters of the decoded license plate number are further matched with the characters in the preset valid character subset; if any character in the decoded license plate number is not in the valid character subset, the decoded license plate number is discarded.

[0073] Step S103: Establish a temporary billing account based on the vehicle identification information, where the temporary billing account is associated with the vehicle entry timestamp and a preset billing rule.

[0074] Specifically, establishing a temporary billing account based on the vehicle identification information may be as follows: Extract the license plate number and vehicle type from the vehicle identification information; obtain the longitude and latitude coordinates of the parking area through the GPS positioning module; query the regional rate database based on the longitude and latitude coordinates of the parking area and load the basic rate standard; mark the holiday rate coefficient according to the special period notice pushed by the municipal management platform; match and bind the license plate number with the preset payment platform account library; when there is no matching account, generate a temporary credit guarantee amount and associate it with the license plate number.

[0075] Step S104: When a vehicle departure trigger event is detected, perform a staged deduction operation based on the parking duration in the temporary billing account.

[0076] Specifically, performing a staged deduction operation based on the parking duration in the temporary billing account may be as follows: Freeze a preset amount of credit guarantee when the vehicle enters; update the accumulated fee according to the real-time billing cycle; preferentially deduct the frozen credit guarantee when leaving; trigger a secondary payment request when the accumulated fee exceeds the credit guarantee. In the above embodiments, when the vehicle parking time exceeds the preset threshold, start a hierarchical reminder mechanism; generate a detention event record when it is detected that the deduction is not successful; dynamically adjust the subsequent deduction policy based on the historical credit rating.

[0077] It should be noted that considering that the traditional solution relies on manual inspections or fixed periods (such as every hour) to count the parking space occupancy rate and cannot capture short-term peaks (such as the end of a shopping mall event), in order to avoid misjudging the parking space occupancy rate in a single dimension, ensure that public resource scheduling takes precedence over business logic, and avoid legal disputes caused by mid-course price adjustments, and prevent database lock conflicts during large-scale concurrent billing. In the embodiments of the present application, the above method may further include dynamically adjusting the rate. Specifically, dynamically adjusting the rate includes: counting the parking space occupancy rate every predetermined time interval (for example, 5 minutes); obtaining the real-time congestion index within a preset range (for example, a radius of 1 km) through the traffic management platform; when the occupancy rate is greater than the preset occupancy rate threshold (for example, 80%) and the congestion index is greater than the preset congestion index threshold (for example, 7), activate the premium rate mode to the following final rate F end :

[0078]

[0079] where F base is the basic rate, P occup is the parking space occupancy rate; when receiving a municipal emergency instruction, forcibly switch to the emergency rate standard; maintain the original rate for the vehicles that have already entered, and apply the emergency rate to the newly entered vehicles.

[0080] Step S105: Generate an electronic release instruction after the deduction verification is passed, and update the parking space status database at the same time.

[0081] Specifically, the conditions for passing the deduction verification include: verifying that the current deduction status is completed; checking that the vehicle identification information is consistent with the entry record; and confirming that the target parking space has been released from the occupied state. After the above conditions are met, an electronic release instruction is generated, and the parking space status database is updated at the same time.

[0082] For safety reasons, in the above embodiments, after the vehicle feature data of the vehicle in the parking area is obtained in real time, the vehicle feature data can also be securely processed, including: performing real-time desensitization on the vehicle feature data of the vehicle in the parking area, hiding sensitive information in non-license plate areas; writing the entry timestamp, license plate number hash value, and deduction amount into the immutable log of the blockchain; and establishing an end-to-end encryption channel through the national cryptographic SM2 / SM4 hybrid encryption algorithm during the payment verification phase. In the above embodiments, writing the entry timestamp, license plate number hash value, and deduction amount into the immutable log of the blockchain specifically includes: performing SHA-256 hash calculation on the entry timestamp, license plate number hash value, and deduction amount to generate the first-level Merkle tree node; verifying the following timing logic through a smart contract: a) the generation time of the entry event hash value is earlier than the deduction event; b) the deduction success status is earlier than the generation of the release instruction; constructing a data index based on the Merkle tree node, and providing a time range filtering query interface.

[0083] From the above-attached Figure 1 Example of the urban on-street parking vehicle charging method, on the one hand, in the above urban on-street parking vehicle charging method, the entire process from vehicle entry to departure is automated, reducing manual intervention nodes; on the other hand, a lightweight backbone network including asymmetric convolution blocks is used to enhance the feature extraction of license plate image information, which can enhance the license plate positioning and character segmentation capabilities in complex scenarios (such as tilt, low light), significantly reducing the misrecognition rate, thereby reducing the probability of disputes; thirdly, based on the residence duration in the temporary charging account, a phased deduction operation is performed to achieve dynamic matching of the rate rule and spatio-temporal parameters, avoiding billing errors caused by rule lag in traditional systems.

[0084] Please refer to the attached Figure 2 , which is an urban on-street parking vehicle charging device provided by an embodiment of the present application. The device may include an acquisition module 201, a processing module 202, an account establishment module 203, a deduction module 204, and a release module 205, which are described in detail as follows:

[0085] The acquisition module 201 is used to obtain the vehicle feature data of the vehicle in the parking area in real time, where the vehicle feature data at least includes the license plate image information of the vehicle in the parking area;

[0086] A processing module 202 is configured to perform enhanced feature extraction on license plate image information by using a lightweight backbone network including an asymmetric convolution block, and perform decoding through a classification module to dynamically select an identification path to obtain vehicle identification information;

[0087] An account establishment module 203 is configured to establish a temporary billing account according to the vehicle identification information, wherein the temporary billing account is associated with a vehicle entry timestamp and a preset billing rule;

[0088] A deduction module 204 is configured to perform a staged deduction operation based on the stay duration in the temporary billing account when a vehicle departure trigger event is detected;

[0089] A release module 205 is configured to generate an electronic release instruction when the deduction verification is passed, and update the parking space status database at the same time.

[0090] From the above Figure 2 As can be seen from the urban on-street parking vehicle billing device in the above example, on the one hand, in the above urban on-street parking vehicle billing method, the entire process from vehicle entry to departure is automated, reducing manual intervention nodes; on the other hand, a lightweight backbone network including an asymmetric convolution block is used to perform enhanced feature extraction on license plate image information, which can enhance the license plate positioning and character segmentation capabilities in complex scenarios (such as tilt, low light), significantly reduce the misrecognition rate, and thus reduce the probability of disputes; on the third hand, a staged deduction operation is performed based on the stay duration in the temporary billing account, realizing the dynamic matching of the rate rule and spatio-temporal parameters, and avoiding billing errors caused by rule lag in traditional systems.

[0091] Figure 3 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application. As Figure 3 shown, the electronic device 3 in this embodiment mainly includes: a processor 30, a memory 31, and a computer program 32 stored in the memory 31 and executable on the processor 30, such as a program for the urban on-street parking vehicle billing method. When the processor 30 executes the computer program 32, the steps in the above urban on-street parking vehicle billing method embodiment are implemented, such as Figure 1 the steps S101 to S105 shown. Or, when the processor 30 executes the computer program 32, the functions of each module / unit in the above device embodiments are implemented, such as Figure 2 the functions of the acquisition module 201, the processing module 202, the account establishment module 203, the deduction module 204, and the release module 205 shown.

[0092] Exemplarily, the computer program 32 for the urban on-street parking vehicle charging method mainly includes: obtaining in real time the vehicle feature data of the vehicles in the parking area, where the vehicle feature data at least includes the license plate image information of the vehicles in the parking area; using a lightweight backbone network including an asymmetric convolution block to perform enhanced feature extraction on the license plate image information, and dynamically selecting an identification path through a classification module to decode to obtain vehicle identification information; establishing a temporary charging account according to the vehicle identification information, where the temporary charging account is associated with the vehicle entry timestamp and a preset charging rule; when a vehicle departure trigger event is detected, performing a staged deduction operation based on the parking duration in the temporary charging account; when the deduction verification passes, generating an electronic release instruction and simultaneously updating the parking space status database. The computer program 32 can be divided into one or more modules / units, and one or more modules / units are stored in the memory 31 and executed by the processor 30 to complete this application. One or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program 32 in the electronic device 3. For example, the computer program 32 can be divided into the functions of an acquisition module 201, a processing module 202, an account establishment module 203, a deduction module 204, and a release module 205 (modules in the virtual device), and the specific functions of each module are as follows: The acquisition module 201 is used to obtain in real time the vehicle feature data of the vehicles in the parking area, where the vehicle feature data at least includes the license plate image information of the vehicles in the parking area; the processing module 202 is used to use a lightweight backbone network including an asymmetric convolution block to perform enhanced feature extraction on the license plate image information, and dynamically select an identification path through a classification module to decode to obtain vehicle identification information; the account establishment module 203 is used to establish a temporary charging account according to the vehicle identification information, where the temporary charging account is associated with the vehicle entry timestamp and a preset charging rule; the deduction module 204 is used to perform a staged deduction operation based on the parking duration in the temporary charging account when a vehicle departure trigger event is detected; the release module 205 is used to generate an electronic release instruction when the deduction verification passes and simultaneously update the parking space status database.

[0093] The electronic device 3 may include but is not limited to the processor 30 and the memory 31. Those skilled in the art can understand that Figure 3 merely examples of the electronic device 3, which do not constitute a limitation on the electronic device 3, and may include more or fewer components than shown in the figure, or combine some components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0094] The so-called processor 30 may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0095] The memory 31 may be an internal storage unit of the electronic device 3, such as the hard disk or memory of the electronic device 3. The memory 31 may also be an external storage device of the electronic device 3, such as a plug-in hard disk equipped on the electronic device 3, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 31 may also include both the internal storage unit and the external storage device of the electronic device 3. The memory 31 is used to store computer programs and other programs and data required by the electronic device. The memory 31 may also be used to temporarily store data that has been output or is to be output.

[0096] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is used as an example. In actual applications, the above-mentioned functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment may be integrated in a processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working processes of the units and modules in the above-mentioned device can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0097] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0098] Those of ordinary skill in the art will realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled professionals can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this application.

[0099] In the embodiments provided in this application, it should be understood that the disclosed apparatus / devices and methods can be implemented in other ways. For example, the apparatus / device embodiments described above are only illustrative. For example, the division of modules or units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in electrical, mechanical or other forms.

[0100] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0101] In addition, the functional units in each embodiment of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0102] When the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, to implement all or part of the processes in the above-described embodiment methods of this application, it can also be completed by instructing relevant hardware through a computer program. The computer program of the urban on-street parking vehicle charging method can be stored in a storage medium. When the computer program is executed by a processor, it can implement the steps of the above-mentioned various method embodiments, that is, to obtain in real time the vehicle feature data of the vehicles in the parking area, where the vehicle feature data at least includes the license plate image information of the vehicles in the parking area; use a lightweight backbone network including an asymmetric convolution block to perform enhanced feature extraction on the license plate image information, and dynamically select an identification path through a classification module for decoding to obtain vehicle identification information; establish a temporary charging account according to the vehicle identification information, where the temporary charging account is associated with the vehicle entry timestamp and a preset charging rule; when a vehicle departure trigger event is detected, perform a phased deduction operation based on the stay duration in the temporary charging account; when the deduction verification passes, generate an electronic release instruction and update the parking space status database at the same time. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The storage medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice within the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the storage medium does not include electrical carrier signals and telecommunication signals.

[0103] The above embodiments are only used to illustrate the technical solutions of this application, rather than to limit them; although this application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of this application, and should all be included in the protection scope of this application. The specific implementation manners described above have further elaborated on the purpose, technical solutions and beneficial effects of this application. It should be understood that the above is only the specific implementation manner of this application, and is not used to limit the protection scope of this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this application should all be included in the protection scope of this invention.

Claims

1. A method for charging temporarily parked vehicles in a city, characterized in that, The method includes: Obtaining vehicle feature data of vehicles in the parking area in real time, where the vehicle feature data at least includes license plate image information of the vehicles in the parking area; Performing enhanced feature extraction on the license plate image information by using a lightweight backbone network including an asymmetric convolution block, and dynamically selecting an identification path through a classification module for decoding to obtain vehicle identification information; Establishing a temporary charging account according to the vehicle identification information, where the temporary charging account is associated with the vehicle entry timestamp and a preset charging rule; When a vehicle departure trigger event is detected, performing a staged deduction operation based on the stay duration in the temporary charging account; Generating an electronic release instruction when the deduction verification passes, and simultaneously updating the parking space status database.

2. The method according to claim 1, wherein The performing enhanced feature extraction on the license plate image information by using a lightweight backbone network including an asymmetric convolution block, and dynamically selecting an identification path through a classification module for decoding to obtain vehicle identification information includes: Extracting an enhanced feature map of the license plate image information through the lightweight backbone network, where the lightweight backbone network includes cascaded asymmetric convolution blocks, and each asymmetric convolution block is composed of a first convolution layer with a convolution kernel size of m×n and a second convolution layer with a convolution kernel size of j×k; Performing the following operations in parallel in the classification module: predicting license plate slicing parameters through a Softmax activation branch, where the slicing parameters are used for feature map segmentation; and calculating a double-line license plate probability value through a Sigmoid activation branch, where the probability value represents whether the current license plate is of a double-line type; According to the double-line license plate probability value and the slicing parameters, dynamically activating a feature branch in the identification module that matches the license plate type for decoding to obtain vehicle identification information.

3. The method according to claim 2, wherein The extracting an enhanced feature map of the license plate image information through the lightweight backbone network includes: Performing a first convolution and a second convolution operation on the initial feature map output by the lightweight backbone network by the first convolution layer and the second convolution layer respectively; Performing channel dimension concatenation on the first feature map and the second feature map output by the first convolution operation and the second convolution operation to generate a concatenated feature map; Using a global context enhancement module to perform spatial attention weighting on the concatenated feature map to generate a weighted feature map; Adding the weighted feature map and the initial feature map through a residual connection to obtain an enhanced feature map.

4. The method according to claim 3, wherein The using a global context enhancement module to perform spatial attention weighting on the concatenated feature map to generate a weighted feature map includes: Performing a global average pooling operation on the concatenated feature map to obtain a spatially compressed feature map; Performing L2 norm normalization processing on the spatially compressed feature map to generate a normalized feature map; Performing spatial dimension splicing on the normalized feature map and the concatenated feature map to generate the weighted feature map; Adjusting the channel dimension of the weighted feature map through a 1×1 convolution layer to make it consistent with the channel dimension of the initial feature map.

5. The method according to claim 2, wherein The predicting license plate slicing parameters through a Softmax activation branch includes: Evenly dividing the height of the initial feature map output by the lightweight backbone network into T candidate slicing intervals; Determine the optimal slice position based on the discrete classification results output by the Softmax activation branch.

6. The method according to claim 1, wherein The phased deduction based on the parking duration in the temporary billing account includes: Freeze the credit guarantee of a preset amount when the vehicle enters the lot; Update the cumulative fee according to the real-time billing cycle; Deduct the frozen credit guarantee preferentially when the vehicle leaves the lot; Trigger a secondary payment request when the cumulative fee exceeds the credit guarantee.

7. The method according to claim 6, characterized in that, It also includes an exception handling step: Start a hierarchical reminder mechanism when the vehicle's parking time exceeds the preset threshold; Generate a detention event record when it is detected that the deduction is not successful; Dynamically adjust the subsequent deduction policy based on the historical credit rating.

8. An urban temporary parking vehicle charging device, characterized in that, The device includes: An acquisition module, configured to acquire in real time the vehicle feature data of the vehicle in the parking area, where the vehicle feature data at least includes the license plate image information of the vehicle in the parking area; A processing module, configured to perform enhanced feature extraction on the license plate image information by using a lightweight backbone network including an asymmetric convolution block, and decode through a classification module to dynamically select an identification path to obtain vehicle identification information; An account establishment module, configured to establish a temporary billing account according to the vehicle identification information, where the temporary billing account is associated with the vehicle entry timestamp and a preset billing rule; A deduction module, configured to perform a phased deduction operation based on the parking duration in the temporary billing account when a vehicle departure trigger event is detected; A release module, configured to generate an electronic release instruction when the deduction verification is passed, and update the parking space status database at the same time.

9. An electronic device, the device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images