Liquefied gas filling equipment control method, system, equipment, medium and program product

Abstract

The present application provides a liquefied gas filling equipment control method, system, equipment, medium and program product, the method comprising: performing image acquisition on a target cylinder through an image acquisition device to obtain a cylinder image including a two-dimensional code printed on a cylinder shield, and obtaining cylinder verification information bound to the two-dimensional code in the cylinder image; performing feature point recognition on the cylinder image using a target recognition model to obtain coordinate information of multiple first feature points in the two-dimensional code and coordinate information of multiple second feature points in the cylinder shield; determining a target position distance based on the coordinate information of the multiple first feature points and the coordinate information of the multiple second feature points; when the target position distance is consistent with the standard position distance in the cylinder verification information, controlling the liquefied gas filling equipment to fill the target cylinder with liquefied gas; improving the accuracy of liquefied gas cylinder information verification, improving the detection rate of unqualified liquefied gas cylinders, and reducing the safety risks caused by the use of unqualified liquefied gas cylinders.

Description

Technical Field

[0001] The present invention relates to the technical field of liquefied gas filling, and in particular to a liquefied gas filling equipment control method, system, equipment, medium and program product. Background Art

[0002] In order to strengthen the safety management of liquefied gas cylinders and reduce safety hazards, relevant departments have made regulations on the use time of liquefied gas cylinders. Unqualified liquefied gas cylinders that have exceeded the use time cannot be filled with liquefied gas and put into use. Therefore, it is necessary to test the safety of liquefied gas cylinders during the liquefied gas filling stage of liquefied gas cylinders to reduce the safety risks caused by filling unqualified liquefied gas cylinders with liquefied gas and putting them into use.

[0003] The existing method of verifying the information of liquefied gas cylinders is usually to scan and identify the QR code of the liquefied gas cylinder during the filling stage, and match the scanned QR code with the pre-stored cylinder data to determine the qualification of the liquefied gas cylinder based on the matching data. However, the accuracy of this cylinder qualification detection method is low. When relevant personnel use the QR code of other qualified liquefied gas cylinders to evade detection, it is difficult to detect unqualified liquefied gas cylinders, and the safety risk is high. Summary of the invention

[0004] The present invention provides a liquefied gas filling equipment control method, system, equipment, medium and program product to solve the problem that the existing liquefied gas cylinder information verification method has low accuracy, is difficult to detect unqualified liquefied gas cylinders, and has high safety risks.

[0005] In a first aspect, an embodiment of the present application provides a method for controlling a liquefied gas filling device, comprising:

[0006] The target cylinder is imaged by an image acquisition device to obtain an image of the cylinder including a two-dimensional code printed on a cylinder shield, and the cylinder verification information bound to the two-dimensional code in the cylinder image is obtained. The cylinder verification information includes a standard position distance generated based on the positional relationship between the shield and the two-dimensional code in a standard image of the liquefied gas cylinder when it leaves the factory;

[0007] A pre-trained target recognition model is used to classify objects and identify object feature points on the cylinder image, and coordinate information of multiple first feature points in the QR code and coordinate information of multiple second feature points in the cylinder shield are obtained. The target recognition model is a feature point recognition model trained based on multiple standard images of liquefied gas cylinders and used for classifying objects and identifying object feature points on the shield of the liquefied gas cylinder and the QR code on the shield.

[0008] Determine the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points;

[0009] When the target position distance is consistent with the standard position distance, the liquefied gas filling device is controlled to unlock to fill the target cylinder with liquefied gas.

[0010] Optionally, after determining the target position distance between the two-dimensional code and the cylinder shield, the method further includes:

[0011] When the target position distance is inconsistent with the standard position distance, it is prompted that the target cylinder is an unqualified cylinder, and it is prohibited to fill the target cylinder with liquefied gas this time.

[0012] Optionally, based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, determining the target position distance between the two-dimensional code and the cylinder shield includes:

[0013] When the cylinder image includes a cylinder label with a QR code printed thereon, coordinate information of a plurality of third feature points in the cylinder label is obtained, and the coordinate information of the plurality of third feature points is obtained by performing object classification and object feature point recognition on the cylinder image using a target recognition model;

[0014] Based on the coordinate information of the plurality of third feature points, the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, a target position distance between the two-dimensional code and the cylinder shield is determined.

[0015] Optionally, based on the coordinate information of the plurality of third feature points, the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, determining the target position distance between the two-dimensional code and the cylinder shield comprises:

[0016] Based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, determine the vertical distance between the two-dimensional code and the cylinder label to obtain the label-code distance;

[0017] Based on the coordinate information of the plurality of second feature points and the coordinate information of the plurality of third feature points, determine the vertical distance between the steel cylinder shield and the steel cylinder label to obtain the shield-label distance;

[0018] Based on the label code distance and the shield label distance, the target position distance is determined.

[0019] Optionally, based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, determining the vertical distance between the two-dimensional code and the cylinder label to obtain the label code distance includes:

[0020] Based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, respectively determine the vertical coordinate difference between each first feature point and each third feature point;

[0021] The vertical coordinate difference with the smallest absolute value is taken as the vertical distance between the QR code and the cylinder label to obtain the label code distance.

[0022] Optionally, the cylinder verification information further includes a standard image of a liquefied gas cylinder bound to the QR code in the cylinder image. Before determining the target position distance between the QR code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, the method further includes:

[0023] When it is detected that the cylinder image does not include the first target object, the target position distance between the two-dimensional code and the cylinder shield is determined based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, and the first target object is an object that does not exist in the standard image of the cylinder verification information;

[0024] When it is detected that the cylinder image includes the first target object, or the cylinder image does not contain the second target object, it is prompted that the target cylinder is an unqualified cylinder, and the filling of liquefied gas to the target cylinder is prohibited this time. The second target object is an object that should exist in the standard image of the cylinder verification information.

[0025] Optionally, the cylinder verification information also includes the production date of the liquefied gas cylinder bound to the QR code in the cylinder image. Before using the pre-trained target recognition model to perform object classification and object feature point recognition on the cylinder image, the method also includes:

[0026] Determine the usage time of the liquefied gas cylinder bound to the QR code in the cylinder image based on the production date;

[0027] When the usage time is less than the calibrated service life, the pre-trained target recognition model is used to classify objects and identify object feature points on the cylinder image;

[0028] When the usage time is greater than or equal to the calibrated service life, it is indicated that the target cylinder is an unqualified cylinder, and it is prohibited to fill the target cylinder with liquefied gas this time.

[0029] In a second aspect, an embodiment of the present application provides a liquefied gas filling system, the liquefied gas filling system comprising a control device, an image acquisition device and a liquefied gas filling device, the control device being used to:

[0030] The target cylinder is imaged by an image acquisition device to obtain an image of the cylinder including a two-dimensional code printed on a cylinder shield, and the cylinder verification information bound to the two-dimensional code in the cylinder image is obtained. The cylinder verification information includes a standard position distance generated based on the positional relationship between the shield and the two-dimensional code in a standard image of the liquefied gas cylinder when it leaves the factory;

[0031] A pre-trained target recognition model is used to classify objects and identify object feature points on the cylinder image, and coordinate information of multiple first feature points in the QR code and coordinate information of multiple second feature points in the cylinder shield are obtained. The target recognition model is a feature point recognition model trained based on multiple standard images of liquefied gas cylinders and used for classifying objects and identifying object feature points on the shield of the liquefied gas cylinder and the QR code on the shield.

[0032] Determine the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points;

[0033] When the target position distance is consistent with the standard position distance, the liquefied gas filling device is controlled to unlock to fill the target cylinder with liquefied gas.

[0034] In a third aspect, an embodiment of the present application provides an electronic device, including 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 electronic device implements the above-mentioned gas tank information verification method.

[0035] In a fourth aspect, an embodiment of the present application provides a readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned gas tank information verification method is executed.

[0036] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, and when the computer program is run, the above-mentioned gas tank information verification method is executed.

[0037] In one solution provided by the above-mentioned liquefied gas filling equipment control method, system, equipment, medium and program product, an image acquisition device is used to acquire an image of a target cylinder to obtain a cylinder image including a two-dimensional code printed on a cylinder shield, and the cylinder verification information bound to the two-dimensional code in the cylinder image is obtained. The cylinder verification information includes a standard position distance generated based on the positional relationship between the shield and the two-dimensional code in the standard image of the liquefied gas cylinder when leaving the factory; a pre-trained target recognition model is used to perform object classification and object feature point recognition on the cylinder image to obtain coordinate information of multiple first feature points in the two-dimensional code and coordinate information of multiple second feature points in the cylinder shield. The target recognition model is trained based on multiple standard images of liquefied gas cylinders and is used to perform object classification and object feature point recognition on the shield of the liquefied gas cylinder and the two-dimensional code on the shield. Point recognition model; based on the coordinate information of multiple first feature points and the coordinate information of multiple second feature points, determine the target position distance between the two-dimensional code and the cylinder shield; when the target position distance is consistent with the standard position distance, control the liquefied gas filling equipment to unlock to fill the target cylinder with liquefied gas; in this embodiment, by pre-training the target recognition model and identifying the standard position distance between the two-dimensional code and the shield in the liquefied gas cylinder, and then in the liquefied gas cylinder filling stage, the target recognition model is used to identify the feature point coordinate information of the two-dimensional code and the shield, and determine the target position distance between the two-dimensional code and the cylinder shield, and fill the liquefied gas when the target position distance is consistent with the standard position distance, thereby improving the accuracy of liquefied gas cylinder information verification, improving the detection rate of unqualified liquefied gas cylinders, and reducing the safety risks caused by the use of unqualified liquefied gas cylinders. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0039] Figure 1 is a structural schematic diagram of a liquefied gas filling system in one embodiment of the present invention;

[0040] Figure 2 is a flow chart of a method for controlling a liquefied gas filling device in one embodiment of the present invention;

[0041] Figure 3 is a schematic diagram of the positions of the characteristic points of the steel cylinder image in one embodiment of the present invention;

[0042] Figure 4 yes Figure 2 A schematic diagram of an implementation process of step S30;

[0043] Figure 5 yes Figure 4 A schematic diagram of an implementation flow of step S32;

[0044] Figure 6 is another flow chart of a method for controlling a liquefied gas filling device in one embodiment of the present invention;

[0045] Figure 7 It is another flow chart of a method for controlling a liquefied gas filling device in one embodiment of the present invention;

[0046] Figure 8 yes Figure 1 A schematic diagram of the structure of the control device;

[0047] Fig. 9 It is a schematic diagram of the structure of an electronic device in one embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or their collections. It should also be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0050] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0051] References to "one embodiment" or "some embodiments" etc. described in the present specification mean that one or more embodiments of the present invention include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0052] It should be understood that the order of execution of the steps in the following embodiments does not imply a precedence of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0053] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0054] It is important to understand that in order to strengthen the safety management of liquefied gas cylinders and reduce potential safety hazards, relevant departments have set regulations on the use time of liquefied gas cylinders. Unqualified liquefied gas cylinders that have exceeded the use time cannot be filled with liquefied gas and put into use. Therefore, it is necessary to test the safety of liquefied gas cylinders during the liquefied gas filling stage to reduce the safety risks caused by unqualified liquefied gas cylinders being filled with liquefied gas and put into use.

[0055] The existing safety detection method for liquefied gas cylinders is usually to scan and identify the QR code of the liquefied gas cylinder during the filling stage, match the scanned and identified QR code with the pre-stored cylinder data, and determine the qualification of the liquefied gas cylinder based on the matching data. However, the accuracy of this cylinder qualification detection method is low. When relevant personnel use the QR code of other qualified liquefied gas cylinders to evade detection, it is difficult to detect unqualified liquefied gas cylinders, and the safety risk is relatively high. For example, in order to evade supervision, the filling personnel often print the QR codes of some qualified liquefied gas cylinders and put them in their pockets. These printed QR codes of qualified liquefied gas cylinders are called pocket codes; when filling liquefied gas, the filling staff will scan the pocket code instead of the QR code of the unqualified liquefied gas cylinder to evade qualification inspection, and illegally fill the unqualified cylinders, so that the unqualified liquefied gas cylinders can be put into use, which is easy to cause safety accidents and has a high safety risk. Therefore, it is necessary to perform visual recognition on the QR code image of the liquefied gas cylinder when filling the liquefied gas to determine the authenticity of the QR code image of the cylinder, improve the detection rate of unqualified liquefied gas cylinders, and reduce safety risks.

[0056] In addition, when filling liquefied gas, due to the complex environment and poor lighting conditions at the liquefied filling site, the scanned and collected cylinder images are prone to deformation and distortion, and the traditional visual recognition effect based on feature matching is poor, resulting in poor image recognition results, prone to false detection and missed detection, low accuracy of cylinder information verification, and low detection rate of unqualified liquefied gas cylinders.

[0057] In view of the above problems, the liquefied gas filling equipment control method provided in the embodiment of the present invention can be applied in Figure 1 In the liquefied gas filling system shown, the liquefied gas filling system includes a control device, an image acquisition device and a liquefied gas filling device, wherein the image acquisition device and the liquefied gas filling device communicate with the control device via a bus or a network.

[0058] In the liquefied gas filling stage, the control device uses an image acquisition device to acquire an image of a target cylinder (i.e., a liquefied gas cylinder that needs to be filled with liquefied gas) to obtain a cylinder image, which includes a cylinder shield and a two-dimensional code printed on the cylinder shield. In other embodiments, a cylinder label including a two-dimensional code can be installed on the cylinder shield, and the cylinder verification information bound to the two-dimensional code in the cylinder image is obtained, wherein the cylinder verification information includes a standard position distance generated based on the positional relationship between the shield and the two-dimensional code in the standard image of the liquefied gas cylinder when it leaves the factory. Then, the control device uses a pre-trained target recognition model to classify objects and identify object feature points on the cylinder image, and obtains coordinate information of multiple first feature points in the two-dimensional code and coordinate information of multiple second feature points in the cylinder shield, wherein the target recognition model is a feature point recognition model trained based on multiple standard images of liquefied gas cylinders, and is used for object classification and object feature point recognition of the shield of the liquefied gas cylinder and the two-dimensional code on the shield. Finally, the control device determines the target position distance between the QR code and the cylinder shield based on the coordinate information of multiple first feature points and the coordinate information of multiple second feature points. When the target position distance is consistent with the standard position distance, it indicates that the cylinder image is a true image of the target cylinder. If the target cylinder passes the verification, the control device controls the liquefied gas filling device to unlock to fill the target cylinder with liquefied gas.

[0059] Among them, the first feature point is the key feature point of the QR code that is obtained by the target recognition model through identifying the feature points (specific pixel points) of the QR code in the cylinder image, which can clearly characterize the position and characteristics of the QR code; the second feature point is the key feature point of the cylinder cover that is obtained by the target recognition model through identifying the feature points (specific pixel points) of the cylinder cover in the cylinder image, which can clearly characterize the position and characteristics of the cylinder cover.

[0060] In this embodiment, by pre-training the target recognition model and identifying the standard position distance between the two-dimensional code in the liquefied gas cylinder and the protective cover, then in the liquefied gas cylinder filling stage, the target recognition model is used to identify the feature point coordinate information of the two-dimensional code and the protective cover, and the target position distance between the two-dimensional code and the cylinder protective cover is determined, and the liquefied gas is filled when the target position distance is consistent with the standard position distance. The authenticity of the collected two-dimensional code image can be accurately detected, thereby identifying the situation where the two-dimensional code of a qualified liquefied gas cylinder such as a pocket code evades detection, thereby improving the accuracy of liquefied gas cylinder information verification; in addition, compared with the traditional visual feature matching method, the position distance of the feature points can also reduce the influence of factors such as complex environment and lighting changes on the recognition effect, further improving the accuracy of liquefied gas cylinder information verification, improving the detection rate of unqualified liquefied gas cylinders, and reducing the safety risks caused by the use of unqualified liquefied gas cylinders.

[0061] In this embodiment, the control device can be an electronic control unit. In other embodiments, the control device can also be a computer device, including but not limited to various personal computers, laptops, smart phones, tablet computers and other terminal devices, or it can be a server device implemented by an independent server or a server cluster composed of multiple servers.

[0062] In one embodiment, if Figure 2 As shown, a liquefied gas filling equipment control method is provided, and the method is applied in Figure 1 The liquefied gas filling system in the example is used as an example to illustrate, including the following steps:

[0063] S10: Capture an image of the target steel cylinder through an image acquisition device to obtain a steel cylinder image including a two-dimensional code printed on a steel cylinder shield, and obtain steel cylinder verification information bound to the two-dimensional code in the steel cylinder image.

[0064] Before filling a liquefied gas cylinder with liquefied gas, it is necessary to collect images of the target cylinder shield and the QR code thereon to obtain a cylinder image including the QR code printed on the cylinder shield, so as to subsequently verify the information of the target cylinder based on the collected cylinder image to ensure the authenticity of the cylinder image and the eligibility of the target cylinder. The target cylinder is a liquefied gas cylinder that currently needs to be filled with liquefied gas. When the automated liquefied gas filling system is used for liquefied gas filling, the liquefied gas cylinder placed on the filling station of the liquefied gas filling system can be directly used as the target cylinder.

[0065] The control device controls the image acquisition device of the liquefied gas filling system to acquire images of the target cylinder shield and the QR code thereon, and obtains a cylinder image including the QR code printed on the cylinder shield, that is, the cylinder image at least includes the cylinder shield and the QR code printed on the cylinder shield. The cylinder shield is a protective cap or shield on the body of the liquefied gas cylinder used to protect the angle valve and facilitate transportation.

[0066] After obtaining the cylinder image, the control device needs to sample the QR code recognition model to perform QR code recognition on the cylinder image, obtain the QR code in the cylinder image, and obtain the cylinder verification information bound to the QR code in the database. The cylinder verification information includes the standard position distance generated based on the position relationship between the shield and the QR code in the standard image of the liquefied gas cylinder when it leaves the factory. The standard image of the liquefied gas cylinder includes the shield of the liquefied gas cylinder and the QR code printed on the shield.

[0067] Among them, when liquefied gas cylinders leave the factory, it is necessary to perform standard image acquisition and verification information extraction on all liquefied gas cylinders leaving the factory, and the extracted cylinder verification information is bound to the QR code of the liquefied gas cylinder one by one and stored in the database for subsequent liquefied gas cylinder information management and monitoring.

[0068] Among them, since the image acquisition distance, acquisition angle and device image parameters will affect the imaging effect, when controlling the image acquisition device to acquire the cylinder image of the target cylinder, it is necessary to set the relevant parameters of the image acquisition device to be the same as the relevant parameters when the standard image of the liquefied gas cylinder is acquired at the factory (such as image acquisition distance, acquisition angle and device image parameters), to ensure that the acquisition and imaging effects are similar, improve the accuracy of the cylinder image, and also improve the accuracy and robustness of subsequent image-based recognition.

[0069] S20: Using a pre-trained target recognition model to perform object classification and object feature point recognition on the cylinder image, and obtain coordinate information of multiple first feature points in the two-dimensional code and coordinate information of multiple second feature points in the cylinder shield.

[0070] After obtaining the cylinder image and the cylinder verification information bound to the QR code in the cylinder image, the control device needs to call the pre-trained target recognition model, and use the pre-trained target recognition model to classify objects and identify object feature points on the cylinder image, and identify all objects in the cylinder image and the coordinate position information of each object. The coordinate position information of each object in the cylinder image includes the coordinate information of multiple first feature points in the QR code and the coordinate information of multiple second feature points in the cylinder shield.

[0071] In other embodiments, the cylinder image may further include a cylinder label with a QR code printed on it and mounted on the cylinder shield; the coordinate position information of each object in the cylinder image may further include the coordinate information of multiple third feature points in the cylinder label.

[0072] Among them, multiple first feature points in the QR code may include feature points on the contour line of the QR code; similarly, multiple second feature points in the cylinder shield may include feature points on the contour line of the cylinder shield; multiple third feature points on the cylinder label may include feature points on the contour line of the cylinder label.

[0073] For example, Figure 3 The cylinder image shown in the figure is subjected to object classification and object feature point recognition using a pre-trained target recognition model to obtain coordinate information of multiple first feature points in the cylinder shield, coordinate information of multiple second feature points in the QR code, and coordinate information of multiple third feature points on the cylinder label. The coordinate information of these feature points can be the coordinate information of the feature points on the contour lines of each object. Figure 3 As shown, the multiple second feature points in the cylinder shield can be the four feature points on the cylinder shield corresponding to the position of the cylinder label: a, b, c and d; the multiple first feature points in the QR code can be the points corresponding to the four corners of the QR code: o, e, f, and g; the multiple third feature points of the cylinder label can be the points corresponding to the four corners of the cylinder label: j, k, p, and q.

[0074] In other embodiments, the multiple first feature points in the two-dimensional code, the multiple first feature points in the cylinder shield, and the multiple third feature points in the cylinder label also include points at other positions; for example, the multiple first feature points in the two-dimensional code may also include the geometric center point of the two-dimensional code, and the multiple second feature points in the cylinder shield may also include the geometric center point of the two-dimensional code. The multiple third feature points in the cylinder label may also include the geometric center point of the cylinder label.

[0075] Among them, the target recognition model is obtained by deep learning training based on multiple standard images of liquefied gas cylinders. It is a feature point recognition model used to classify objects and recognize object feature points of the liquefied gas cylinder shields and the QR codes on the shields. By performing deep learning on multiple standard images of liquefied gas cylinders, a target recognition model that is more suitable for liquefied gas cylinder recognition scenarios can be obtained, thereby improving the accuracy of model recognition.

[0076] S30: Determine the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points.

[0077] After obtaining the coordinate information of the first feature points in the two-dimensional code in the cylinder image and the coordinate information of the second feature points in the cylinder shield, the control device determines the target position distance between the two-dimensional code in the cylinder image and the cylinder shield based on the coordinate information of the first feature points and the coordinate information of the second feature points. The coordinate information of each feature point is two-dimensional coordinate information, that is, the coordinate information of each feature point includes a horizontal coordinate (i.e., an x-axis coordinate) and a vertical coordinate (i.e., a y-axis coordinate).

[0078] Among them, the control device can directly use a certain feature point in the QR code (such as the upper left corner or upper right corner of the QR code) as the reference point to determine the difference between the vertical coordinates of the reference point and the vertical coordinates of each feature point of the liquefied gas cylinder guard among multiple feature points of the liquefied gas cylinder guard, and use the vertical coordinate difference with the smallest absolute value as the target position distance between the QR code and the cylinder guard, which is simple and convenient.

[0079] For example, a cylinder image such as Figure 3 As shown, a pre-trained target recognition model is used to perform object classification and object feature point recognition on the cylinder image to obtain the coordinate information of multiple first feature points in the two-dimensional code and the coordinate information of multiple second feature points in the cylinder shield; wherein, Figure 3 As shown, the multiple second feature points in the cylinder shield include four feature points a, b, c and d, and the multiple first feature points in the two-dimensional code include four feature points o, e, f and g. Among them, the point o in the upper left corner of the two-dimensional code can be directly used as the reference point, and the absolute value of the difference between the vertical coordinate of the reference point o and the vertical coordinate of the point b in the cylinder shield is the smallest, then the absolute value of the difference between the vertical coordinates of point o and point b is used as the target position distance between the two-dimensional code and the cylinder shield.

[0080] S40: When the target position distance is consistent with the standard position distance in the cylinder verification information, the liquefied gas filling device is controlled to unlock to fill the target cylinder with liquefied gas.

[0081] After determining the target position distance between the QR code in the cylinder image and the cylinder shield, the control device needs to determine the consistency of the target position distance and the standard position distance in the cylinder verification information. When it is determined that the target position distance is consistent with the standard position distance in the cylinder verification information, it means that the difference between the two is not large, and the QR code in the cylinder image is the real QR code of the target cylinder. At this time, the control device controls the liquefied gas filling equipment to unlock to fill the target cylinder with liquefied gas. It can identify the situation where others use other QR codes to replace the target cylinder QR code, thereby reducing the situation where the target cylinder filled with liquefied gas is an unqualified cylinder.

[0082] When the target position distance is inconsistent with the standard position distance in the cylinder verification information, it means that the difference between the two is large, the QR code position has changed, and there is a possibility that other QR codes will be used to replace the target cylinder QR code. Then the control device prompts that the target cylinder is an unqualified cylinder and prohibits the target cylinder from being filled with liquefied gas this time, thereby reducing the situation of filling unqualified cylinders with liquefied gas. For example, the control device can generate prompt information for unqualified cylinders, and then the control device uses the sound and light equipment in the liquefied gas filling system to give an audible and visual prompt of the prompt information of the unqualified cylinder.

[0083] In this embodiment, a target recognition model is pre-trained and the standard position distance between the two-dimensional code in the liquefied gas cylinder and the protective cover is obtained. Then, during the liquefied gas cylinder filling stage, the target recognition model is used to obtain the feature point coordinate information of the two-dimensional code and the protective cover, and the target position distance between the two-dimensional code and the cylinder protective cover is determined. When the target position distance is consistent with the standard position distance, the liquefied gas is filled. This improves the accuracy of the liquefied gas cylinder information verification, increases the detection rate of unqualified liquefied gas cylinders, and reduces the safety risks caused by the use of unqualified liquefied gas cylinders.

[0084] In one embodiment, before step S20, that is, before using the pre-trained target recognition model to classify objects and identify object feature points on the cylinder image, it is necessary to first perform deep training on object classification and object feature point recognition on multiple standard images of liquefied gas cylinders, the shields of the liquefied gas cylinders, and the two-dimensional codes on the shields, so as to obtain a target recognition model for subsequent image recognition of liquefied gas cylinders. The target recognition model is trained in the following manner:

[0085] SA01: Obtain standard images of multiple liquefied gas cylinders, wherein the standard image includes at least a protective cover of the liquefied gas cylinder and a QR code printed on the protective cover, and each standard image corresponds to a set of feature point coordinate data, which includes the standard coordinates of multiple standard feature points of the liquefied gas cylinder protective cover and the standard coordinates of multiple standard feature points of the QR code on the protective cover.

[0086] Among them, the standard images of multiple liquefied gas cylinders are historical images of liquefied gas cylinders collected from liquefied gas filling points at different locations. Among them, the standard image includes the shield of the liquefied gas cylinder and the cylinder label with a QR code installed on the shield, that is, the standard image at least includes the shield of the liquefied gas cylinder, the cylinder label and the QR code. Each standard image corresponds to a set of feature point coordinate data, which includes the standard coordinates of multiple standard feature points of the liquefied gas cylinder shield, the standard coordinates of multiple standard feature points of the cylinder label on the shield, and the standard coordinates of multiple standard feature points of the QR code on the cylinder label.

[0087] SA02: Input a standard image into a preset model including initial parameters, and use the preset model to classify objects and identify object feature points in the standard image to obtain the coordinate information of multiple feature points of the liquefied gas cylinder cover in the standard image and the coordinate information of multiple feature points of the QR code on the cover.

[0088] A standard image is input into a preset model including initial parameters, and the objects in the standard image are classified and the object feature points are identified by the preset model to identify all objects in the standard image and the coordinate position information of each object. The coordinate position information of each object in the standard image includes the coordinate information of multiple feature points of the hood of the liquefied gas cylinder in the standard image, the coordinate information of multiple feature points of the cylinder label on the hood, and the coordinate information of multiple feature points of the QR code in the cylinder label.

[0089] The preset model may be a convolutional neural network. A standard image is input into the preset model, and the preset model first classifies and identifies the objects in the standard image, and determines each object in the standard image, such as the shield of a liquefied gas cylinder, the cylinder label on the shield, and the QR code; then, the key feature points of each object in the standard image are identified to obtain the coordinate information of the feature points of each object in the standard image, and the coordinate information of the feature points of each object in the standard image includes the coordinate information of multiple feature points of the liquefied gas cylinder shield, the coordinate information of multiple feature points of the cylinder label on the shield, and the coordinate information of multiple feature points of the QR code.

[0090] SA03: Determine the total loss value based on the coordinate information of multiple feature points of the liquefied gas cylinder cover, the coordinate information of multiple feature points of the QR code, and the standard coordinates of multiple standard feature points of the liquefied gas cylinder cover and the standard coordinates of multiple standard feature points of the QR code in the feature point coordinate data.

[0091] Among them, based on the coordinate information of multiple feature points of the liquefied gas cylinder shield and the standard coordinates of multiple standard feature points of the liquefied gas cylinder shield, the coordinate similarity between each feature point and the standard feature point in the liquefied gas cylinder shield can be determined to obtain the coordinate similarity of multiple feature points of the liquefied gas cylinder shield, and then the coordinate similarities of the multiple feature points of the liquefied gas cylinder shield are averaged to obtain the feature point similarity of the liquefied gas cylinder shield; based on the coordinate information of multiple feature points of the two-dimensional code and the standard coordinates of multiple standard feature points of the two-dimensional code, the coordinate similarity between each feature point in the two-dimensional code and the standard feature point is determined to obtain the coordinate similarity of multiple feature points of the two-dimensional code, and then the coordinate similarities of the multiple feature points of the two-dimensional code are averaged to obtain the feature point similarity of the two-dimensional code; finally, the feature point similarity of the liquefied gas cylinder shield is added to or averaged with the feature point similarity of the two-dimensional code to obtain the total loss value.

[0092] In other embodiments, the coordinate similarity between each feature point of the cylinder label and the standard feature point can be determined based on the coordinate information of multiple feature points of the cylinder label on the shield and the standard coordinates of multiple standard feature points of the cylinder label on the shield to obtain the coordinate similarity of the multiple feature points of the cylinder label, and then the average of the coordinate similarities of the multiple feature points of the cylinder label is taken to obtain the feature point similarity of the cylinder label. Finally, the feature point similarity of the liquefied gas cylinder shield, the feature point similarity to the cylinder label, and the feature point similarity of the QR code are added or averaged to obtain the total loss value.

[0093] S04: When the total loss value does not meet the model convergence condition, continue to train the preset model based on multiple standard images to update the parameters of the iterative preset model until the total loss value meets the model convergence condition, and output the preset model with converged parameters as the target recognition model.

[0094] After obtaining the total loss value, when the total loss value does not meet the model convergence conditions, continue to train the preset model based on multiple standard images, that is, repeat steps S02-S03 to update the parameters of the iterative preset model until the total loss value meets the model convergence conditions, and output the preset model with converged parameters as the target recognition model.

[0095] Among them, when the total loss value is less than the preset value, or the number of model iterations is less than the preset number, it is determined that the total loss value does not meet the model convergence conditions; when the total loss value is greater than or equal to the preset value, or the number of model iterations reaches the preset number, it is determined that the total loss value meets the model convergence conditions.

[0096] In this embodiment, by acquiring a plurality of standard images of liquefied gas cylinders, wherein the standard image at least includes a shield of the liquefied gas cylinder and a two-dimensional code printed on the shield, each standard image corresponds to a set of feature point coordinate data, and the feature point coordinate data includes the standard coordinates of a plurality of standard feature points of the liquefied gas cylinder shield and the standard coordinates of a plurality of standard feature points of the two-dimensional code on the shield; a standard image is input into a preset model including initial parameters, and the objects in the standard image are classified and the object feature points are recognized by the preset model, so as to obtain the coordinate information of the plurality of feature points of the liquefied gas cylinder shield in the standard image, and the standard coordinates of the plurality of feature points of the shield The coordinate information of multiple feature points of the QR code above; based on the coordinate information of multiple feature points of the liquefied gas cylinder shield, the coordinate information of multiple feature points of the QR code, and the standard coordinates of multiple standard feature points of the liquefied gas cylinder shield in the feature point coordinate data, the standard coordinates of multiple standard feature points of the QR code, determine the total loss value; when the total loss value does not meet the model convergence condition, continue to train the preset model based on multiple standard images to update the parameters of the iterative preset model until the total loss value meets the model convergence condition, output the preset model with converged parameters as the target recognition model, and clarify the training process of the target recognition model. Through deep learning of target object classification and feature point recognition using multiple standard images of liquefied gas cylinders, a high-accuracy feature recognition model suitable for liquefied gas cylinder scenes can be trained, which can recognize cylinder images under different environments and lighting conditions, ensuring the stability and efficiency of visual recognition under complex backgrounds and lighting conditions.

[0097] In an optimized embodiment, in order to improve the accuracy of subsequent network recognition data, other solutions can also be used for model training. For example:

[0098] SB01: Obtain multiple standard images of liquefied gas cylinders, and annotate each standard image to obtain a label information map of each standard image, in which multiple standard feature points of the shield, multiple standard feature points of the cylinder label, and multiple standard feature points of the QR code are annotated.

[0099] SB02: Input a standard image into a preset model including initial parameters, and use the preset model to identify the object area and object feature points in the standard image to obtain an identification information map. The identification image is marked with multiple identification feature points of the protective cover, multiple identification feature points of the cylinder label, and multiple identification feature points of the QR code.

[0100] The standard image is input into a preset model including initial parameters. The preset model divides the input standard image into multiple grids, predicts the objects falling within each grid, and predicts the bounding box of the objects within each grid, wherein each grid predicts multiple bounding boxes, each bounding box contains the following information: the center coordinates (x, y) of the bounding box, the width and height (w, h) of the bounding box, and the confidence score of the object category of the bounding box, so as to identify different objects in the standard image according to the confidence score of the object category, such as the shield, cylinder label and QR code of the liquefied gas cylinder in the standard image; output the coordinate information of each bounding box, and post-process the coordinate information of these bounding boxes (such as non-maximum suppression processing), filter out the most likely feature point position of each object and mark it, and obtain a recognition information map. Among them, the recognition image is marked with multiple recognition feature points of the shield, multiple recognition feature points of the cylinder label and multiple recognition feature points of the QR code.

[0101] SB03: Calculate the confidence of the recognition information graph based on the label information graph to obtain the confidence of the recognition information graph.

[0102] After obtaining the recognition information graph including the recognition feature points of different objects, the confidence of the recognition information graph is calculated based on the label information graph to obtain the confidence of the recognition information graph.

[0103] SB04: When the confidence of the recognition information graph does not meet the model convergence conditions, the parameters of the preset model are iteratively trained based on multiple standard images, and the back propagation algorithm is used to optimize the weights of different recognition feature points during the training process until the confidence of the recognition information graph meets the model convergence conditions. The preset model with converged parameters is output as the target recognition model.

[0104] When the confidence of the recognition information graph does not meet the model convergence conditions, the parameters of the preset model are iteratively trained based on multiple standard images, that is, steps S02-S03 are repeatedly executed, and the back propagation algorithm is used to optimize the weights of different recognition feature points during the training process until the confidence of the recognition information graph meets the model convergence conditions, and the preset model with converged parameters is output as the target recognition model.

[0105] During the training process, the similarity of each object or feature point is not compared separately. Instead, the positional relationship of each object in the entire image and the recognition results of its feature points are considered to determine the confidence of the recognition result. Then, the gradient direction of the next step is determined based on the confidence of the recognition result. The possible positions of the feature points are learned through training data, and the weights of different feature points in the recognition results are optimized through the back propagation algorithm so that the positions of these feature points can be accurately predicted, which improves the accuracy of the target recognition model and the training efficiency, thereby improving the accuracy of subsequent object classification and feature point recognition using the target recognition model.

[0106] In one embodiment, if Figure 4 As shown, in step S30, based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, the target position distance between the two-dimensional code and the cylinder shield is determined, which specifically includes the following steps:

[0107] S31: When the cylinder image also includes a cylinder label printed with a two-dimensional code, coordinate information of a plurality of third feature points in the cylinder label is obtained.

[0108] After obtaining the coordinate information of multiple first feature points in the two-dimensional code in the cylinder image and the coordinate information of multiple second feature points in the cylinder shield, it is determined whether the cylinder image includes a cylinder label printed with a two-dimensional code. When the cylinder image also includes a cylinder label printed with a two-dimensional code, the coordinate information of multiple third feature points in the cylinder label is obtained, wherein the coordinate information of the multiple third feature points is obtained by performing object classification and object feature point recognition on the cylinder image using a target recognition model.

[0109] When the cylinder image does not contain a cylinder label with a QR code printed on it, determine whether the standard image in the cylinder verification information bound to the QR code in the cylinder image contains the cylinder label. If the standard image in the cylinder verification information does not contain the cylinder label, it means that the target cylinder itself does not have a cylinder label. Then, based on the coordinate information of the multiple first feature points and the coordinate information of the multiple second feature points, determine the target position distance between the QR code and the cylinder shield. Specifically, the vertical coordinate difference (i.e., the absolute value of the vertical coordinate difference) between each first feature point and each second feature point can be determined based on the coordinate information of the multiple first feature points and the coordinate information of the multiple second feature points, and the shortest distance among the vertical coordinate differences is used as the target position distance between the QR code and the cylinder shield.

[0110] If the standard image in the cylinder verification information contains a cylinder label, it means that the target cylinder itself has a cylinder label. Then the cylinder image is not the real image of the target cylinder. There may be a situation where the QR code of the liquefied gas cylinder is replaced. In this case, it is prompted that the target cylinder is an unqualified liquefied gas cylinder, and it is prohibited to fill the target cylinder with liquefied gas this time.

[0111] S32: Determine the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of third feature points, the coordinate information of the plurality of first feature points, and the coordinate information of the plurality of second feature points.

[0112] After obtaining the coordinate information of multiple third feature points in the cylinder label, the control device determines the target position distance between the QR code and the cylinder shield based on the coordinate information of the multiple third feature points, the coordinate information of the multiple first feature points and the coordinate information of the multiple second feature points.

[0113] For example, the position distance between the QR code and the cylinder label is first determined based on the coordinate information of multiple first feature points and the coordinate information of multiple third feature points, and then the position distance between the cylinder label and the cylinder cover is determined based on the coordinate information of multiple third feature points and the coordinate information of multiple second feature points. Furthermore, based on the position distance between the QR code and the cylinder label, the target position distance is determined, such as directly adding the two together as the target position distance.

[0114] In one embodiment, the target position distance may be the ratio of the label code distance to the shield label distance, wherein the label code distance is the position distance between the two-dimensional code and the cylinder label (such as the shortest distance), and the shield label distance is the position distance between the cylinder label and the cylinder shield (such as the shortest distance); the calculation process of the standard position distance in the cylinder verification information is consistent with that of the target position distance, and correspondingly, the standard position distance in the cylinder verification information may include the ratio of the label code distance to the shield label distance obtained by standard image recognition when the liquefied gas cylinder leaves the factory. The distance ratio of the two-dimensional code, the cylinder label and the cylinder shield is used as the basis for verification to judge the authenticity of the captured image of the liquefied gas cylinder. Compared with directly using the recognition distance between the three as the basis for information verification, the distance error generated by the image acquisition device capturing the same object at different distances can be reduced, the accuracy of the target position distance and the standard position distance is improved, and the accuracy of the qualified judgment of the liquefied gas can be improved.

[0115] In this embodiment, when the cylinder image also includes a cylinder label with a QR code printed on it, the coordinate information of multiple third feature points in the cylinder label is obtained, and the coordinate information of the multiple third feature points is obtained by classifying the cylinder image and identifying the object feature points using the target recognition model; based on the coordinate information of the multiple third feature points, the coordinate information of the multiple first feature points, and the coordinate information of the multiple second feature points, the target position distance between the QR code and the cylinder shield is determined. This process refines the determination process of the target position distance, adds the calculation of the coordinate information of the multiple third feature points in the cylinder label when calculating the target position distance, and improves the accuracy of the target position distance.

[0116] In one embodiment, if Figure 5As shown, in step S32, based on the coordinate information of the plurality of third feature points, the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, the target position distance between the two-dimensional code and the cylinder shield is determined, which specifically includes the following steps:

[0117] S321: Based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, determine the vertical distance between the two-dimensional code and the cylinder label to obtain the label code distance.

[0118] For example, a reference point (such as the point corresponding to the upper left corner of the QR code) can be determined from multiple first feature points in the QR code, and the vertical coordinate difference between the reference point and each third feature point can be determined to obtain multiple vertical coordinate differences. The vertical coordinate difference with the smallest absolute value can be determined among the vertical coordinate differences, and the vertical coordinate difference with the smallest absolute value can be used as the vertical distance between the QR code and the cylinder label, that is, the label code distance, which is simple and convenient.

[0119] In other embodiments, the vertical coordinate difference between each first feature point and each third feature point can be determined respectively to obtain multiple vertical coordinate differences, and the absolute values ​​of the multiple vertical coordinate differences can be averaged to obtain the average value of the absolute values ​​of the multiple vertical coordinate differences as the vertical distance between the QR code and the cylinder label, that is, the label code distance, to improve the accuracy of the label code distance and the subsequent target position distance.

[0120] S322: Based on the coordinate information of the plurality of second feature points and the coordinate information of the plurality of third feature points, determine the vertical distance between the cylinder shield and the cylinder label to obtain the shield-label distance.

[0121] For example, a reference point (such as the point corresponding to the upper left corner of the cylinder label) can be determined from multiple third feature points in the cylinder label, and the vertical coordinate difference between the reference point and each second feature point can be determined to obtain multiple vertical coordinate differences. The vertical coordinate difference with the smallest absolute value can be determined among the vertical coordinate differences, and the vertical coordinate difference with the smallest absolute value can be used as the vertical distance between the cylinder shield and the cylinder label, that is, the shield label distance, which is simple and convenient.

[0122] In other embodiments, the vertical coordinate difference between each third feature point and each second feature point can be determined respectively to obtain multiple vertical coordinate differences, and the absolute values ​​of the multiple vertical coordinate differences can be averaged to obtain the average value of the absolute values ​​of the multiple vertical coordinate differences as the vertical distance between the QR code and the cylinder label. The vertical distance between the QR code and the cylinder label is calculated using the vertical coordinate difference of the multiple feature points, and the influence of multiple feature points is taken into account, thereby reducing the distance calculation error caused by the recognition error of a single feature point when using the coordinate calculation of a single feature point, thereby improving the accuracy of the vertical distance between the QR code and the cylinder label, thereby improving the accuracy of the distance between the cylinder shield and the subsequent target position.

[0123] S323: Determine the target position distance based on the label code distance and the shield label distance.

[0124] After obtaining the tag code distance and the shield label distance, the target position distance is determined based on the tag code distance and the shield label distance. The target position distance can be the ratio of the tag code distance to the shield label distance, and the target position distance can also be the ratio of the shield label distance to the tag code distance. The relative position relationship between the QR code, the cylinder label and the cylinder shield is characterized by the relationship between the tag code distance and the shield label distance, with less data and high accuracy.

[0125] In other embodiments, after obtaining the tag code distance and the shield label distance, the sum of the tag code distance and the shield label distance can be used as the total distance, and the target position distance can be determined based on the tag code distance and the total distance. The target position distance includes the ratio of the tag code distance to the total distance.

[0126] In this embodiment, based on the coordinate information of multiple first feature points and the coordinate information of multiple third feature points, the vertical distance between the two-dimensional code and the cylinder label is determined to obtain the label code distance; based on the coordinate information of multiple second feature points and the coordinate information of multiple third feature points, the vertical distance between the cylinder shield and the cylinder label is determined to obtain the shield label distance; based on the label code distance and the shield label distance, the target position distance is determined, and the determination process of the target position distance is clarified. The relative position relationship between the two-dimensional code, the cylinder label and the cylinder shield is characterized by the relationship between the label code distance and the shield label distance, with a small amount of data and high accuracy, which improves the accuracy of the target position distance and the accuracy of subsequent liquefied gas cylinder information verification.

[0127] In one embodiment, in step S321, based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, the vertical distance between the two-dimensional code and the cylinder label is determined to obtain the label code distance, which specifically includes the following steps:

[0128] S3211: Based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, respectively determine the vertical coordinate difference between each first feature point and each third feature point.

[0129] Wherein, based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, the vertical coordinate difference between each first feature point and each third feature point is determined respectively, and a plurality of vertical coordinate difference between the first feature point and the third feature point is obtained. The vertical coordinate difference is the coordinate value difference between the vertical coordinate of the first feature point and the vertical coordinate of the third feature point.

[0130] S3212: The vertical coordinate difference between the first feature point with the smallest absolute value and the third feature point is used as the vertical distance between the two-dimensional code and the cylinder label to obtain the label code distance.

[0131] After determining the vertical coordinate difference between each first feature point and each third feature point, determine the vertical coordinate difference with the smallest absolute value among the multiple vertical coordinate differences, and use the vertical coordinate difference with the smallest absolute value among the multiple vertical coordinate differences as the vertical distance between the QR code and the cylinder label, that is, use the vertical coordinate difference between the first feature point with the smallest absolute value and the third feature point as the vertical distance between the QR code and the cylinder label to obtain the label code distance.

[0132] In other embodiments, the absolute values ​​of multiple vertical coordinate differences between the first feature point and the third feature point can also be averaged, and the average value of the absolute values ​​of multiple vertical coordinate differences between the first feature point and the third feature point can be used as the vertical distance between the QR code and the cylinder label to obtain the label code distance.

[0133] In this embodiment, based on the coordinate information of multiple first feature points and the coordinate information of multiple third feature points, the vertical distances between each first feature point and each third feature point are determined respectively, and the vertical coordinate difference between the first feature point and the third feature point with the smallest absolute value is used as the vertical distance between the QR code and the cylinder label to obtain the label code distance. The process of determining the label code distance is clarified, and the vertical coordinate difference between the first feature point and the third feature point with the smallest absolute value is used as the label code distance, which makes it easier to determine the movement of part of the QR code area more accurately.

[0134] In one embodiment, in step S322, based on the coordinate information of the plurality of second feature points and the coordinate information of the plurality of third feature points, the vertical distance between the cylinder shield and the cylinder label is determined to obtain the shield-label distance, which specifically includes the following steps:

[0135] S3221: Based on the coordinate information of the plurality of second feature points and the coordinate information of the plurality of third feature points, respectively determine the vertical coordinate difference between each second feature point and each third feature point.

[0136] Wherein, based on the coordinate information of the plurality of second feature points and the coordinate information of the plurality of third feature points, the vertical coordinate difference between each second feature point and each third feature point is determined respectively, and a plurality of vertical coordinate difference between the second feature point and the third feature point is obtained. The vertical coordinate difference is the coordinate value difference between the vertical coordinate of the second feature point and the vertical coordinate of the third feature point.

[0137] S3222: The vertical coordinate difference between the second feature point with the smallest absolute value and the third feature point is used as the vertical distance between the cylinder shield and the cylinder label to obtain the shield-label distance.

[0138] After determining the vertical coordinate difference between each second feature point and each third feature point, determine the vertical coordinate difference with the smallest absolute value among the multiple vertical coordinate differences, and use the vertical coordinate difference with the smallest absolute value among the multiple vertical coordinate differences as the vertical distance between the cylinder shield and the cylinder label. That is, use the vertical coordinate difference between the second feature point with the smallest absolute value and the third feature point as the vertical distance between the cylinder shield and the cylinder label to obtain the shield-label distance.

[0139] In other embodiments, the absolute values ​​of multiple vertical coordinate differences between the second feature point and the third feature point can also be averaged, and the average value of the absolute values ​​of multiple vertical coordinate differences between the second feature point and the third feature point can be used as the vertical distance between the cylinder shield and the cylinder label to obtain the shield-label distance.

[0140] In this embodiment, based on the coordinate information of multiple second feature points and the coordinate information of multiple third feature points, the vertical coordinate differences between each second feature point and each third feature point are determined respectively, and the vertical coordinate difference between the second feature point and the third feature point with the smallest absolute value is used as the vertical distance between the cylinder shield and the cylinder label to obtain the shield label distance. The process of determining the shield label distance is clarified, and the vertical coordinate difference between the second feature point and the third feature point with the smallest absolute value is used as the shield label distance, which makes it easier to determine the movement of part of the cylinder label area more accurately.

[0141] In one embodiment, the cylinder verification information includes a standard image of the liquefied gas cylinder bound to the QR code in the cylinder image. Figure 6 As shown, after step S10, that is, before determining the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, the method further includes the following steps:

[0142] S50: When it is detected that the cylinder image does not include the first target object, a target position distance between the two-dimensional code and the cylinder shield is determined based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points.

[0143] After using a pre-trained target recognition model to perform object classification and object feature point recognition on the cylinder image, all objects in the cylinder image are identified, and the control device needs to perform object detection on all objects in the cylinder image to determine whether the cylinder image includes the first target object and the second target object.

[0144] The control device matches all objects in the cylinder image with the first target object. When the first target object is matched, it is determined that the cylinder image includes the first target object; when the first target object is not matched, it is determined that the cylinder image does not include the first target object. At the same time, the control device matches all objects in the cylinder image with the second target object. When the second target object is matched, it is determined that the cylinder image includes the second target object; when the second target object is not matched, it is determined that the cylinder image does not include the second target object.

[0145] When it is detected that the cylinder image does not include the first target object, the target position distance between the two-dimensional code and the cylinder shield is determined based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points. In other embodiments, when it is detected that the cylinder image does not include the first target object and the cylinder image includes all the second target objects, it indicates that the cylinder image may not be a true image of the target cylinder, and the target position distance between the two-dimensional code and the cylinder shield is determined based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, which is more accurate.

[0146] The first target object is an object that does not exist in the standard image of the cylinder verification information, such as human body parts (palms, fingers), gas cylinder identification codes or other entities around the identification code labels (such as tape, ropes, etc.). The second target object is an object that should exist in the standard image of the cylinder verification information, such as gas cylinder shields, gas cylinder identification codes, identification code labels, and fixed objects on identification code labels (such as bolts, screws), etc.

[0147] S60: When it is detected that the cylinder image includes the first target object, or the cylinder image does not include the second target object, it is indicated that the target cylinder is an unqualified cylinder, and filling of the target cylinder with liquefied gas is prohibited this time.

[0148] When it is detected that the cylinder image includes the first target object, or the cylinder image does not include the second target object, it means that the cylinder image may not be a real image of the target cylinder, and the cylinder label and / or QR code is forged or replaced, then the target cylinder is prompted as an unqualified cylinder, and the target cylinder is prohibited from being filled with liquefied gas this time. There is no need for further distance identification and determination, which reduces data processing, improves the efficiency of information verification of liquefied gas cylinders, and improves the accuracy of information verification.

[0149] In this embodiment, before determining the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of multiple first feature points and the coordinate information of multiple second feature points, when it is detected that the cylinder image does not include the first target object, the target position distance between the two-dimensional code and the cylinder shield is determined based on the coordinate information of the multiple first feature points and the coordinate information of the multiple second feature points, and the first target object is an object that does not exist in the standard image of the cylinder verification information; when it is detected that the cylinder image includes the first target object, or the cylinder image does not contain the second target object, it is prompted that the target cylinder is an unqualified cylinder, and the filling of liquefied gas to the target cylinder is prohibited this time, and the second target object is an object that should exist in the standard image of the cylinder verification information, which can reduce data processing, improve the information verification efficiency of liquefied gas cylinders, and improve the accuracy of information verification.

[0150] In one embodiment, the cylinder verification information also includes the production date of the liquefied gas cylinder bound to the QR code in the cylinder image. Figure 7 As shown, before step S20, that is, before using the pre-trained target recognition model to perform object classification and object feature point recognition on the cylinder image, the method further includes the following steps:

[0151] S70: Determine the usage time of the liquefied gas cylinder bound to the QR code in the cylinder image based on the production date of the liquefied gas cylinder in the cylinder verification information.

[0152] Before using the pre-trained target recognition model to classify objects and identify object feature points in the cylinder image, the control device needs to obtain the cylinder verification information of the liquefied gas cylinder bound to the QR code in the cylinder image, and the cylinder verification information also includes the production date of the liquefied gas cylinder bound to the QR code in the cylinder image. The control device needs to determine the usage time of the liquefied gas cylinder bound to the QR code in the cylinder image based on the production date and current date of the liquefied gas cylinder in the cylinder verification information, and confirm whether the usage time is greater than the calibrated service life of liquefied gas cylinders of this type or manufacturer.

[0153] S80: When the usage time is less than the calibrated service life, a pre-trained target recognition model is used to perform object classification and object feature point recognition on the cylinder image.

[0154] When the usage time is less than the calibrated service life, a pre-trained target recognition model is used to perform object classification and object feature point recognition on the cylinder image, that is, step S20 to step S40 are executed to ensure the accuracy of subsequent distance data verification.

[0155] S90: When the usage time is greater than or equal to the calibrated service life, it is indicated that the target cylinder is an unqualified cylinder and it is prohibited to fill the target cylinder with liquefied gas this time.

[0156] When the usage time is greater than or equal to the calibrated service life, the target cylinder is prompted as an unqualified cylinder, and the target cylinder is prohibited from being filled with liquefied gas this time, which can reduce data processing, improve the information verification efficiency of liquefied gas cylinders, and improve the accuracy of information verification.

[0157] In this embodiment, the usage time of the liquefied gas cylinder bound to the QR code in the cylinder image is determined based on the production date. When the usage time is less than the calibrated service life, a pre-trained target recognition model is used to perform object classification and object feature point recognition on the cylinder image; when the usage time is greater than or equal to the calibrated service life, it is prompted that the target cylinder is an unqualified cylinder, and the filling of liquefied gas to the target cylinder is prohibited this time, which can reduce data processing, improve the information verification efficiency of the liquefied gas cylinder, and improve the accuracy of information verification.

[0158] In other embodiments, the cylinder verification information also includes the inspection date (i.e., the date on which the most recent safety inspection was completed), the next inspection date (i.e., the date on which the safety inspection is required), the manufacturer, the scrapping date of the liquefied gas cylinder bound to the QR code in the cylinder image, as well as the pressure, wall thickness, volume, calibrated weight and other information of the cylinder.

[0159] Before step S20, that is, before using the pre-trained target recognition model to classify objects and identify object feature points on the steel cylinder image, if it is determined that the current date does not exceed the inspection date and / or scrap date in the steel cylinder verification information, and the manufacturer in the steel cylinder verification information is a pre-calibrated qualified manufacturer, and the pressure, wall thickness, volume and weight of the target steel cylinder are detected to be consistent with the pressure, wall thickness, volume and calibrated weight in the steel cylinder verification information, it means that the basic information of the target steel cylinder meets the qualification test requirements and it is necessary to further judge the authenticity of the steel cylinder image. Then, the pre-trained target recognition model is used to classify objects and identify object feature points on the steel cylinder image, that is, execute steps S20 to S40 to ensure the accuracy of the subsequent distance data verification. After the distance data verification is passed, it is determined that the steel cylinder image is the true image of the target steel cylinder. Only then can it be determined that the target steel cylinder is a qualified steel cylinder, and the liquefied gas filling equipment can be controlled to fill the target steel cylinder with liquefied gas, reducing the situation where unqualified liquefied gas cylinders are filled with liquefied gas and put into use, thereby improving safety.

[0160] If it is determined that the current date exceeds the next inspection date and / or scrapping date in the cylinder verification information, or the manufacturer in the cylinder verification information is not a pre-calibrated qualified manufacturer, and the pressure, wall thickness, volume and weight of the target cylinder are detected to be inconsistent with any of the pressure, wall thickness, volume and calibrated weight in the cylinder verification information, indicating that the basic information of the target cylinder does not meet the qualification test requirements, there is no need to further judge the authenticity of the cylinder image, and the target cylinder is directly prompted as an unqualified cylinder, and the target cylinder is prohibited from being filled with liquefied gas this time, thereby reducing the situation where unqualified liquefied gas cylinders are filled with liquefied gas and put into use, thereby improving safety.

[0161] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0162] In one embodiment, a control device is provided, which corresponds one-to-one to the control method of the liquefied gas filling device in the above embodiment. Figure 8 As shown, the control device includes a collection module 801, an identification module 802, a determination module 803 and a control module 804. Each functional module is described in detail as follows:

[0163] The acquisition module 801 is used to acquire an image of the target cylinder through an image acquisition device, obtain an image of the cylinder including a two-dimensional code printed on a cylinder shield, and obtain the cylinder verification information bound to the two-dimensional code in the cylinder image, wherein the cylinder verification information includes a standard position distance generated based on the positional relationship between the shield and the two-dimensional code in a standard image of the liquefied gas cylinder when leaving the factory;

[0164] The recognition module 802 is used to classify objects and identify object feature points on the cylinder image using a pre-trained target recognition model to obtain coordinate information of multiple first feature points in the two-dimensional code and coordinate information of multiple second feature points in the cylinder shield. The target recognition model is a feature point recognition model trained based on multiple standard images of liquefied gas cylinders and used for classifying objects and identifying object feature points on the shield of the liquefied gas cylinder and the two-dimensional code on the shield.

[0165] A determination module 803, for determining a target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points;

[0166] The control module 804 is used to control the liquefied gas filling device to unlock so as to fill the target cylinder with liquefied gas when the target position distance is consistent with the standard position distance.

[0167] Optionally, the control device further includes a prompt module 805, which is used to, after determining the target position distance between the two-dimensional code and the cylinder shield:

[0168] When the target position distance is inconsistent with the standard position distance, it is prompted that the target cylinder is an unqualified cylinder, and it is prohibited to fill the target cylinder with liquefied gas this time.

[0169] Optionally, the determination module 803 is specifically configured to:

[0170] When the cylinder image includes a cylinder label with a QR code printed thereon, coordinate information of a plurality of third feature points in the cylinder label is obtained, and the coordinate information of the plurality of third feature points is obtained by performing object classification and object feature point recognition on the cylinder image using a target recognition model;

[0171] Based on the coordinate information of the plurality of third feature points, the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, a target position distance between the two-dimensional code and the cylinder shield is determined.

[0172] Optionally, the determining module 803 is further configured to:

[0173] Based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, determine the vertical distance between the two-dimensional code and the cylinder label to obtain the label-code distance;

[0174] Based on the coordinate information of the plurality of second feature points and the coordinate information of the plurality of third feature points, determine the vertical distance between the steel cylinder shield and the steel cylinder label to obtain the shield-label distance;

[0175] Based on the label code distance and the shield label distance, the target position distance is determined.

[0176] Optionally, the determining module 803 is further configured to:

[0177] Based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, respectively determine the vertical coordinate difference between each first feature point and each third feature point;

[0178] The vertical coordinate difference with the smallest absolute value is taken as the vertical distance between the QR code and the cylinder label to obtain the label code distance.

[0179] Optionally, the cylinder verification information also includes a standard image of the liquefied gas cylinder bound to the QR code in the cylinder image:

[0180] The determination module 803 is also used to, before determining the target position distance between the two-dimensional code and the cylinder shield, determine the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points when it is detected that the cylinder image does not include the first target object, wherein the first target object is an object that does not exist in the standard image of the cylinder verification information;

[0181] The prompt module 805 is also used to, before determining the target position distance between the QR code and the cylinder shield, when it is detected that the cylinder image includes the first target object, or the cylinder image does not contain the second target object, prompt that the target cylinder is an unqualified cylinder and prohibit the target cylinder from being filled with liquefied gas this time. The second target object is an object that should exist in the standard image of the cylinder verification information.

[0182] Optionally, the cylinder verification information also includes the production date of the liquefied gas cylinder bound to the QR code in the cylinder image;

[0183] The determination module 803 is also used to determine the usage time of the liquefied gas cylinder bound to the QR code in the cylinder image based on the production date before using the pre-trained target recognition model to perform object classification and object feature point recognition on the cylinder image;

[0184] The prompt module 805 is also used to prompt that the target cylinder is an unqualified cylinder and prohibit the target cylinder from being filled with liquefied gas when the usage time is greater than or equal to the calibrated usage period.

[0185] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0186] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned function allocation can be completed by different functional units and modules based on needs, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0187] The embodiment of the present application also provides an electronic device, which may be an electronic control unit, a terminal device or a server. Fig. 9 As shown, the electronic device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps in any of the above-mentioned method embodiments when executing the computer program, or implements the functions of the modules / units in the above-mentioned device embodiments when executing the computer program.

[0188] Those skilled in the art will understand that Fig. 9 These are merely examples of electronic devices and do not constitute a limitation of the electronic device. The electronic device may include more or fewer components than those shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0189] The processor may be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, field programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0190] The memory may be an internal storage unit of an electronic device, such as a hard disk or memory of the electronic device. The memory may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart memory card, a secure digital card, a flash memory card, etc. equipped on the electronic device. Furthermore, the memory may also include both an internal storage unit of the electronic device and an external storage device.

[0191] An embodiment of the present application also provides a readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned liquefied gas tank qualification detection method is executed.

[0192] An embodiment of the present application provides a computer program product, which, when executed on an electronic device, enables the above-mentioned liquefied gas tank qualification detection method to be executed.

[0193] If the integrated 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 computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, which can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to the camera device / terminal device, recording medium, computer memory, read-only memory, random access memory, electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk. In some jurisdictions, based on legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0194] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0195] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0196] In the embodiments provided in the present application, it should be understood that the disclosed devices / equipment and methods can be implemented in other ways. For example, the device / equipment embodiments described above are merely schematic, for example, the division of the modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0197] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected based on actual needs to achieve the purpose of the solution of this embodiment.

[0198] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for controlling liquefied gas filling equipment, characterized in that: include: Capturing the target cylinder through an image acquisition device, obtaining a cylinder image including a two-dimensional code printed on a cylinder shield, and obtaining cylinder verification information bound to the two-dimensional code in the cylinder image, wherein the cylinder verification information includes a standard position distance generated based on a positional relationship between the shield and the two-dimensional code in a standard image of a liquefied gas cylinder when leaving the factory; A pre-trained target recognition model is used to perform object classification and object feature point recognition on the cylinder image to obtain coordinate information of multiple first feature points in the two-dimensional code and coordinate information of multiple second feature points in the cylinder shield. The target recognition model is a feature point recognition model trained based on multiple standard images of the liquefied gas cylinder and used for object classification and object feature point recognition of the shield of the liquefied gas cylinder and the two-dimensional code on the shield. Determine the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of the first feature points and the coordinate information of the plurality of the second feature points; When the target position distance is consistent with the standard position distance, controlling the liquefied gas filling device to unlock so as to fill the target cylinder with liquefied gas; Wherein, determining the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points includes: When the cylinder image includes a cylinder label printed with the two-dimensional code, coordinate information of a plurality of third feature points in the cylinder label is obtained, and the coordinate information of the plurality of third feature points is obtained by performing object classification and object feature point recognition on the cylinder image using the target recognition model; Based on the coordinate information of the plurality of third feature points, the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, the target position distance between the two-dimensional code and the cylinder shield is determined.

2. The liquefied gas filling equipment control method according to claim 1, characterized in that: The determining the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of third feature points, the coordinate information of the plurality of first feature points, and the coordinate information of the plurality of second feature points comprises: Based on the coordinate information of the first feature points and the coordinate information of the third feature points, determine the vertical distance between the two-dimensional code and the cylinder label to obtain the label code distance; Based on the coordinate information of the plurality of second feature points and the coordinate information of the plurality of third feature points, determining the vertical distance between the cylinder shield and the cylinder label to obtain the shield-label distance; The target position distance is determined based on the tag code distance and the shield tag distance.

3. The liquefied gas filling equipment control method according to claim 2, characterized in that: The determining the vertical distance between the two-dimensional code and the cylinder label based on the coordinate information of the plurality of the first feature points and the coordinate information of the plurality of the third feature points to obtain the label code distance includes: Based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of third feature points, respectively determine the vertical coordinate difference between each of the first feature points and each of the third feature points; The vertical coordinate difference with the smallest absolute value is used as the vertical distance between the two-dimensional code and the cylinder label to obtain the label code distance.

4. The liquefied gas filling equipment control method according to any one of claims 1 to 3, characterized in that: The cylinder verification information also includes a standard image of the liquefied gas cylinder bound to the two-dimensional code in the cylinder image. Before determining the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points, the method further includes: When it is detected that the cylinder image does not include the first target object, the target position distance between the two-dimensional code and the cylinder shield is determined based on the coordinate information of multiple first feature points and the coordinate information of multiple second feature points. The first target object is an object that does not exist in the standard image of the cylinder verification information.

5. The liquefied gas filling equipment control method according to any one of claims 1 to 3, characterized in that: The cylinder verification information also includes the production date of the liquefied gas cylinder bound to the two-dimensional code in the cylinder image. Before using the pre-trained target recognition model to perform object classification and object feature point recognition on the cylinder image, the method also includes: Determine the usage time of the liquefied gas cylinder bound to the QR code in the cylinder image based on the production date; When the usage time is less than the calibrated usage period, the pre-trained target recognition model is used to perform object classification and object feature point recognition on the cylinder image.

6. A liquefied gas filling system, characterized in that: The liquefied gas filling system includes a control device, an image acquisition device and a liquefied gas filling device, wherein the control device is used to: The target cylinder is imaged by the image acquisition device to obtain an image of the cylinder including a two-dimensional code printed on a cylinder shield, and the cylinder verification information bound to the two-dimensional code in the cylinder image is obtained, wherein the cylinder verification information includes a standard position distance generated based on the positional relationship between the shield and the two-dimensional code in a standard image of the liquefied gas cylinder when it leaves the factory; A pre-trained target recognition model is used to perform object classification and object feature point recognition on the cylinder image to obtain coordinate information of multiple first feature points in the two-dimensional code and coordinate information of multiple second feature points in the cylinder shield. The target recognition model is a feature point recognition model trained based on multiple standard images of the liquefied gas cylinder and used for object classification and object feature point recognition of the shield of the liquefied gas cylinder and the two-dimensional code on the shield. Determining a target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of first feature points and the coordinate information of the plurality of second feature points includes: When the cylinder image includes a cylinder label printed with the two-dimensional code, coordinate information of a plurality of third feature points in the cylinder label is obtained, and the coordinate information of the plurality of third feature points is obtained by performing object classification and object feature point recognition on the cylinder image using the target recognition model; Determine the target position distance between the two-dimensional code and the cylinder shield based on the coordinate information of the plurality of third feature points, the coordinate information of the plurality of first feature points, and the coordinate information of the plurality of second feature points; When the target position distance is consistent with the standard position distance, the liquefied gas filling device is controlled to unlock to fill the target cylinder with liquefied gas.

7. An electronic 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, the electronic device implements the liquefied gas filling equipment control method according to any one of claims 1 to 5.

8. A readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the liquefied gas filling equipment control method according to any one of claims 1 to 5 is executed.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed, the liquefied gas filling equipment control method according to any one of claims 1 to 5 is executed.

Images