A gantry for edge computing billing and a gantry billing method for edge computing using the gantry

Abstract

A gantry for edge computing billing and a gantry billing method for edge computing using the gantry, comprising a first support column and a second support column, a crossbeam is provided between the first support column and the second support column, a maintenance installation platform is provided on the crossbeam, the maintenance installation platform comprises a U-shaped guard plate provided along the length direction of the maintenance installation platform, a plurality of fixed connecting rods are provided on the lower bottom plate of the U-shaped guard plate, a support frame is provided at the end of the fixed connecting rod, a monitoring unit is provided at the end of the support frame, and the monitoring unit is connected to the signal of the gantry industrial computer provided in the U-shaped guard plate. The U-shaped guard plate disclosed in the present application is also suitable for containing more structural units, so the edge computing components can be arranged in the U-shaped guard plate, and it is also convenient for maintenance and operation, providing convenience for the maintenance of the edge computing gantry, and does not affect normal road traffic.

Description

Technical Field

[0001] The present application relates to a gantry for edge computing billing and a gantry billing method for edge computing using the gantry. Background Art

[0002] Edge computing refers to the use of an open platform that integrates network, computing, storage, and application core capabilities to provide the nearest service on the side close to the source of objects or data. Its application is initiated on the edge side, resulting in faster network service response, meeting the basic needs of the industry in terms of real-time business, application intelligence, security and privacy protection. Edge computing is between the physical entity and the industrial connection, or on top of the physical entity. After the domestic highway toll collection was changed from end-to-end to gantry billing, how to better calculate vehicle expenses has become a research focus. Simply setting up an industrial computer at the front end of the gantry for monitoring obviously cannot meet the requirements. Summary of the invention

[0003] In order to solve the above problems, the present application discloses, on the one hand, a gantry for edge computing billing, including a first support column and a second support column, a crossbeam is provided between the first support column and the second support column, a maintenance installation platform is provided on the crossbeam, the maintenance installation platform includes a U-shaped guard plate arranged along the length direction of the maintenance installation platform, a plurality of fixed connecting rods are provided on the lower bottom plate of the U-shaped guard plate, a support frame is provided at the end of the fixed connecting rod, a monitoring unit is provided at the end of the support frame, and the monitoring unit is connected to the signal of the gantry industrial computer arranged in the U-shaped guard plate. The U-shaped guard plate disclosed in the present application itself is also suitable for holding more structural units, so the edge computing components can be arranged in the U-shaped guard plate, and it is also conducive to maintenance and operation, providing convenience for the maintenance of the edge computing gantry, and does not affect normal road traffic.

[0004] Preferably, the bottom of the connecting rod is fixedly connected to the bottom of the U-shaped guard plate, the connecting rod extends out of the U-shaped guard plate and is provided with an inner connecting plate, and the support frame is provided with an outer connecting plate fixed to the inner connecting plate.

[0005] Preferably, a plug-in plate cooperating with the inner connecting plate is provided on the U-shaped guard plate, and a groove for accommodating the inner connecting plate is provided below the plug-in plate.

[0006] Preferably, a vertical ladder is provided on the side of the first supporting column or the second supporting column, and a safety baffle is provided on the side of the U-shaped guard plate away from the vertical ladder.

[0007] On the other hand, the present application also discloses a gantry billing method based on edge computing, comprising the following steps:

[0008] Acquiring vehicle information through a monitoring unit;

[0009] Transmit vehicle information to the gantry industrial computer;

[0010] The gantry industrial computer analyzes and summarizes vehicle information;

[0011] The gantry industrial computer queries the temporary vehicle flow table through the vehicle information. If the vehicle information is found, it will be updated. If not, the vehicle information will be inserted into the temporary vehicle flow table. After the gantry working condition machine obtains information from the monitoring unit, the first step of the application is to analyze and summarize the carriages, and obtain the temporary vehicle flow table from the cloud to obtain its movement path. If it cannot be obtained, the path can be supplemented at any time.

[0012] Preferably, the vehicle information analysis and summary include summary time, gantry HEX code, gantry number, total traffic volume, total billing amount, traffic volume by vehicle type, billing amount by vehicle type, traffic volume by axle number, billing amount by axle number, traffic volume by truck weight, and billing amount; the summary time interval is not less than 5 minutes.

[0013] Preferably, the vehicle information is analyzed and summarized to form a matching record, which includes the gantry HEX code, gantry number, transaction license plate, identification license plate, billing vehicle model, capture time, transaction time, matching mark, special information, and capture information; the capture information includes the front photo and the rear photo.

[0014] Preferably, the monitoring unit obtains the vehicle information according to the following method:

[0015] Acquire an image frame containing a target vehicle through a surveillance camera;

[0016] Performing identification of the vehicle re-identification model on the image frame: performing vehicle identification from the image frame using the vehicle re-identification model to obtain a first identification domain;

[0017] Recognize the multidimensional feature model of the image frame: obtain the multidimensional features of the vehicle from the image frame using the multidimensional feature model, encode the multidimensional features, and obtain the second recognition domain by matching based on the encoded multidimensional features;

[0018] The intersection of the first identification domain and the second identification domain is obtained to obtain the target vehicle. This application adopts a combination of a re-identification model and a multi-dimensional feature model to improve the recognition accuracy, and can obtain different cargo capacities under different loading conditions of trucks, thereby improving the overall recognition accuracy, and can also be applied to large trucks with different cargo capacities for billing operations in the gantry billing system.

[0019] Preferably, it also includes a training set for training a vehicle re-identification model, wherein the vehicle re-identification training set is in the form of a triple, and each pair of training data in the triple form includes two identical vehicle images taken by different surveillance cameras and one different vehicle image; the vehicle re-identification model uses a ternary loss function and a gradient descent algorithm to calculate gradients and update network parameters on the vehicle re-identification training set, and uses a test set for testing to select a model whose recognition accuracy exceeds a re-identification threshold, so as to obtain a final vehicle re-identification model.

[0020] Preferably, the multi-dimensional feature recognition model is trained according to the following steps: establishing a vehicle data set from the acquired image frames, the vehicle data set including vehicle images and real vehicle information;

[0021] Construct a multi-dimensional feature training set and a multi-dimensional feature test set according to vehicle characteristics;

[0022] The backbone network of the multi-dimensional feature recognition model adopts the Resnet101 network as the basic feature extraction network, and adds the attention mechanism. The basic feature extraction network serves as the starting point of other attribute recognition branches. The multi-branch vehicle multi-attribute recognition network constructed by the neural network structure search technology recognizes the multi-dimensional features of the vehicle at one time; the multi-dimensional features obtained by the multi-dimensional feature recognition model for the vehicle to be identified are encoded in a hash manner to obtain the vehicle structured feature vector of the vehicle to be identified;

[0023] Also included is a database to be identified, the database to be identified including a vehicle structured feature vector of the target vehicle obtained by hash coding the target vehicle;

[0024] The cosine similarity of the vehicle structured feature vector of the vehicle to be identified and the vehicle structured feature vector of the target vehicle is calculated, and the most similar recognition results are obtained by sorting according to the size of the cosine similarity to form a second recognition domain.

[0025] This application can bring the following beneficial effects:

[0026] 1. The U-shaped guard plate disclosed in the present application is also suitable for accommodating more structural units, so the edge computing components can be set in the U-shaped guard plate, which is also conducive to maintenance and operation, and provides convenience for the inspection and maintenance of the edge computing gantry without affecting normal road traffic;

[0027] 2. After the application obtains information from the monitoring unit through the gantry condition machine, the first step is to analyze and summarize the carriages, and obtain the temporary table of vehicle flow from the cloud to obtain its movement path. If it cannot be obtained, the path can be supplemented at any time;

[0028] 3. This application adopts a combination of re-identification model and multi-dimensional feature model to improve the recognition accuracy, and can obtain different cargo capacities under different loading conditions of the truck, thereby improving the overall recognition accuracy, and can also be used for large trucks with different cargo capacities to perform billing operations in the gantry billing system. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0030] Figure 1 is a schematic diagram of the door frame;

[0031] Figure 2 It is a schematic diagram of the U-shaped guard plate part;

[0032] Figure 3 Schematic diagram of the second embodiment. DETAILED DESCRIPTION

[0033] In order to clearly illustrate the technical features of the present solution, the present application is described in detail below through specific implementation methods and in conjunction with the accompanying drawings.

[0034] In the first embodiment, if Figure 1-2 As shown, a gantry for edge computing billing includes a first support column 1 and a second support column 2, a crossbeam 3 is provided between the first support column 1 and the second support column 2, and a maintenance installation platform is provided on the crossbeam 3, the maintenance installation platform includes a U-shaped guard plate 4 arranged along the length direction of the maintenance installation platform, a plurality of fixed connecting rods 5 are provided on the lower bottom plate of the U-shaped guard plate 4, a support frame 6 is provided at the end of the fixed connecting rod 5, and a monitoring unit 7 is provided at the end of the support frame 6, and the monitoring unit 7 is connected to the gantry industrial control computer 8 arranged in the U-shaped guard plate 4 by signal. The bottom of the fixed connecting rod 5 is fixedly connected to the bottom of the U-shaped guard plate 4, the fixed connecting rod 5 extends out of the U-shaped guard plate 4 and is provided with an inner connecting plate 9, and the support frame 6 is provided with an outer connecting plate 10 fixedly arranged with the inner connecting plate 9. A plug-in board 11 is provided on the U-shaped guard plate 4 to cooperate with the inner connecting plate 9, and a groove for accommodating the inner connecting plate 9 is provided below the plug-in board 11. A vertical ladder 12 is provided on the side of the first support column 1 or the second support column 2, and a safety baffle 13 is provided on the side of the U-shaped guard plate 4 away from the vertical ladder.

[0035] When in use, place the gantry industrial computer in the U-shaped guard plate 4, place the monitoring unit 7 (usually a camera, but can also be other Internet of Things monitoring components) on the support frame 6, and then connect the gantry industrial computer with the control center by signal. When maintenance and adjustment are needed, enter the U-shaped guard plate 4 from the vertical ladder 12.

[0036] In a second embodiment, a gantry billing method based on edge computing includes the following steps:

[0037] S201. Obtain vehicle information through the monitoring unit;

[0038] The monitoring unit obtains vehicle information according to the following method: an image frame containing a target vehicle is acquired through a surveillance camera; a vehicle re-identification model is performed on the image frame: a vehicle is identified from the image frame using the vehicle re-identification model to obtain a first identification domain; a multi-dimensional feature model is performed on the image frame: a multi-dimensional feature of the vehicle is acquired from the image frame using the multi-dimensional feature model, and the multi-dimensional feature is encoded, and a second identification domain is obtained by matching based on the encoded multi-dimensional feature; and an intersection of the first identification domain and the second identification domain is obtained to obtain the target vehicle.

[0039] It also includes a training set for training a vehicle re-identification model, wherein the vehicle re-identification training set is in the form of a triple, and each pair of training data in the triple form includes two identical vehicle images taken by different surveillance cameras and one different vehicle image; the vehicle re-identification model uses a ternary loss function and a gradient descent algorithm to calculate gradients and update network parameters on the vehicle re-identification training set, and uses a test set for testing to select a model whose recognition accuracy exceeds a re-identification threshold, so as to obtain a final vehicle re-identification model.

[0040] The multidimensional feature recognition model is trained according to the following steps: a vehicle data set is established from the obtained image frames, and the vehicle data set includes vehicle images and real vehicle information; a multidimensional feature training set and a multidimensional feature test set are constructed according to vehicle features; the backbone network of the multidimensional feature recognition model adopts the Resnet101 network as the basic feature extraction network, and adds an attention mechanism. The basic feature extraction network serves as the starting point of other attribute recognition branches. The multi-branch vehicle multi-attribute recognition network constructed by the neural network structure search technology recognizes the multidimensional features of the vehicle at one time; the multidimensional features obtained by the multidimensional feature recognition model for the vehicle to be identified are encoded in a hash manner to obtain a vehicle structured feature vector of the vehicle to be identified; a database to be identified is also included, and the database to be identified includes a vehicle structured feature vector of the target vehicle obtained by hash encoding of the target vehicle; the cosine similarity of the vehicle structured feature vector of the vehicle to be identified and the vehicle structured feature vector of the target vehicle is calculated, and the most similar recognition results are obtained by sorting according to the size of the cosine similarity to form a second recognition domain.

[0041] S202. Transmitting vehicle information to the gantry industrial computer;

[0042] S203. The gantry industrial computer analyzes and summarizes vehicle information;

[0043] The vehicle information analysis and summary includes summary time, gantry HEX code, gantry number, total traffic volume, total billing amount, traffic volume by vehicle type, billing amount by vehicle type, traffic volume by axle number, billing amount by axle number, traffic volume by weight of trucks, and billing amount; the summary time interval is not less than 5 minutes. After the vehicle information and analysis and summary, a matching record is formed, and the matching record includes the gantry HEX code, gantry number, transaction license plate, identification license plate, billing vehicle type, capture time, transaction time, matching mark, special information, and capture information; the capture information includes the front photo and the rear photo.

[0044] S204. The gantry industrial computer searches the temporary vehicle flow table through the vehicle information. If the vehicle information is found, the vehicle information is updated. If the vehicle information is not found, the vehicle information is inserted into the temporary vehicle flow table.

[0045] The temporary table of vehicle flow is stored in the control center, which serves as the core for connecting and controlling the industrial computers of each gantry.

[0046] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A gantry billing method based on edge computing, applied to a gantry for edge computing billing, characterized in that: The method comprises the following steps: Acquiring vehicle information through a monitoring unit includes: acquiring an image frame containing a target vehicle through a monitoring camera; performing vehicle re-identification model recognition on the image frame: performing vehicle recognition from the image frame using the vehicle re-identification model to obtain a first recognition domain; performing multi-dimensional feature model recognition on the image frame: obtaining multi-dimensional features of the vehicle from the image frame using the multi-dimensional feature model, encoding the multi-dimensional features, and matching based on the encoded multi-dimensional features to obtain a second recognition domain; obtaining the intersection of the first recognition domain and the second recognition domain to obtain the target vehicle; Transmit vehicle information to the gantry industrial computer; The gantry industrial computer analyzes and summarizes the vehicle information; the vehicle information analysis and summary includes summary time, gantry HEX code, gantry number, total traffic volume, total billing amount, traffic volume by vehicle type, billing amount by vehicle type, traffic volume by axle number, billing amount by axle number, traffic volume by truck weight, and billing amount; the summary time interval is not less than 5 minutes; The gantry industrial computer searches the temporary vehicle flow table through the vehicle information. If the vehicle information is found, the vehicle information is updated. If the vehicle information is not found, the vehicle information is inserted into the temporary vehicle flow table. The gantry includes a first support column and a second support column, a cross beam is provided between the first support column and the second support column, a maintenance installation platform is provided on the cross beam, the maintenance installation platform includes a U-shaped guard plate arranged along the length direction of the maintenance installation platform, a plurality of fixed connecting rods are provided on the lower bottom plate of the U-shaped guard plate, a support frame is provided at the end of the fixed connecting rod, and a monitoring unit is provided at the end of the support frame, and the monitoring unit is connected to the gantry industrial control computer signal arranged in the U-shaped guard plate.

2. According to the edge computing-based gantry billing method of claim 1, it is characterized in that: After vehicle information is analyzed and summarized, a matching record is formed, which includes the gantry HEX code, gantry number, transaction license plate, identification license plate, billing vehicle model, capture time, transaction time, matching mark, special information, and capture information; the capture information includes the front photo and the rear photo.

3. According to the edge computing-based gantry billing method of claim 1, it is characterized in that: It also includes a training set for training a vehicle re-identification model, wherein the vehicle re-identification training set is in the form of a triple, and each pair of training data in the triple form includes two identical vehicle images taken by different surveillance cameras and one different vehicle image; the vehicle re-identification model uses a ternary loss function and a gradient descent algorithm to calculate gradients and update network parameters on the vehicle re-identification training set, and uses a test set for testing to select a model whose recognition accuracy exceeds a re-identification threshold, so as to obtain a final vehicle re-identification model.

4. According to the edge computing-based gantry billing method of claim 1, it is characterized in that: The multi-dimensional feature recognition model is trained according to the following steps: a vehicle data set is established from the obtained image frames, wherein the vehicle data set includes vehicle images and real vehicle information; Construct a multi-dimensional feature training set and a multi-dimensional feature test set according to vehicle characteristics; The backbone network of the multi-dimensional feature recognition model adopts the Resnet101 network as the basic feature extraction network, and adds the attention mechanism. The basic feature extraction network serves as the starting point of other attribute recognition branches. The multi-branch vehicle multi-attribute recognition network constructed by the neural network structure search technology recognizes the multi-dimensional features of the vehicle at one time. The multidimensional features obtained by the multidimensional feature recognition model for the vehicle to be identified are encoded in a hash manner to obtain a vehicle structured feature vector of the vehicle to be identified; Also included is a database to be identified, the database to be identified including a vehicle structured feature vector of the target vehicle obtained by hash coding the target vehicle; The cosine similarity of the vehicle structured feature vector of the vehicle to be identified and the vehicle structured feature vector of the target vehicle is calculated, and the most similar recognition results are obtained by sorting according to the size of the cosine similarity to form a second recognition domain.

5. According to the edge computing-based gantry billing method of claim 1, it is characterized in that: The bottom of the connecting rod is fixedly connected to the bottom of the U-shaped guard plate, the connecting rod extends out of the U-shaped guard plate and is provided with an inner connecting plate, and the support frame is provided with an outer connecting plate fixed to the inner connecting plate.

6. According to the edge computing-based gantry billing method of claim 5, it is characterized in that: A plug-in board matched with the inner connecting board is arranged on the U-shaped guard plate, and a groove for accommodating the inner connecting board is arranged below the plug-in board.

7. A gantry billing method based on edge computing according to claim 1, characterized in that: A vertical ladder is arranged on the side of the first supporting column or the second supporting column, and a safety baffle is arranged on the side of the U-shaped guard plate away from the vertical ladder.

Images