Equipment point location mapping method, electronic equipment and storage medium

Abstract

The invention discloses an equipment point location mapping method, electronic equipment and a storage medium. The equipment point location mapping method comprises the following steps: acquiring an engineering image; target recognition is carried out on the to-be-monitored equipment in the engineering image through a preset image recognition model, an image recognition result is generated, the type and position information of the to-be-monitored equipment is marked in the image recognition result, and the preset image recognition model is obtained through training of a deep learning model; performing coordinate conversion on the position information of the to-be-monitored equipment in the image recognition result to generate a coordinate conversion result; and binding the equipment to be monitored on the digital twinborn map based on a coordinate conversion result. According to the equipment point location mapping method, the problem that a large amount of manpower and time cost need to be consumed in the configuration process of mapping the real world equipment by the digital twin map can be solved.

Description

Technical Field

[0001] This application relates to the field of digital twin technology, and particularly to a method for mapping device points on a map, an electronic device, and a storage medium. Background Art

[0002] A digital twin map is a virtual map constructed using digital twin technology (a technology that uses digital means to comprehensively and accurately simulate and map physical entities), corresponding to the real geographical space. It can perform high-precision digital modeling on various elements such as terrain, landforms, buildings, and roads in the real geographical space, and can access various dynamic data in real time, enabling users to timely understand the changes in the geographical space. When applied to the field of building automation, managers can centrally monitor and manage each system of a building on a unified platform, without the need to switch between different systems and devices, greatly improving management efficiency, and can reduce losses caused by equipment failures, reduce energy consumption, thereby reducing the operating costs of the building. In addition, digital twin technology can provide a more comfortable and convenient working and living environment for users in the building, enhancing the user experience.

[0003] However, to achieve the mapping of the digital twin map to the real world, engineers need to bind each device to be monitored to the digital twin map one by one by referring to the CAD engineering drawings. For relatively large projects, this process requires a large amount of human and time costs. Summary of the Invention

[0004] Embodiments of this application provide a method for mapping device points on a map, a device point mapping device, an electronic device, and a non-volatile computer-readable storage medium.

[0005] The method for mapping device points on a map provided by the embodiments of this application includes:

[0006] Obtain an engineering image;

[0007] Perform target recognition on the devices to be monitored in the engineering image through a preset image recognition model to generate an image recognition result, where the type and location information of the devices to be monitored are marked in the image recognition result, and the preset image recognition model is obtained by training a deep learning model;

[0008] Perform coordinate conversion on the location information of the devices to be monitored in the image recognition result to generate a coordinate conversion result; and

[0009] Based on the coordinate conversion result, bind the devices to be monitored on the digital twin map.

[0010] In some embodiments, coordinate conversion is performed on the position information of the device to be monitored in the image recognition result to generate a coordinate conversion result, including:

[0011] Mark a reference point in the image recognition result;

[0012] Obtain a target reference point at the position corresponding to the reference point in the digital twin map;

[0013] Calculate a standard parameter set for converting the reference point to the target reference point based on the coordinates of the reference point and the coordinates of the target reference point;

[0014] Perform coordinate conversion on the position information of the device to be monitored according to the standard parameter set to generate a coordinate conversion result.

[0015] In some embodiments, the coordinate conversion of the position information of the device to be monitored in the image recognition result to generate a coordinate conversion result further includes:

[0016] Adjust the offset parameters of the X and / or Y axes in the coordinate system of the digital twin map to adjust the aggregation degree of the device to be monitored in the digital twin map.

[0017] In some embodiments, based on the coordinate conversion result, binding the device to be monitored on the digital twin map includes:

[0018] Determine the target output format of the coordinate conversion result;

[0019] Generate a table file in the target output format with the coordinate conversion result, where the table file defines the device type identifier, coordinate input format, keys and key values of each column in the table file;

[0020] Import the table file into the digital twin map to complete the binding of the device to be monitored on the digital twin map.

[0021] In some embodiments, target recognition is performed on the device to be monitored in the engineering image through a preset image recognition model to generate an image recognition result, including:

[0022] Determine the floor type of the engineering image;

[0023] Detect the bounding boxes of each target object in the engineering image;

[0024] Outline the contours of the target objects based on the bounding boxes to generate the image recognition result.

[0025] In some embodiments, the method for mapping device points further includes:

[0026] Obtain a training sample set, where the training sample set includes a training set and a validation set;

[0027] Perform labeling processing on the sample set;

[0028] Use the labeled training set to train the deep learning model to obtain a training result;

[0029] Based on the training result and the validation set, correct the deep learning model;

[0030] When the test result of the deep learning model meets the accuracy requirement, use the corrected deep learning model as the preset image recognition model.

[0031] In some embodiments, the method for mapping device points on the map further includes:

[0032] Record the training data during the training process of the deep learning model;

[0033] Output the training data in the form of charts and tables.

[0034] In some embodiments, using the labeled training set to train the deep learning model to obtain a training result includes:

[0035] Perform image enhancement processing on the training sample set, where the image enhancement processing includes at least one of mosaic enhancement, hybrid enhancement, random perturbation, or color perturbation..

[0036] In some embodiments, the deep learning model includes one of YOLOv8, YOLOv5, and large artificial intelligence models.

[0037] The device point mapping device according to the embodiments of the present application includes:

[0038] An acquisition module, configured to acquire engineering images;

[0039] A prediction module, configured to perform target recognition on the devices to be monitored in the engineering images through a preset image recognition model, and generate an image recognition result, where the type and location information of the devices to be monitored are marked in the image recognition result;

[0040] A conversion module, configured to perform coordinate conversion on the location information of the devices in the image recognition result to generate a coordinate conversion result; and

[0041] An import module, configured to import the coordinate conversion result into the digital twin map to implement the binding of the devices to be monitored on the digital twin map.

[0042] An electronic device according to an embodiment of the present application includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the processor implements the method for mapping device points on the map.

[0043] A non - volatile computer - readable storage medium according to an embodiment of the present application includes a computer program. When the computer program is executed by a processor, the processor implements the method for mapping device points on the map.

[0044] In the method for mapping device points on the map, the device for mapping device points on the map, the electronic device, and the storage medium according to the embodiments of the present application, a preset image recognition model obtained by training a deep - learning model is used to perform target recognition on the device to be monitored in the engineering image, generate an image recognition result, then perform coordinate conversion on the position information of the device to be monitored in the image recognition result to generate a coordinate conversion result in the digital - twin map coordinate system, and finally, based on the coordinate conversion result, bind the device to be monitored on the digital - twin map. In this way, the problem that the configuration process of mapping the device to be monitored in the real world on the digital - twin map requires a large amount of manpower and time costs is solved, the configuration efficiency is improved, and the cost is greatly reduced. Description of the Drawings

[0045] The above - mentioned and / or additional aspects and advantages of the present application will become obvious and easy to understand from the following description of the embodiments in conjunction with the drawings, where:

[0046] Figure 1 is a schematic flow chart of the method for mapping device points on the map according to some embodiments of the present application;

[0047] Figure 2 is a schematic block diagram of the device for mapping device points on the map according to some embodiments of the present application;

[0048] Figure 3 is a schematic diagram of the GUI interface for selecting an engineering image according to some embodiments of the present application;

[0049] Figure 4 is a schematic diagram of the GUI interface for selecting a preset image recognition model according to some embodiments of the present application;

[0050] Figure 5 is a schematic diagram of the image recognition result according to some embodiments of the present application;

[0051] Figure 6 is a schematic flow chart of the method for mapping device points on the map according to some embodiments of the present application;

[0052] Figure 7 and Figure 8 is a schematic diagram of the GUI interface for marking reference points according to some embodiments of the present application;

[0053] Figure 9 It is a schematic flowchart of the method for mapping device points in some embodiments of the present application;

[0054] Figure 10 It is a schematic diagram of adjusting the offset parameters of the X and / or Y axes in some embodiments of the present application;

[0055] Figure 11 It is a schematic flowchart of the method for mapping device points in some embodiments of the present application;

[0056] Figures 12 - 14 It is a GUI schematic diagram of customizing the output format of coordinate conversion results in some embodiments of the present application;

[0057] Figure 15 It is a schematic diagram of a tabular file in some embodiments of the present application;

[0058] Figure 16 It is a schematic flowchart of the method for mapping device points in some embodiments of the present application;

[0059] Figure 17 It is a schematic diagram of the modules of the device point mapping device in some embodiments of the present application;

[0060] Figure 18 It is a schematic flowchart of the method for mapping device points in some embodiments of the present application;

[0061] Figure 19 It is a schematic diagram of adjusting the parameters of the online data enhancement algorithm in some embodiments of the present application;

[0062] Figure 20 It is a schematic flowchart of the method for mapping device points in some embodiments of the present application;

[0063] Figure 21 It is a schematic diagram of the training data in some embodiments of the present application. Specific Embodiments

[0064] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary only for explaining the embodiments of the present invention and should not be construed as limiting the embodiments of the present invention.

[0065] Please refer to Figure 1 , the embodiments of the present application provide a method for mapping device points, and the method for mapping device points includes:

[0066] 01: Obtain engineering images;

[0067] 02: Perform object recognition on the device to be monitored in the engineering image through a preset image recognition model to generate an image recognition result, where the position information of the device to be monitored is marked in the image recognition result, and the preset image recognition model is obtained by training a deep learning model;

[0068] 03: Perform coordinate transformation on the position information of the device to be monitored in the image recognition result to generate a coordinate transformation result; and

[0069] 04: Bind the device to be monitored on the digital twin map based on the coordinate transformation result.

[0070] Please refer to Figure 2 This embodiment also provides a device location mapping device 10, which includes an acquisition module 110, a prediction module 120, a conversion module 130, and an import module 140.

[0071] Step 01 can be implemented by the acquisition module 110, step 02 can be implemented by the prediction module 120, step 03 can be implemented by the conversion module 130, and step 04 can be implemented by the import module 140. That is to say, the acquisition module 110 is used to obtain the engineering image of the building; the prediction module 120 can be used to perform object recognition on the device to be monitored in the engineering image through a preset image recognition model to generate an image recognition result, where the position information of the device to be monitored is marked in the image recognition result; the conversion module 130 is used to perform coordinate transformation on the position information of the device in the image recognition result to generate a coordinate transformation result; the import module 140 is used to bind the device to be monitored on the digital twin map based on the coordinate transformation result.

[0072] This application embodiment also provides an electronic device, which includes a processor and a memory. Among them, the memory stores a computer program. When the computer program is executed by the processor, the processor is caused to execute the above device location mapping method. That is, the processor can be used to obtain the engineering image; perform object recognition on the device to be monitored in the engineering image through a preset image recognition model to generate an image recognition result, where the position information of the device to be monitored is marked in the image recognition result, and the preset image recognition model is obtained by training a deep learning model; perform coordinate transformation on the position information of the device to be monitored in the image recognition result to generate a coordinate transformation result; and bind the device to be monitored on the digital twin map based on the coordinate transformation result.

[0073] In the device location mapping method, device location mapping apparatus 10, and electronic device of the present application, a preset image recognition model obtained by training a deep learning model is used to perform target recognition on the devices to be monitored in the engineering image, generate an image recognition result, then perform coordinate conversion on the location information of the devices to be monitored in the image recognition result to generate a coordinate conversion result in the digital twin map coordinate system, and finally, based on the coordinate conversion result, bind the devices to be monitored on the digital twin map. In this way, the problem that the configuration process of mapping the devices to be monitored in the real world on the digital twin map requires a large amount of human and time costs is solved, the configuration efficiency is improved, and the cost is greatly reduced.

[0074] In some embodiments, the device location mapping apparatus 10 may be a part of the electronic device. Or rather, the electronic device includes the device location mapping apparatus 10.

[0075] In some embodiments, the device location mapping apparatus 10 may be discrete components assembled in a certain way to have the aforementioned functions, or a chip in the form of an integrated circuit having the aforementioned functions, or a computer software code segment that, when running on a computer, enables the computer to have the aforementioned functions.

[0076] In some embodiments, as hardware, the device location mapping apparatus 10 may be independent or added as an additional peripheral component to the electronic device. The device location mapping apparatus 10 may also be integrated into the electronic device. For example, when the device location mapping apparatus 10 is a part of the electronic device, the device location mapping apparatus 10 may be integrated into the processor.

[0077] The electronic device may be a computer (such as a personal computer PC, a laptop computer, etc.). The digital twin map can run in the electronic device. The digital twin map is a virtual map corresponding to the real building and is used to achieve point-to-point real-time control and operation and maintenance monitoring of certain systems in the building (such as the HVAC system, the power system, the lighting system).

[0078] Engineering images refer to CAD engineering drawings. Understandably, CAD engineering drawings are drawn using computer-aided design (CAD) software and are used to represent the design scheme of engineering projects. CAD engineering drawings can be, but are not limited to, architectural construction drawings, structural construction drawings, electrical equipment construction drawings, and heating, ventilation, and air conditioning (HVAC) equipment construction drawings. The area represented by the engineering images is the same as the area represented by the digital twin map. The engineering images can be one or multiple, and one or more devices to be monitored are shown in the engineering images. Understandably, when the digital twin map is used to control and perform operation and maintenance monitoring of the HVAC system, the devices to be monitored can be HVAC devices such as air-conditioning hosts, indoor units, ventilation equipment, water pumps, coil units, fans, and thermostats. When the digital twin map is used to control and perform operation and maintenance monitoring of the power system, the devices to be monitored can be electricity meters, distribution boxes, wires and cables, sockets, switches, electrical equipment, etc.

[0079] The preset image recognition model is an image recognition model pre-set in the device potential mapping device or electronic device. For example, the preset image recognition model can be stored in the memory of the electronic device and then called by the processor to implement the above step 02. The preset image recognition model is obtained by training a deep learning model. Understandably, a deep learning model is a type of machine learning model based on artificial neural networks. By constructing a neural network with multiple layers, it automatically learns the features and patterns in the data to achieve tasks such as data classification, prediction, and generation. The deep learning model can include, but is not limited to, one of YOLOv8, YOLOv5, and large artificial intelligence models. Both YOLOv8 and YOLOv5 are object detection algorithms in the YOLO series in the field of object detection. Among them, YOLOv8 has the advantages of high efficiency, high accuracy, and model lightweight. Therefore, in this embodiment, the deep learning model can be illustrated by taking YOLOv8 as an example. That is, the preset image recognition model can be obtained by training with YOLOv8.

[0080] The preset image recognition model can be one or multiple. When the preset image recognition model is one, the preset image recognition model can perform object recognition on all devices to be monitored in the engineering image and generate an image recognition result. When the preset image recognition model is multiple, each preset image recognition model can only perform object recognition on one type of device to be monitored in the engineering image. That is, each preset image recognition model is only used to recognize one type of device to be monitored. For example, the preset image recognition model can include two, and the devices to be monitored include outdoor air conditioners and indoor units. One preset image recognition model is used to recognize the outdoor air conditioners in the engineering image, and the other preset image recognition model is used to recognize the indoor units in the engineering image. Thus, by using the two preset image recognition models to perform object recognition on the devices to be monitored in the engineering image respectively, the final image recognition result can be obtained.

[0081] Furthermore, in the process of the preset image recognition model performing object recognition on all devices to be monitored in the engineering image, tasks such as classification, detection, and segmentation can be carried out, thereby generating an image recognition result. Specifically, the preset image recognition model can first perform classification recognition on the engineering image to determine the floor type of the engineering image, and then detect the bounding boxes of each target object (device to be monitored) in the engineering image; and based on the bounding boxes, outline the contours of the target objects to generate an image recognition result.

[0082] The image recognition result can be output in the form of a picture. The image recognition result marks the types and location information of the devices to be monitored. The types of the devices to be monitored in the image recognition result can be represented in the form of text, letters, or numbers, and the location information can be represented in the form of coordinates.

[0083] The coordinate conversion result is used to represent the position coordinates of the devices to be monitored in the digital twin map. Thus, based on the coordinate conversion result, the devices to be monitored can be bound on the digital twin map.

[0084] The steps of the above method for mapping device points can be implemented based on the interaction between the user and the device point mapping device or electronic device. Specifically, the electronic device includes a desktop graphical user interface (GUI) for each step in the above method for mapping device points. The user can interact with the electronic device by interacting with the desktop GUI, so that the processor of the electronic device implements the above method for mapping device points. Thus, the user only needs to complete each operation to bind the devices to be monitored on the digital twin map.

[0085] For example, in some examples, in step 01, an engineering image can be obtained based on the user's selection (as Figure 3 shown), and in step 02, the user can select a preset image recognition model to perform object recognition on the selected engineering image and generate an image recognition result (as Figure 4 and Figure 5 shown).

[0086] Please refer to Figure 6 , in some embodiments, step 03 includes:

[0087] 031: Mark a reference point in the image recognition result;

[0088] 032: Obtain a target reference point at the position corresponding to the reference point in the digital twin map;

[0089] 033: Calculate a standard parameter set for converting the reference point to the target reference point according to the coordinates of the reference point and the coordinates of the target reference point;

[0090] 034: Perform coordinate transformation on the location information of the device to be monitored according to the standard parameter set, and generate a coordinate transformation result.

[0091] Please further combine Figure 2 , in some embodiments, 031 - 034 can be implemented by the conversion module 130. Or rather, the conversion module 130 can be used to mark reference points in the image recognition result; obtain target reference points corresponding to the reference points in the digital twin map; calculate the standard parameter set for converting the reference points to the target reference points according to the coordinates of the reference points and the target reference points; perform coordinate transformation on the location information of the device to be monitored according to the standard parameter set, and generate a coordinate transformation result.

[0092] In some embodiments, the processor can be used to mark reference points in the image recognition result; and obtain target reference points corresponding to the reference points in the digital twin map; the processor can also be used to calculate the standard parameter set for converting the reference points to the target reference points according to the coordinates of the reference points and the target reference points, and perform coordinate transformation on the location information of the device to be monitored according to the standard parameter set, and generate a coordinate transformation result.

[0093] Specifically, after generating the image recognition result, the electronic device can generate a table text corresponding to the image recognition result (please refer to Figure 7 ), and the table text includes information such as the type and location coordinates of each identified device to be monitored. The user can select two reference points for coordinate transformation as the origin in the table text (as shown in Figure 8 ), then obtain the coordinates corresponding to the origin on the corresponding digital twin map, and calculate the standard parameter set required to convert the absolute coordinates in the image recognition result to the target twin map coordinate system. After obtaining the standard parameter set, the coordinates of all the identified devices to be monitored in the image recognition result can be converted into the coordinates in the coordinate system of the digital twin map through the standard parameter set, so as to generate a coordinate transformation result.

[0094] Please refer to Figure 9 , in some embodiments, before sub-step 034, step 03 further includes:

[0095] 035, adjust the offset parameters of the X and / or Y axes in the coordinate system of the digital twin map to adjust the aggregation degree of the devices to be monitored in the digital twin map.

[0096] In some embodiments, sub-step 035 can be implemented by the conversion module 130. Or rather, the conversion module 130 is also used to adjust the offset parameters of the X and / or Y axes in the coordinate system of the digital twin map to adjust the aggregation degree of the devices to be monitored in the digital twin map.

[0097] In some embodiments, the processor is further configured to adjust the offset parameters of the X and / or Y axes in the coordinate system of the digital twin map to adjust the degree of aggregation of the devices to be monitored in the digital twin map.

[0098] The offset parameters of the X and Y axes affect the degree of aggregation of each point in the final generated coordinate conversion result in the digital twin map. The smaller the offset value, the more aggregated, and the larger the offset value, the more dispersed. The aggregation direction is controlled separately by the X and Y axes. Therefore, by adjusting the offset parameters of the X and / or Y axes, the degree of aggregation of each device to be monitored on the X / Y axis in the final coordinate conversion result is adjusted respectively (as Figure 10 shown), so as to ensure the accuracy of the position coordinates of the devices to be monitored in the digital twin map in the coordinate conversion result.

[0099] In this way, by adjusting the offset parameters of the X and / or Y axes in the coordinate system of the digital twin map, the accuracy of the devices to be monitored in the digital twin map is ensured.

[0100] Please refer to Figure 11 , in some embodiments, 04 includes:

[0101] 041: Determine the target output format of the coordinate conversion result;

[0102] 042: Generate a table file with the coordinate conversion result in the target output format. The table file defines the device type identifier, coordinate input format, keys and key values of each column in the table file of the coordinate conversion result;

[0103] 043: Import the table file into the digital twin map to complete the binding of the devices to be monitored on the digital twin map.

[0104] Please further refer to Figure 2 , in some embodiments, 041 - 044 can be implemented by the import module 140. Or rather, the import module 140 can be used to determine the target output format of the coordinate conversion result, generate a table file with the coordinate conversion result in the target output format. The table file defines the device type identifier, coordinate input format, keys and key values of each column in the table file of the coordinate conversion result; import the table file into the digital twin map to complete the binding of the devices to be monitored on the digital twin map.

[0105] In some embodiments, the processor can be used to determine the target output format of the coordinate conversion result, generate a table file with the coordinate conversion result in the target output format. The table file defines the device type identifier, coordinate input format, keys and key values of each column in the table file of the coordinate conversion result; import the table file into the digital twin map to complete the binding of the devices to be monitored on the digital twin map.

[0106] Specifically, the electronic device may include a desktop GUI for defining the output format of the coordinate conversion result. The user can customize the output format of the coordinate conversion result through the desktop GUI (such as Figures 12 - 14 shown). Among them, the customized output format defines the device type identifier, the X and Y coordinate output templates, the keys (Key) of each column of the output table, and the key values (value) filled in each Key, etc.

[0107] In sub-step 041, the processor can determine whether the user has defined the output format of the coordinate conversion result according to the input of the desktop GUI. When the user defines the output format of the coordinate conversion result through the desktop GUI, the user-defined format is used as the target output format of the coordinate conversion result. When the user does not define the output format of the coordinate conversion result through the desktop GUI, the default format is used as the target output format of the coordinate conversion result.

[0108] In sub-step 042. The processor can generate and save an Excel table file for the coordinate conversion result according to the target output format (such as Figure 15 shown). In sub-step 043, the output table file can be imported into the digital twin map through a browser, or the table file can be directly imported into the database of the digital twin map, so as to bind the device to be monitored to the digital twin map.

[0109] Please refer to Figure 16 , in some embodiments, the method for mapping device points to the map further includes:

[0110] 001: Obtain a training sample set, where the training sample set includes a training set and a validation set;

[0111] 002: Perform labeling processing on the sample set;

[0112] 003: Train the deep learning model with the labeled training set to obtain a training result;

[0113] 004: Correct the deep learning model based on the training result and the validation set;

[0114] 005: When the test result of the deep learning model meets the accuracy requirement, use the corrected deep learning model as the preset image recognition model.

[0115] Please refer to Figure 17In some embodiments, the device 10 for mapping equipment points may further include a training module 150, wherein step 001 may be implemented by the acquisition module 110, and steps 002-005 may be implemented by the training module 150. In other words, the acquisition module 110 may also be used to acquire a training sample set, which includes a training set and a validation set, and the training module 150 may be used to label the sample set; train the deep learning model with the labeled training set to obtain a training result; modify the deep learning model based on the training result and the validation set; and use the modified deep learning model as a preset image recognition model when the test result of the deep learning model meets the accuracy requirement.

[0116] In some embodiments, the processor can be used to obtain a training sample set, which includes a training set and a validation set. The processor can also be used to label the sample set, train the deep learning model with the labeled training set to obtain training results, and modify the deep learning model based on the training results and the validation set; and when the test results of the deep learning model meet the accuracy requirements, the modified deep learning model is used as the preset image recognition model.

[0117] In some examples, 500 to 1000 engineering image samples are prepared for each device to be monitored, and the engineering image samples are divided into a training set and a validation set, wherein the ratio of training objects in the training set to validation objects in the validation set is recommended to be 7:3 or 8:2. For example, for the image recognition model of three types of monitored devices, namely coil units, fans and thermostats, CAD drawings related to these three types of monitored devices are required. At least 500 samples are found for each object in these CAD drawings, and 70% of each 500 samples are extracted as a test set, and 30% are extracted as a validation set. The training set and validation set paths can be selected through the browser provided by the GUI.

[0118] The electronic device also includes a labeling tool, which can efficiently label the target graphic objects (devices to be monitored) on the CAD drawing and save the labeled information in a format readable by the deep learning model. After the labeling tool completes the labeling work, the output label set can be put together with the corresponding training set and verification set. The label set path can be selected through the browser provided by the GUI.

[0119] After the training set, validation set, and label set are prepared, the deep learning model training can be started directly, and the labeled training set is used to train the deep learning model to obtain the training results. In order to improve the efficiency of training, the electronic device can be an electronic device with a GPU that supports CUDA12.

[0120] It should be noted that the deep learning model can include, but is not limited to, training tasks such as classification, detection, and segmentation. Among them, the classification task refers to learning the global parsing of the entire image and outputting a model that can distinguish this type of image. In this embodiment, the classification training task is used to train a floor classification model. The detection task refers to locating the target bounding boxes in the dataset images and classifying their categories. The detection task will output a model that can identify the bounding boxes, types, and confidence levels of each target object in the image. The segmentation task refers to further distinguishing the boundaries between the target objects and the background on the basis of the detection task and outlining the approximate contours of the objects.

[0121] In addition, it should also be noted that the style standards of various device types on the CAD engineering drawings can be predefined, and strict drawing according to these standards can be carried out during the project implementation process. In this way, the number of times of customizing the training model can be reduced, and a set of models can cover the working conditions of most devices on the drawings. In this way, the human resources and time costs can be greatly reduced, the work efficiency can be significantly improved, the working hours of this project can be shortened, and even the model can be reused in future other projects.

[0122] Please refer to Figure 18 , in some embodiments, before step 002 or 003, the method for placing device points on the drawing further includes:

[0123] 006: Perform image enhancement processing on the training sample set. The image enhancement processing includes at least one of mosaic enhancement, hybrid enhancement, random perturbation, or color perturbation;

[0124] Please further refer to Figure 17 , in some embodiments, step 006 can be implemented by the training module 150, or rather, the training module 150 can be used to perform image enhancement processing on the training sample set. The image enhancement processing includes at least one of mosaic enhancement, hybrid enhancement, random perturbation, or color perturbation.

[0125] In some embodiments, the processor can be used to perform image enhancement processing on the training sample set. The image enhancement processing includes at least one of mosaic enhancement, hybrid enhancement, random perturbation, or color perturbation.

[0126] It should be noted that the processor can integrate multiple online data enhancement algorithms. The online data enhancement algorithms include at least one of mosaic enhancement, hybrid enhancement, random perturbation, or color perturbation. The online data enhancement algorithms can adjust the brightness, contrast, color saturation, etc. of the images in the training sample set, or perform operations such as adding noise, to improve the diversity and generalization ability of the dataset, so that the deep learning model can better understand the essential features in the training sample set.

[0127] Please refer to Figure 19, Further, the electronic device may include a desktop GUI for implementing online data enhancement. The user can adjust the parameters of the online data enhancement algorithm through the desktop GUI, so that the effect of the training sample set can better meet the user's needs.

[0128] In this way, by performing image enhancement processing on the training sample set, the overfitting risk of the deep learning model can be reduced, the robustness of the deep learning model can be enhanced, and more abundant information can be provided for model training, enabling the model to learn more subtle features and patterns.

[0129] Please refer to Figure 20 and Figure 21 , In some embodiments, the device point-on-map method further includes:

[0130] 007: Record the training data during the training process of the deep learning model;

[0131] 008: Output the training data in the form of charts and tables.

[0132] In some embodiments, step 007 and step 008 can be implemented by the training module 150. Or rather, the training module 150 can also be used to record the training data during the training process of the deep learning model and output the training data in the form of charts and tables.

[0133] In some embodiments, the processor can also be used to record the training data during the training process of the deep learning model and output the training data in the form of charts and tables.

[0134] The training data can be output together with the preset image recognition model after the deep learning model is trained to obtain the preset image recognition model. The training data may include but is not limited to the training results, accuracy, number of training times, etc. of the deep learning model.

[0135] In this way, by outputting the training data related to training in the form of charts and tables, the user can master the training situation of the preset image recognition model.

[0136] The embodiments of the present application also provide a non-transitory computer-readable storage medium, including a computer program, which when executed by a processor, enables the processor to implement the device point-on-map method described above.

[0137] In the storage medium of the embodiment of the present application, a preset image recognition model obtained by training a deep learning model is used to perform target recognition on the device to be monitored in the engineering image, generate an image recognition result, then perform coordinate conversion on the position information of the device to be monitored in the image recognition result to generate a coordinate conversion result in the digital twin map coordinate system, and finally import the coordinate conversion result into the digital twin map, so that the device to be monitored can be bound on the digital twin map. In this way, the problem that the configuration process of mapping the device to be monitored in the real world in the digital twin map requires a large amount of manpower and time costs is solved, the configuration efficiency is improved and the cost is greatly reduced.

[0138] In the description of this specification, the descriptions referring to terms such as "one embodiment", "certain embodiments", "schematic embodiments", "examples", "specific examples", or "some examples" etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0139] Although the embodiments of the present application have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present application, and the scope of the present application is defined by the claims and their equivalents.

[0140] The above is only the specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for plotting device points, characterized in that, The method for mapping the device points onto the map includes: Obtaining an engineering image; Performing target recognition on the devices to be monitored in the engineering image through a preset image recognition model to generate an image recognition result, where the position information of the devices to be monitored is marked in the image recognition result, and the preset image recognition model is obtained by training a deep learning model; Performing coordinate transformation on the position information of the devices to be monitored in the image recognition result to generate a coordinate transformation result; and Binding the devices to be monitored on the digital twin map based on the coordinate transformation result.

2. The method for positioning the device points as described in claim 1, wherein, Performing coordinate transformation on the position information of the devices to be monitored in the image recognition result to generate a coordinate transformation result, including: Marking reference points in the image recognition result; Obtaining target reference points at the corresponding positions in the digital twin map for the reference points; Calculating a standard parameter set for converting the reference points to the target reference points according to the coordinates of the reference points and the coordinates of the target reference points; Performing coordinate transformation on the position information of the devices to be monitored according to the standard parameter set to generate a coordinate transformation result.

3. The method for positioning the device points according to claim 2, wherein Performing coordinate transformation on the position information of the devices to be monitored in the image recognition result to generate a coordinate transformation result further includes: Adjusting the offset parameters of the X and / or Y axes in the coordinate system of the digital twin map to adjust the aggregation degree of the devices to be monitored in the digital twin map.

4. The method for mapping device points according to claim 2, wherein Binding the devices to be monitored on the digital twin map based on the coordinate transformation result includes: Determining the target output format of the coordinate transformation result; Generating a table file in the target output format for the coordinate transformation result, where the table file defines the device type identifier, coordinate input format, keys and key values of each column in the table file for the coordinate transformation result; Importing the table file into the digital twin map to complete the binding of the devices to be monitored on the digital twin map.

5. The method for positioning the device points as described in claim 1, wherein, Performing target recognition on the devices to be monitored in the engineering image through a preset image recognition model to generate an image recognition result, including: Determining the floor type of the engineering image; Detecting the bounding boxes of each target object in the engineering image; Outlining the contours of the target objects based on the bounding boxes to generate the image recognition result.

6. The method for positioning the device points on the map according to claim 1, characterized in that The method for mapping the device points onto the map further includes: Obtaining a training sample set, where the training sample set includes a training set and a validation set; Performing labeling processing on the sample set; Training the deep learning model with the labeled training set to obtain a training result; Correcting the deep learning model based on the training result and the validation set; When the test result of the deep learning model meets the accuracy requirement, using the corrected deep learning model as the preset image recognition model.

7. The method for mapping device points according to claim 6, wherein The method for mapping the device points onto the map further includes: Performing image enhancement processing on the training sample set, where the image enhancement processing includes at least one of mosaic enhancement, hybrid enhancement, random perturbation, or color perturbation.

8. The method for plotting device points according to claim 6, wherein The method for mapping the device points onto the map further includes: Recording the training data during the training process of the deep learning model; Output the training data in the form of charts and tables.

9. An electronic device, characterized in that, It includes a processor and a memory, and the memory stores a computer program. When the computer program is executed by the processor, the processor implements the device point mapping method according to any one of claims 1-8.

10. A non-volatile computer-readable storage medium, characterized in that, It includes a computer program. When the computer program is executed by a processor, the processor implements the device point mapping method according to any one of claims 1-8.

Images