A method and device for identifying a dial value of an analog instrument

Abstract

The application provides a dial value recognition method and device of a pointer instrument, and the method comprises the following steps: acquiring a dial image of the pointer instrument; inputting the dial image into a region recognition model to determine the position of a dial scale strip region in the dial image; extracting a multi-scale feature map from the inference result of the region recognition model, and extracting a first dial feature image and a second dial feature image from a target feature map with the largest scale in the multi-scale feature map based on the determined dial scale strip region and the target feature map; inputting the first dial feature image and the second dial feature image into a key point recognition model to determine four key points; and determining the dial value of the pointer instrument according to the determined four key points. The effect of accurately determining the dial reading of the instrument in the presence of strong interference is achieved.

Description

Technical Field

[0001] This application relates to the field of image recognition technology, and more specifically, to a method and apparatus for recognizing the value of a pointer-type instrument. Background Technology

[0002] Pointer instruments come in a wide variety of types, including common ones such as oil level gauges, sulfur hexafluoride pressure gauges, barometers, and thermometers. The shapes and measuring ranges of various types of pointer instruments differ, and even instruments of the same type and function can vary in appearance depending on the manufacturer. Furthermore, even for the same instrument, images taken at different times can be affected by factors such as varying lighting conditions, shooting angles, and the degree of obstruction and blurring caused by inclement weather. These factors severely interfere with dial readings and reading recognition, making it impossible to accurately interpret the instrument's readings. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method and apparatus for identifying the meter reading of a pointer instrument, which can determine the meter reading of the pointer instrument by recognizing and processing the dial image, thus solving the problem in the prior art that the meter reading cannot be correctly identified under the presence of multiple interferences, and achieving the effect of accurately determining the meter reading even under the presence of strong interference.

[0004] In a first aspect, embodiments of this application provide a method for recognizing the meter reading of a pointer-type instrument. The method includes: acquiring a dial image of the pointer-type instrument; inputting the dial image into a region recognition model to determine the position of the dial scale strip region in the dial image; extracting a multi-scale feature map from the inference result of the region recognition model, and based on the determined dial scale strip region and the target feature map of the largest scale in the multi-scale feature map, extracting a first dial feature image and a second dial feature image from the target feature map, wherein the first dial feature image includes regions other than the dial scale strip region, and the second dial feature image includes the dial scale strip region; inputting the first dial feature image and the second dial feature image into a key point recognition model to determine four key points; and determining the meter reading of the pointer-type instrument based on the determined four key points.

[0005] Optionally, the region recognition model is trained as follows: Multiple subsets of region location data are determined from a pre-set region location dataset. Each subset includes a training image and the location of a dial scale strip region within that training image. For each subset, a training image from that subset is input into the region recognition model to obtain the predicted location of the dial scale strip. For each subset, a first degree of overlap between the predicted location of the dial scale strip and the location of the dial scale strip region is calculated. Based on the first degree of overlap, the loss function value of the region recognition model is calculated. Based on the obtained loss function value, the initial region recognition model is trained until the loss function value of the initial region recognition model converges to a minimum. The initial region recognition model whose loss function value converges to a minimum is determined as the trained target region recognition model.

[0006] Optionally, the keypoint recognition model is trained as follows: Multiple keypoint location data subsets are determined from a pre-set keypoint location dataset. Each keypoint location data subset includes a training image and the locations of four keypoints in the training image. For each keypoint location data subset, one training image from that subset is input into the region recognition model to obtain a first dial feature map and a second dial feature map. The first and second dial feature maps are input into the keypoint recognition model to obtain the predicted locations of four keypoints. For each region location data subset, a second overlap degree is calculated between the predicted locations of the four keypoints and the locations of the four keypoints. Based on the second overlap degree, the loss function value of the keypoint recognition model is calculated. Based on the obtained loss function value, the initial keypoint recognition model is trained until the loss function value of the initial keypoint recognition model converges to its minimum value.

[0007] The initial keypoint recognition model that converges to the minimum value of the loss function is determined as the trained target keypoint recognition model.

[0008] Optionally, the first dial feature image is extracted from the target feature map in the following manner: based on the dial scale strip area, the target area of ​​the dial scale strip area is determined in the multi-scale feature map; the target area is mapped onto the target feature map at the largest scale to obtain a first superimposed image; the target area in the first superimposed image is set to zero to obtain the first dial feature image.

[0009] Optionally, the second dial feature image is extracted from the target feature map in the following manner: based on the dial scale strip area, the target area of ​​the dial scale strip area is determined in the multi-scale feature map; the target area is mapped onto the target feature map at the largest scale to obtain a second superimposed image; the area outside the target area in the second superimposed image is set to zero to obtain the second dial feature image.

[0010] Optionally, the four key points include a first key point, a second key point, a third key point, and a fourth key point, wherein the first key point is used to indicate the position of the pointer tip in the dial image, the second key point is used to indicate the position of the pointer center in the dial image, the third key point is used to indicate the position of the scale head in the dial image, and the fourth key point is used to indicate the position of the scale tail in the dial image.

[0011] Optionally, the key point recognition model determines four key points from the first dial feature image and the second dial feature image in the following manner: A first convolution calculation, a second convolution calculation, and a third convolution calculation are performed on the first dial feature image respectively to obtain three first dial key point images. The first dial key point image obtained through the first convolution calculation is used to determine the second key point, the first dial key point image obtained through the second convolution calculation is used to determine the third key point, and the first dial key point image obtained through the third convolution calculation is used to determine the fourth key point; a fourth convolution calculation is performed on the second dial feature image. The process involves obtaining a second dial keypoint image, wherein the second dial keypoint image obtained through a fourth convolution is used to determine the first keypoint; the pixel with the highest feature value in the first dial keypoint image obtained through the first convolution is determined as the second keypoint; the pixel with the highest feature value in the first dial keypoint image obtained through the second convolution is determined as the third keypoint; the pixel with the highest feature value in the first dial keypoint image obtained through the third convolution is determined as the fourth keypoint; and the pixel with the highest feature value in the second dial keypoint image obtained through the fourth convolution is determined as the first keypoint.

[0012] Optionally, the step of determining the meter value of the pointer instrument based on the four determined key points includes: determining the ratio of the first key point between the third key point and the fourth key point based on the first key point, the second key point, the third key point, and the fourth key point; and determining the meter value of the pointer instrument based on the ratio and the dial range of the pointer instrument.

[0013] Optionally, the step of determining the proportion of the first key point between the third and fourth key points based on the first, second, third, and fourth key points includes: determining the pointer tip rotation angle value based on the third, fourth, and first key points; determining the range angle value based on the second, third, and fourth key points; calculating the angle ratio between the pointer tip rotation angle value and the range angle value; and determining the proportion of the first key point between the third and fourth key points based on the angle ratio.

[0014] Secondly, embodiments of this application also provide a meter reading recognition device for a pointer-type instrument, the device comprising:

[0015] The dial image acquisition module is used to acquire dial images of analog instruments;

[0016] The region recognition model calculation module is used to input the dial image into the region recognition model to obtain the position of the dial scale strip region in the dial image;

[0017] The feature image extraction module is used to extract multi-scale feature maps from the region recognition model, and based on the determined dial scale strip region and the target feature map of the largest scale in the multi-scale feature map, extract a first dial feature image and a second dial feature image from the target feature map. The first dial feature image includes other regions besides the dial scale strip region, and the second dial feature image includes the dial scale strip region.

[0018] The key point recognition model calculation module is used to input the first dial feature map and the second dial feature map into the key point recognition model to determine four key points;

[0019] The meter value determination module is used to determine the meter value of the pointer-type instrument based on four determined key points.

[0020] Thirdly, embodiments of this application also provide an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the above-described pointer-type instrument value recognition method.

[0021] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the above-described pointer-type instrument value recognition method.

[0022] The present application provides a method and apparatus for identifying the meter reading of a pointer-type instrument. This method and apparatus can determine the meter reading of the pointer-type instrument by recognizing and processing the dial image. This solves the problem in the prior art that the meter reading cannot be correctly identified under the presence of multiple interferences, and achieves the effect of accurately determining the meter reading even under strong interference.

[0023] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 A flowchart illustrating a method for identifying the value of a pointer-type instrument provided in an embodiment of this application;

[0026] Figure 2 A schematic diagram of a pointer-type instrument provided in an embodiment of this application;

[0027] Figure 3 A schematic diagram illustrating the steps of extracting a first dial feature image and a second dial feature image from the dial image provided in this application embodiment;

[0028] Figure 4 A schematic diagram of the structure of a pointer-type instrument value recognition device provided in an embodiment of this application;

[0029] Figure 5 This application provides a schematic diagram of the structure of an electronic device. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0031] First, the applicable scenarios for this application will be introduced. This application can be applied to the field of image recognition technology.

[0032] Research has revealed a wide variety of pointer-type instruments, including common ones such as oil level gauges, sulfur hexafluoride pressure gauges, barometers, and thermometers. These different types of pointer instruments vary in shape and range; even instruments of the same type and function can differ in appearance due to different manufacturers. Furthermore, even for the same instrument, images taken at different times can vary due to factors such as lighting conditions, shooting angles, and the degree of obstruction and blurring caused by inclement weather. These factors severely interfere with dial readings and reading recognition, making it impossible to accurately interpret the instrument's readings.

[0033] Based on this, the present application provides a method and apparatus for identifying the meter reading of a pointer-type instrument. It can determine the meter reading of the pointer-type instrument by recognizing and processing the dial image, which solves the problem in the prior art that the meter reading cannot be correctly identified under the presence of multiple interferences, and achieves the effect of accurately determining the meter reading even under the presence of strong interference.

[0034] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for identifying the value of a pointer-type instrument, as provided in an embodiment of this application. Figure 1 As shown in the figure, the meter value identification method for pointer-type instruments provided in this application includes:

[0035] S101. Obtain the dial image of the pointer-type instrument.

[0036] Here, the dial image of the pointer instrument can be obtained in a variety of ways, such as by using a camera set above the pointer instrument or a moving image acquisition device.

[0037] For example, please refer to Figure 2 , Figure 2 This is a schematic diagram of a pointer-type instrument provided in an embodiment of this application. Figure 2 As shown in the figure, the pointer-type instrument provided in this application embodiment includes: pointer tip recognition point D, pointer center recognition point C, scale head recognition point A, and scale tail recognition point B, wherein the shaded area is the dial scale strip area.

[0038] S102. Input the dial image into the region recognition model to determine the position of the dial scale strip region in the dial image.

[0039] Specifically, the region recognition model can be trained as follows: Multiple subsets of region location data are determined from a pre-set region location dataset. Each subset includes a training image and the location of the dial scale strip region within that training image. For each subset, one training image from that subset is input into the region recognition model to obtain the predicted location of the dial scale strip. For each subset, a first degree of overlap between the predicted location of the dial scale strip and the location of the actual dial scale strip region is calculated. Based on the first degree of overlap, the loss function value of the region recognition model is calculated. Based on the obtained loss function value, the initial region recognition model is trained until the loss function value of the initial region recognition model converges to a minimum. The initial region recognition model whose loss function value converges to a minimum is determined as the completed target region recognition model.

[0040] It should be noted that after the region recognition model receives the dial image, it can obtain four feature maps at different scales. The feature values ​​of each pixel in two feature maps at different scales are compared, and the pixel with the higher feature value is retained in the two feature maps. In this way, one scale feature map is a feature map of the striped region, and the other scale feature map is a feature map without the striped region.

[0041] In this way, the location of the strip region can be determined.

[0042] For example, the region identification model can be the HRNet model.

[0043] S103. Extract multi-scale feature maps from the inference results of the region recognition model, and extract the first dial feature image and the second dial feature image from the target feature map based on the determined dial scale strip region and the target feature map of the largest scale in the multi-scale feature map.

[0044] The first dial feature image includes areas other than the dial scale strip area, and the second dial feature image includes the dial scale strip area.

[0045] Specifically, the first dial feature image can be extracted from the target feature map in the following way: based on the dial scale strip area, the target area of ​​the dial scale strip area is determined in the multi-scale feature map; the target area is mapped onto the target feature map of the largest scale to obtain a first superimposed image; the target area in the first superimposed image is set to zero to obtain the first dial feature image.

[0046] Specifically, the second dial feature image can be extracted from the target feature map in the following way: based on the dial scale strip area, the target area of ​​the dial scale strip area is determined in the multi-scale feature map; the target area is mapped onto the target feature map of the largest scale to obtain a second superimposed image; the area outside the target area in the second superimposed image is set to zero to obtain the second dial feature image.

[0047] For example, please refer to Figure 3 , Figure 3 This is a schematic diagram illustrating the steps for extracting a first dial feature image and a second dial feature image from a dial image, provided as an embodiment of this application. Figure 3 As shown in the figure, the step of extracting the first dial feature image and the second dial feature image from the dial image provided in this application embodiment includes: dial image 301, dial scale strip area 302, superimposed image 304, target feature image 303, first dial feature image 305, and second dial feature image 306.

[0048] For example, the dial image is input into the region recognition model to obtain the dial scale strip region 302; the dial scale strip region 302 is projected onto the target feature image 303 to obtain the superimposed image 304; the target region in the superimposed image is set to zero to obtain the first dial feature image 305; the region in the superimposed image outside the target region is set to zero to obtain the second dial feature image 306.

[0049] S104. Input the first dial feature map and the second dial feature map into the key point recognition model to determine four key points.

[0050] The four key points include the first key point, the second key point, the third key point, and the fourth key point.

[0051] For details, please refer to Figure 2 The first key point is used to indicate the position of the pointer tip in the dial image, i.e. Figure 2 Point D in the image; the second key point is used to indicate the position of the pointer center in the dial image, i.e. Figure 2Point C in the diagram; the third key point is used to indicate the position of the scale head in the dial image, i.e. Figure 2 Point A in the image; the fourth key point is used to indicate the position of the scale tail in the dial image, i.e. Figure 2 Point B in the diagram.

[0052] Specifically, the key point recognition model determines four key points from the first and second dial feature images in the following way: First, a first convolution calculation, a second convolution calculation, and a third convolution calculation are performed on the first dial feature image to obtain three first dial key point images. The first dial key point image obtained through the first convolution calculation is used to determine the second key point, the first dial key point image obtained through the second convolution calculation is used to determine the third key point, and the first dial key point image obtained through the third convolution calculation is used to determine the fourth key point; a fourth convolution calculation is performed on the second dial feature image. The process involves obtaining a second dial keypoint image, wherein the second dial keypoint image obtained through a fourth convolution is used to determine the first keypoint; the pixel with the highest feature value in the first dial keypoint image obtained through the first convolution is determined as the second keypoint; the pixel with the highest feature value in the first dial keypoint image obtained through the second convolution is determined as the third keypoint; the pixel with the highest feature value in the first dial keypoint image obtained through the third convolution is determined as the fourth keypoint; and the pixel with the highest feature value in the second dial keypoint image obtained through the fourth convolution is determined as the first keypoint.

[0053] The first, second, third, and fourth convolution calculations are used to calculate the positions of four key points. For example, the first convolution calculation is used to determine the second key point. The first convolution calculation can be performed on the first dial feature image to obtain a first dial key point image. The pixel with the largest response value is then identified in the first dial key point image, and the position of this pixel is the position of the second key point.

[0054] Optionally, the keypoint recognition model can be trained as follows: Multiple keypoint location data subsets are determined from a pre-set keypoint location dataset. Each keypoint location data subset includes a training image and the locations of four keypoints in the training image. For each keypoint location data subset, one training image from that subset is input into the region recognition model to obtain a first dial feature map and a second dial feature map. The first and second dial feature maps are input into the keypoint recognition model to obtain the predicted locations of four keypoints. For each region location data subset, a second overlap degree is calculated between the predicted locations of the four keypoints and the locations of the four keypoints. Based on the second overlap degree, the loss function value of the keypoint recognition model is calculated. Based on the obtained loss function value, the initial keypoint recognition model is trained until the loss function value of the initial keypoint recognition model converges to a minimum value. The initial keypoint recognition model whose loss function value converges to a minimum value is determined as the target keypoint recognition model after training.

[0055] S105. Determine the reading of the pointer instrument based on the four key points identified.

[0056] The step of determining the meter value of the pointer instrument based on the four key points includes: determining the ratio of the first key point between the third key point and the fourth key point based on the first key point, the second key point, the third key point, and the fourth key point; and determining the meter value of the pointer instrument based on the ratio and the dial range of the pointer instrument.

[0057] Specifically, the step of determining the proportion of the first key point between the third and fourth key points based on the first, second, third, and fourth key points includes: determining the pointer tip rotation angle value based on the third, fourth, and first key points; determining the range angle value based on the second, third, and fourth key points; calculating the angle ratio between the pointer tip rotation angle value and the range angle value; and determining the proportion of the first key point between the third and fourth key points based on the angle ratio.

[0058] The meter reading identification method for pointer-type instruments provided in this application can determine the meter reading of pointer-type instruments by recognizing and processing the dial image. This solves the problem in the prior art that the meter reading cannot be correctly identified under the presence of multiple interferences, and achieves the effect of accurately determining the meter reading even under strong interference.

[0059] Based on the same inventive concept, this application also provides a pointer meter value identification device corresponding to the pointer meter value identification method. Since the principle of the device in this application is similar to the pointer meter value identification method described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0060] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of a pointer-type instrument value recognition device provided in an embodiment of this application. Figure 4 As shown, the meter reading recognition device 400 of the pointer-type instrument includes:

[0061] The dial image acquisition module 401 is used to acquire the dial image of an analog instrument.

[0062] The region recognition model calculation module 402 is used to input the dial image into the region recognition model to obtain the position of the dial scale strip region in the dial image.

[0063] The feature image extraction module 403 is used to extract multi-scale feature maps from the region recognition model, and based on the determined dial scale strip region and the target feature map of the largest scale in the multi-scale feature map, extract a first dial feature image and a second dial feature image from the target feature map. The first dial feature image includes other regions besides the dial scale strip region, and the second dial feature image includes the dial scale strip region.

[0064] The key point recognition model calculation module 404 is used to input the first dial feature map and the second dial feature map into the key point recognition model to determine four key points.

[0065] The meter value determination module 405 is used to determine the meter value of the pointer-type instrument based on the four determined key points.

[0066] The pointer-type instrument value recognition device provided in this application embodiment can determine the value of the pointer-type instrument by recognizing and processing the dial image. This solves the problem in the prior art that the dial reading of the instrument cannot be correctly identified under the presence of multiple interferences, and achieves the effect of accurately determining the dial reading of the instrument even under the presence of strong interference.

[0067] Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 5 As shown, the electronic device 500 includes a processor 510, a memory 520, and a bus 530.

[0068] The memory 520 stores machine-readable instructions executable by the processor 510. When the electronic device 500 is running, the processor 510 and the memory 520 communicate via the bus 530. When the machine-readable instructions are executed by the processor 510, they can perform the operations described above. Figure 1 The steps of the pointer-type instrument value identification method in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0069] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The steps of the pointer-type instrument value recognition method in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0070] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0071] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0072] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0073] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0074] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0075] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for recognizing the value of a pointer-type instrument, characterized in that, The method includes: Acquire an image of the dial of an analog instrument; The dial image is input into the region recognition model to determine the position of the dial scale strip region in the dial image; After receiving the dial image, the region recognition model obtains multiple feature maps at different scales. Based on the determined dial scale strip region and the target feature map at the largest scale in the multi-scale feature maps, it extracts a first dial feature image and a second dial feature image from the target feature map. The first dial feature image includes regions other than the dial scale strip region, and the second dial feature image includes the dial scale strip region. The first dial feature map and the second dial feature map are input into the key point recognition model to identify four key points; The readings of the pointer-type instrument are determined based on the four key points identified.

2. The method according to claim 1, characterized in that, The region recognition model is trained using the following method: Multiple subsets of regional location data are determined from a pre-set regional location dataset. Each subset of regional location data includes a training image and the location of the dial scale strip region in the training image. For each subset of regional location data, a training image from that subset is input into the region recognition model to obtain the predicted position of the dial scale strips; For each subset of regional location data, calculate the first degree of overlap between the predicted position of the dial scale strip and the position of the dial scale strip region; Based on the first degree of overlap, calculate the loss function value of the region recognition model; Based on the obtained loss function value, the initial region recognition model is trained until the loss function value of the initial region recognition model converges to the minimum value; The initial region recognition model that converges to the minimum value of the loss function is determined as the trained target region recognition model.

3. The method according to claim 1, characterized in that, The key point recognition model is trained using the following method: Multiple keypoint location data subsets are determined from a pre-set keypoint location dataset. Each keypoint location data subset includes a training image and the locations of four keypoints in the training image. For each subset of key point location data, a training image from that subset is input into the region recognition model to obtain a first dial feature map and a second dial feature map. The first dial feature map and the second dial feature map are input into the key point recognition model to obtain the predicted positions of four key points; For each subset of regional location data, calculate the second degree of overlap between the predicted locations of the four key points and the locations of the four key points; Based on the second degree of overlap, calculate the loss function value of the key point recognition model; Based on the obtained loss function value, the initial key point recognition model is trained until the loss function value of the initial key point recognition model converges to the minimum value; The initial keypoint recognition model that converges to the minimum value of the loss function is determined as the trained target keypoint recognition model.

4. The method according to claim 1, characterized in that, The first dial feature image is extracted from the target feature map using the following method: Based on the dial scale strip area, the target area of ​​the dial scale strip area is determined in the multi-scale feature map; The target region is mapped onto the target feature map of the largest scale to obtain a first overlay image; The target region in the first overlay image is set to zero to obtain the first dial feature image.

5. The method according to claim 1, characterized in that, The second dial feature image is extracted from the target feature map using the following method: Based on the dial scale strip area, the target area of ​​the dial scale strip area is determined in the multi-scale feature map; The target region is mapped onto the target feature map of the largest scale to obtain a second overlay image; Set all regions in the second overlay image except the target region to zero to obtain the second dial feature image.

6. The method according to claim 1, characterized in that, The four key points include the first key point, the second key point, the third key point, and the fourth key point. The first key point indicates the position of the pointer tip in the dial image, the second key point indicates the position of the pointer center in the dial image, the third key point indicates the position of the scale head in the dial image, and the fourth key point indicates the position of the scale tail in the dial image.

7. The method according to claim 6, characterized in that, The key point recognition model determines four key points from the first dial feature map and the second dial feature map in the following way: The first convolution calculation, the second convolution calculation, and the third convolution calculation are performed on the first dial feature image respectively to obtain three first dial key point images. The first dial key point image obtained by the first convolution calculation is used to determine the second key point, the first dial key point image obtained by the second convolution calculation is used to determine the third key point, and the first dial key point image obtained by the third convolution calculation is used to determine the fourth key point. Perform a fourth convolution calculation on the second dial feature image to obtain the second dial key point image, wherein the second dial key point image obtained by the fourth convolution calculation is used to determine the first key point; The pixel with the highest feature value in the first dial keypoint image obtained by the first convolution is determined as the second keypoint; the pixel with the highest feature value in the first dial keypoint image obtained by the second convolution is determined as the third keypoint; the pixel with the highest feature value in the first dial keypoint image obtained by the third convolution is determined as the fourth keypoint; and the pixel with the highest feature value in the second dial keypoint image obtained by the fourth convolution is determined as the first keypoint.

8. The method according to claim 6, characterized in that, The steps for determining the meter value of the pointer-type instrument based on the four key points identified include: Based on the first key point, the second key point, the third key point, and the fourth key point, determine the ratio of the first key point between the third key point and the fourth key point; The reading of the pointer instrument is determined based on the ratio and the dial range of the pointer instrument.

9. The method according to claim 8, characterized in that, The step of determining the ratio of the first key point to the third key point based on the first key point, the second key point, the third key point, and the fourth key point includes: Based on the third key point, the fourth key point, and the first key point, determine the pointer tip rotation angle value; Determine the range angle value based on the second, third, and fourth key points; Calculate the angle ratio between the pointer tip rotation angle value and the range angle value; The angle ratio is used to determine the proportion of the first key point between the third key point and the fourth key point.

10. A device for recognizing the value of a pointer-type instrument, characterized in that, The device includes: The dial image acquisition module is used to acquire dial images of analog instruments; The region recognition model calculation module is used to input the dial image into the region recognition model to obtain the position of the dial scale strip region in the dial image; The feature image extraction module is used to obtain multiple feature maps of different scales after the region recognition model receives the dial image, and extract a first dial feature image and a second dial feature image from the target feature map based on the determined dial scale strip area and the target feature map of the largest scale in the multi-scale feature map. The first dial feature image includes other areas except the dial scale strip area, and the second dial feature image includes the dial scale strip area. The key point recognition model calculation module is used to input the first dial feature map and the second dial feature map into the key point recognition model to determine four key points; The meter value determination module is used to determine the meter value of the pointer-type instrument based on four determined key points.

Images