Device point labeling method and apparatus, and device and storage medium

Abstract

Provided in the present disclosure are a device point labeling method and apparatus, and a device and a storage medium. The method comprises: on the basis of metadata corresponding to a three-dimensional scene model, determining a device detection point set corresponding to the three-dimensional scene model, wherein the device detection point set comprises a plurality of device detection points; and matching the device detection points in the device detection point set with device reference points in a preset data source point set, and labeling device points in the three-dimensional scene model on the basis of a matching result, wherein the data source point set is a set of device points in a real scene corresponding to the three-dimensional scene model, and the data source point set comprises a plurality of device reference points.

Description

Equipment point labeling methods, devices, equipment and storage media

[0001] Cross-references to related applications

[0002] This disclosure claims priority to Chinese Patent Application No. 2024105077450, filed on April 25, 2024, entitled "Method, Apparatus, Device and Storage Medium for Device Dotting", which is incorporated herein by reference in its entirety. Technical Field

[0003] This disclosure relates to the field of computer technology, and in particular to a device dot annotation method, apparatus, device, and storage medium. Background Technology

[0004] With the continuous development of the security industry, there is an increasing demand for building 3D scene models and annotating the devices to be displayed in these models.

[0005] When marking the devices to be displayed in a 3D scene model, the process usually involves manually finding the device locations (referring to the positions of the devices) on the map corresponding to the 3D scene model and then manually marking the corresponding devices at those locations.

[0006] However, the above technologies are time-consuming and labor-intensive. Summary of the Invention

[0007] This disclosure provides a method, apparatus, device, and storage medium for labeling device points, which addresses the shortcomings of existing technologies in labeling device points in 3D scene models, resulting in time-consuming and labor-intensive processes and achieving the technical effect of saving manpower and time in labeling.

[0008] This disclosure provides a method for marking equipment points, including:

[0009] Based on the metadata corresponding to the 3D scene model, determine the set of device detection points corresponding to the 3D scene model; the set of device detection points includes multiple device detection points.

[0010] The device detection points in the device detection point set are matched with the device reference points in the preset data source point set, and the device points in the 3D scene model are labeled according to the matching results.

[0011] The aforementioned data source set is a collection of device points in the real scene corresponding to the 3D scene model, including multiple device reference points.

[0012] According to the equipment point annotation method provided in this disclosure, the above-mentioned matching of equipment detection points in the equipment detection point set with equipment reference points in the preset data source point set includes:

[0013] Iterate through each device reference point in the data source point set, calculate the first distance between each device reference point and each device detection point, and obtain the first distance set corresponding to each device reference point;

[0014] Based on the first distance set of each device reference point, determine the candidate detection points that match each device reference point from each device detection point;

[0015] Based on the candidate detection points corresponding to each device reference point, a corresponding set of candidate detection points is determined; the set of candidate detection points includes at least one candidate detection point.

[0016] According to the device point annotation method provided in this disclosure, if the candidate detection point set corresponding to the first device reference point includes a first candidate detection point, then the above-mentioned annotation of device points in the 3D scene model based on the matching result includes:

[0017] Establish a binding relationship between the first device reference point and the device detection point corresponding to the first candidate detection point, and mark the information corresponding to the first device reference point on the device detection point corresponding to the first candidate detection point in the 3D scene model.

[0018] According to the device point annotation method provided in this disclosure, if the candidate detection point set corresponding to the second device reference point includes multiple second candidate detection points, and the second device reference point is a device reference point without established binding relationship, then the annotation of device points in the 3D scene model based on the matching result includes:

[0019] Determine at least two target third device reference points from the data source point set; each target third device reference point is a device reference point that has a binding relationship with a device detection point;

[0020] Construct each second closed shape based on each second candidate detection point and the device detection point corresponding to each target third device reference point;

[0021] Based on the similarity between each second closed shape and the target first closed shape corresponding to at least two target third device reference points, a target detection point matching the second device reference point is determined from each second candidate detection point;

[0022] Establish a binding relationship between the second device reference point and the device detection point corresponding to the target detection point, and mark the information corresponding to the second device reference point on the device detection point corresponding to the target detection point in the 3D scene model.

[0023] According to the device point annotation method provided in this disclosure, the above-mentioned determination of at least two target third device reference points from the data source point set includes:

[0024] Determine at least one set of third device reference points from the set of data source points; the set of third device points includes at least two third device reference points, and each third device reference point is a device reference point that has a binding relationship with a device detection point.

[0025] A first closed figure is constructed based on the second equipment reference point and each set of third equipment reference points, and a target third equipment reference point set is determined from the third equipment reference point set based on the area of ​​each first closed figure, thereby obtaining at least two target third equipment reference points.

[0026] According to the device point annotation method provided in this disclosure, the above-mentioned method for determining target detection points matching the second device reference points from each second candidate detection point based on the similarity between each second closed shape and at least two target third device reference points corresponding to the target first closed shape includes:

[0027] Calculate the ratio of the side length of each second closed figure to the side length of the target first closed figure;

[0028] Calculate the angle difference between the interior angle of each second closed figure and the interior angle of the target first closed figure;

[0029] Based on the side length ratio and angle difference value of each second closed figure, determine the similarity between each second closed figure and the target first closed figure;

[0030] Based on each similarity, the target detection point is determined from each second candidate detection point.

[0031] According to the device point annotation method provided in this disclosure, the above-mentioned determination of the device detection point set corresponding to the 3D scene model based on the metadata corresponding to the 3D scene model includes:

[0032] Obtain at least one set of metadata corresponding to the 3D scene model, and perform target detection and feature extraction on each set of metadata to determine the initial device detection point set corresponding to each set of metadata; the initial device detection point set includes the initial detection points.

[0033] For each initial detection point, the initial detection points belonging to the same device are removed to determine the set of device detection points corresponding to the 3D scene model.

[0034] This disclosure also provides a device for marking points on equipment, comprising:

[0035] The determination module is configured to determine the set of device detection points corresponding to the 3D scene model based on the metadata corresponding to the 3D scene model; the set of device detection points includes multiple device detection points.

[0036] The matching and annotation module is configured to match the device detection points in the device detection point set with the device reference points in the preset data source point set, and to annotate the device points in the 3D scene model according to the matching results; wherein, the aforementioned data source point set is a set of device points in the real scene corresponding to the 3D scene model, including multiple device reference points.

[0037] This disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the device dot annotation method as described above.

[0038] This disclosure also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the device point annotation method as described above.

[0039] This disclosure also provides a computer program product, including a computer program that, when executed by a processor, implements the device point annotation method as described above.

[0040] The device point annotation method, apparatus, device, and storage medium disclosed herein determine the set of device detection points corresponding to the 3D scene model based on the metadata corresponding to the 3D scene model. Then, each device detection point in the device detection point set is matched with each device reference point in the data source point set, and the device points in the 3D scene model are annotated according to the matching results. The data source point set is a set of device points in the real scene corresponding to the 3D scene model, and includes multiple device reference points. This method automatically annotates the device points in the 3D scene model based on the matching results, eliminating the need for manual identification and annotation of device points. This reduces the manpower and time required for device point annotation, saving both labor and time costs. Furthermore, because the device points are annotated after matching them with device reference points in the real scene, the annotated device points are more accurate and reflect the actual situation. Attached Figure Description

[0041] To better describe and illustrate embodiments and / or examples of the inventions disclosed herein, reference may be made to one or more accompanying drawings. The appendices or examples used to describe the drawings should not be considered as limiting the scope of any of the disclosed inventions, the embodiments and / or examples currently described, or the best mode of these inventions as currently understood.

[0042] Figure 1 is a schematic flowchart of one of the device dot annotation methods provided in certain embodiments of this disclosure;

[0043] Figure 2 is a second schematic flowchart of a device dot annotation method provided in some embodiments of this disclosure;

[0044] Figure 3 is a third schematic flowchart of a device dot annotation method provided in some embodiments of this disclosure;

[0045] Figure 4 is a schematic diagram of the distribution of candidate detection points provided in some embodiments of this disclosure;

[0046] Figure 5 is a fourth flowchart illustrating a device dot annotation method provided in certain embodiments of this disclosure;

[0047] Figure 6 is a schematic diagram of a triangular surface formed by device reference points provided in some embodiments of this disclosure;

[0048] Figure 7 is a schematic diagram of graphic similarity comparison provided in some embodiments of this disclosure;

[0049] Figure 8 is a detailed flowchart illustrating the device dot annotation method provided in some embodiments of this disclosure;

[0050] Figure 9 is a schematic diagram of the structure of the device dot marking device provided in some embodiments of this disclosure;

[0051] Figure 10 is a schematic diagram of the structure of an electronic device provided in some embodiments of this disclosure. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this disclosure clearer, the technical solutions of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0053] With the continuous development of the security field, the application of video services in 3D scenes is increasing, and the demand for displaying device points in the scene is also increasing. Currently, there are two solutions for marking points in 3D scenes. One solution is to manually find and mark device points on the map. There are often tens of thousands of device points in the scene model, which requires a lot of manpower and is very inefficient. Moreover, if the marking is incorrect due to human error, it needs to be remarked, which will consume a lot of time. The other solution is to obtain the latitude and longitude list of devices from the platform and directly place the points, combined with manual correction. However, since the scene model is obtained by drone photography and manual processing or by manual modeling, the latitude and longitude of the scene model has deviations. Directly placing points based on the obtained latitude and longitude often results in offsets. Based on this, this disclosure provides a device point marking method, device, equipment, and storage medium to solve this technical problem.

[0054] The equipment point labeling method of this disclosure is described below with reference to Figures 1-8.

[0055] First, let me explain the 3D scene model disclosed herein, which can also be called a 3D map. This 3D map contains corresponding scenes, such as houses and roads, and multiple devices are marked on the 3D map. For example, a camera icon can be used to mark a device point on the 3D map (meaning there is a device at that location). By marking or calibrating the devices in the 3D scene model, backend personnel can quickly and accurately obtain the data of the corresponding devices and perform subsequent operations, such as viewing live playback or alarms, or simply viewing live playback. For device point marking, device points can generally be quickly identified and marked when the devices are relatively dispersed and there is no interference from other similar devices in close proximity. However, there may be cases where some devices are closer together or there are errors in the algorithm's recognition, requiring further marking of the devices in these cases. The embodiments of this disclosure can mark device points in all situations.

[0056] It should be noted that the execution subject of the device point annotation method disclosed herein can be a device point annotation device, an electronic device, or a computer device including such an electronic device. The following embodiments will use an electronic device as an example to illustrate the device point annotation method disclosed herein.

[0057] Figure 1 is a flowchart illustrating one of the device dot annotation methods provided in certain embodiments of this disclosure. Referring to Figure 1, the method may include the following steps:

[0058] S102, Based on the metadata corresponding to the 3D scene model, determine the set of device detection points corresponding to the 3D scene model; the set of device detection points includes multiple device detection points.

[0059] As mentioned above, the 3D scene model can be obtained by modeling a real, fixed scene, such as a factory area, industrial park, or residential area. This fixed scene contains one or more devices, such as cameras, snapshot cameras, or other similar equipment. The metadata of this 3D scene model is mainly used to model the 3D scene; that is, it constructs the 3D scene model through metadata. This metadata can be image data of the fixed scene. The metadata of this 3D scene model can be data obtained by using drones to photograph the real, fixed scene, or it can be data pre-stored in the cloud or other locations and retrieved from the cloud when needed.

[0060] The 3D scene model can be a high-precision oblique photogrammetry or a finely modeled model, or it can be a photogrammetry or modeling model with average precision.

[0061] After obtaining the metadata of the 3D scene model, the devices in the metadata can be detected and their features extracted to obtain device points and device information in the 3D scene model. The device information can include the location of the device, the bounding box information of the device, and the type of the device. Then, all the device points can be combined to form a device point set, denoted as the device detection point set. This device detection point set includes multiple detected device detection points, and each device detection point represents a device or a device location (i.e., a detected device position).

[0062] It should be noted that the set of device detection points obtained here can include all device detection points in a real, fixed scene, or it can include only a portion of the device detection points. Furthermore, this set of device detection points can include duplicate device detection points or it can include non-duplicate device detection points.

[0063] S104: Match the device detection points in the device detection point set with the device reference points in the preset data source point set, and annotate the device points in the 3D scene model according to the matching results.

[0064] The aforementioned data source set is a collection of device points in the real scene corresponding to the 3D scene model, including multiple device reference points. The real scene here refers to the aforementioned fixed real scene, which typically includes multiple devices. When multiple devices are pre-installed or configured, their actual location and device type can be obtained through methods such as device calibration. Each device can then report its actual location and device type to the backend electronic device (or middleware), allowing the electronic device to obtain this information. Alternatively, the electronic device can obtain this information from the middleware when it needs it. Each device can be designated as a device reference point, and the electronic device can combine these reference points to form a set, thus obtaining the data source set.

[0065] After obtaining the data source point set and the device detection point set, each device reference point in the data source point set can be compared and matched with each device detection point in the device detection point set to find a matching detection point for each device detection point. Then, using the relevant information from the device reference point, the matched device detection points are labeled as device points in the 3D scene model. This device point labeling can involve assigning relevant information from the device reference point to the matching device detection point in the 3D scene model. This relevant information may include, for example, the device type, the device's IP (Internet Protocol) address, the device name, and the device's location.

[0066] The aforementioned device point annotation method determines the set of device detection points corresponding to the 3D scene model based on the metadata of the 3D scene model. Then, it matches each device detection point in the device detection point set with each device reference point in the data source point set, and annotates the device points in the 3D scene model based on the matching results. The data source point set is the set of device points in the real scene corresponding to the 3D scene model, and includes multiple device reference points. This method automatically annotates the device points in the 3D scene model based on the matching results, eliminating the need for manual identification and annotation. This reduces the manpower and time required for device point annotation, saving both labor and time costs. Furthermore, because the device points are matched with device reference points in the real scene before annotation, the annotated device points are more accurate and reflect the actual situation.

[0067] The following examples illustrate the process of determining the set of device detection points using metadata.

[0068] Figure 2 is a second schematic flowchart of a device dot annotation method provided in certain embodiments of this disclosure. Referring to Figure 2, the above-mentioned S102 may include the following steps:

[0069] S202, obtain at least one set of metadata corresponding to the 3D scene model, and perform target detection and feature extraction on each set of metadata to determine the initial device detection point set corresponding to each set of metadata; the initial device detection point set includes the initial detection point.

[0070] In this step, a drone with multiple cameras can be used to film a fixed scene. Each camera can obtain a set of image data (a set of image data may include one or more images), thus obtaining multiple sets of image data. Alternatively, multiple sets of image data can be obtained manually. Here, multiple sets of image data are also known as multiple sets of metadata.

[0071] Next, a deep learning model can be used to perform object detection and feature detection on each set of metadata, obtaining the detection results for each set of metadata. The detection results include the set of device points detected in that set of metadata, the detection location of each device point in the set of device points, and the device type. Here, the set of device points is denoted as the initial device detection point, and each device point in it is denoted as the initial detection point. The specific architecture and type of the deep learning model mentioned above are not specifically limited here.

[0072] For example, assuming there are l sets of metadata, then l sets of initial device detection points can be obtained, as follows:

[0073] Where P represents the initial set of equipment detection points, and the subscript of P indicates which group of initial equipment detection point set; (x, y, z, t) represents the location information of an initial detection point, and its subscript indicates which initial detection point in which group of initial equipment detection point set, where x, y, z, and t represent longitude, latitude, altitude, and equipment type, respectively; n is the number of initial detection points in the initial equipment detection point set, which is generally greater than or equal to 1.

[0074] S204, remove initial detection points belonging to the same device from each initial detection point to determine the set of device detection points corresponding to the 3D scene model.

[0075] In this step, the multiple sets of initial equipment detection points obtained above may contain duplicate data for the same equipment point, or there may be interfering points. In order to perform the subsequent matching and labeling process more accurately, the data in the initial equipment detection point set can be optimized using the Euclidean formula to improve the accuracy of the final data.

[0076] Specifically, we can first iterate through all the initial detection points in each group's initial device detection points, and then compare each initial detection point with other initial detection points pairwise. We set a threshold Δt to limit the matching range. Only points with a distance less than the threshold are considered as possible matching detection points. The process is as follows:

[0077] Choose any point 'a', and iterate through the data of all initial detection points in each initial device detection point set in the list. Calculate the distance to point 'a' using the Euclidean distance formula, as follows:

[0078] Where Δd1 represents the Euclidean distance between the initial detection point a and other initial detection points; (x1, y1, z1) represents the position coordinates of the initial detection point a, and (x2, y2, z2) represents the position coordinates of other initial detection points.

[0079] When Δd1 < Δt, and the t values ​​in the initial detection point information are consistent, these points may belong to the same device, so the point is removed (one point 'a' can be retained during removal, and the others can be deleted). To avoid accidental deletion, the Δt value should be as small as possible, and should be within the distance of one device (e.g., the width of the device).

[0080] The distance calculation and threshold comparison steps described above can then be repeated continuously to ultimately obtain the set of device detection points in the 3D scene model, as follows: P (x,y,z,t) =[(x1,y1,z1,t1),(x2,y2,z2,t2),(x3,y3,z3,t3),…(x m ,y m ,z m ,t m )]

[0081] Where m is the number of equipment detection points, which is generally greater than or equal to 1, and m can be less than or equal to n.

[0082] In this embodiment, an initial set of device detection points is obtained by performing target detection and feature extraction on each group of metadata in the 3D scene model. Initial detection points belonging to the same device in the initial set of device detection points are removed, which can improve the accuracy of the final set of device detection points. In addition, the initial detection points are removed by Euclidean distance calculation and threshold comparison. This process is simple and intuitive, so it can also improve the efficiency of obtaining the set of device detection points.

[0083] The following examples illustrate the matching process between device reference points in the data source point set and device detection points in the device detection point set.

[0084] Figure 3 is a third schematic flowchart of a device dot annotation method provided in some embodiments of this disclosure. Referring to Figure 3, the matching step in S104 above may include the following steps:

[0085] S302, traverse each device reference point in the data source point set, calculate the first distance between each device reference point and each device detection point, and obtain the first distance set corresponding to each device reference point.

[0086] The set of data source points can be represented by the following formula: P' (x,y,z,t) =[(x'1,y'1,z'1,t'1),(x'2,y'2,z'2,t'2),…(x' k ,y' k ,z' k ,t' k )]

[0087] Where k refers to the total number of devices in the real scene, and k is generally greater than or equal to 1; P' represents the set of device reference points; (x', y', z', t') represents the location information of a device reference point, where x', y', z', and t' represent longitude, latitude, altitude, and device type, respectively.

[0088] In this step, P' can be traversed. (x,y,z,t) Each device reference point in the P (x,y,z,t) Distance calculations are performed on each device detection point to identify those that meet threshold conditions. Specifically, the distance between each device reference point and each device detection point is calculated based on the location information of each device reference point and each device detection point, and is denoted as the first distance.

[0089] When calculating the distance between two points, the common method is to use Euclidean distance. However, the points in this solution are three-dimensional, and considering that the three-dimensional scene is on a sphere, to calculate the distance between two points more accurately, the curvature and elevation changes of the surface need to be taken into account. Therefore, the distance between the two points can be calculated using the great circle distance formula to improve accuracy. Before calculating the distance, the arc length between the two points on the sphere can be calculated first, and then the great circle distance between the two points can be obtained from the arc length. The formula for calculating the arc length between two points on a sphere is as follows: Δσ=arccos(siny1·siny2+cosy1·cosy2·cos(x1-x2))

[0090] Where Δσ is the arc length between the two points on the sphere, and the great circle distance between the two points (i.e., the first distance between each equipment reference point and each equipment detection point) is denoted as Δd2, and its calculation formula is: Δd2=R·Δσ

[0091] Furthermore, considering the influence of elevation changes, the Δd2 distance can be transformed into the following formula: Δd2'=R·Δσ+|z2-z1|

[0092] Where x1 and x2 are the longitudes of the two points, y1 and y2 are the latitudes of the two points, z2 and z1 are the altitudes of the two points (here, the units of longitude and latitude are radians), and R is the average radius of the Earth, which is generally 6371 kilometers.

[0093] The first distance between each device reference point and each device detection point can be calculated using the above method. Each device reference point will obtain multiple first distances. Then, the multiple first distances of each device reference point can be combined to obtain the set of first distances of each device reference point.

[0094] S304, Based on the first distance set of each device reference point, determine the candidate detection points that match each device reference point from each device detection point.

[0095] In this step, after obtaining the first distance set for each device reference point, multiple first distances for each device reference point can be compared with the great circle distance threshold. Taking the first distance set of a device reference point as an example, if a certain first distance is less than the great circle distance threshold (e.g., Δd), the device detection point corresponding to the first distance can be regarded as a suspected matching point with the device reference point, that is, a point that may match the device reference point; otherwise, it is not regarded as a suspected matching point.

[0096] By repeating the above steps, a possible / probable matching point can be determined for each device reference point at each device detection point, and these points are recorded as candidate detection points.

[0097] In addition, the above-mentioned device type parameter t can be used to remove candidate detection points that do not match the type of each device reference point (generally, the device types of the matching device reference point and the device detection point should be the same), which can further improve the accuracy of the obtained matching candidate detection points.

[0098] S306, determine the corresponding set of candidate detection points based on the candidate detection points corresponding to each device reference point; the set of candidate detection points includes at least one candidate detection point.

[0099] In this step, each of the aforementioned device reference points may have one or more candidate detection points. Refer to Figure 4, which shows a schematic diagram of the distribution of candidate detection points provided in certain embodiments of this disclosure. The black dots represent device reference points in the data source point set, such as P'1, P'2, P'3, P'5, P'6, and P' ... 11 、P' 12 Hollow circles represent candidate detection points in the device detection point set that meet the threshold condition (i.e., the large circle distance threshold mentioned above), such as point P1. Hollow circles that are relatively widely dispersed indicate a one-to-one match. Black dots corresponding to hollow circles with confirmed matches are not drawn in the diagram. The circular filled area in the diagram illustrates the case where there are multiple candidate detection points that meet the threshold condition, meaning that one device reference point has multiple potentially matching candidate detection points.

[0100] Taking a candidate detection point for a device reference point as an example, the candidate detection points corresponding to the device reference point can be sorted in ascending order according to their first distance and combined into a set to obtain the candidate detection point set corresponding to the device reference point. The candidate detection point set corresponding to each device reference point can be obtained in this way, as follows:

[0101] Where p' represents the device reference point, and the subscript indicates which device reference point it is.

[0102] In this embodiment, the distance between each device reference point and each device detection point is calculated, and candidate detection points and a set of candidate detection points that are likely to match each device reference point are determined from each device detection point based on the distance. Matching by distance calculation can improve the efficiency and accuracy of matching between device reference points and device detection points.

[0103] The above embodiments propose that the candidate detection point set may include one candidate detection point or multiple candidate detection points. The following embodiments will explain the device point annotation process in these two cases.

[0104] In one embodiment, if the set of candidate detection points corresponding to the first device reference point includes a first candidate detection point, then the step of labeling based on the matching result in S104 above may include the following steps:

[0105] Establish a binding relationship between the first device reference point and the device detection point corresponding to the first candidate detection point, and mark the information corresponding to the first device reference point on the device detection point corresponding to the first candidate detection point in the 3D scene model.

[0106] In this step, if the set of candidate detection points corresponding to a device reference point includes a candidate detection point, the device reference point can be recorded as the first device reference point, and the corresponding candidate detection point can be recorded as the first candidate detection point.

[0107] For a single device reference point, when there is only one candidate detection point that meets the threshold condition, as shown in Figure 4, P'1 and P' 11 、P' 12 If a candidate detection point is identified, it can be directly determined that the candidate detection point is the matching point corresponding to the corresponding device reference point, and the relationship between the device reference point and the corresponding candidate detection point can be directly bound.

[0108] It's important to note that the punctuation / annotation is performed within the 3D scene model. Therefore, this involves reverse-engineering the device reference points from the data source point set onto the map corresponding to the 3D scene model and binding them to the corresponding devices on the map. For example, point P'1 can be bound as follows:

[0109] In other words, in the 3D scene model, the name, IP address, and other information of the device reference point can be assigned to a matching candidate detection point. However, the position of the candidate detection point can be the position detected by the above target detection. This allows the point to be accurately marked on the 3D scene model, making the two positions more visually matched.

[0110] In this embodiment, when there is only one candidate detection point corresponding to the device reference point, the two can be directly bound together and the information of the device reference point can be marked on the corresponding candidate detection point. This allows the point to be accurately marked on the 3D scene model, making the positions of the two more visually matched.

[0111] In another scenario, if the candidate detection point set corresponding to the second device reference point includes multiple second candidate detection points, and the aforementioned second device reference point is a device reference point without an established binding relationship, referring to the flowchart of the device point annotation method provided in certain embodiments of this disclosure shown in Figure 5, the step of annotating based on the matching result in S104 may include the following steps:

[0112] S402, determine at least two target third device reference points from the data source point set; each target third device reference point is a device reference point that has a binding relationship with the device detection point.

[0113] In this step, if a set of candidate detection points corresponding to a device reference point includes multiple candidate detection points, the device reference point can be designated as the second device reference point, and all the corresponding candidate detection points can be designated as second candidate detection points. Here, the second device reference point is a point that has not established a binding relationship with a device detection point.

[0114] Since there must be some kind of relationship between the pairwise matching points in the set of device detection points corresponding to the 3D scene model and the set of data source points, and simply calculating one or two points to determine the relationship is prone to errors, this embodiment adopts the method of simultaneously confirming the relationship between multiple points to find the matching point. Triangular faces with longer sides have higher geometric clarity and are less likely to degenerate into lines or points, making them easier to identify and process, and they have greater numerical stability, reducing the impact of calculation errors. Therefore, this embodiment mainly uses the method of establishing the largest triangular face, and through the principle of triangle similarity (but not limited to triangles), finds the point that matches the device reference point from multiple candidate detection points. Of course, this is not limited to triangular faces / triangles; polygons can also be used.

[0115] The following explanation uses triangular facet / triangle matching as an example. First, at least two device reference points that have already established a binding relationship with the device detection point can be determined from the data source point set; these are denoted as target third device reference points. As an optional embodiment, the process of determining the target third device reference points may include the following steps:

[0116] Step A1: Determine at least one set of third device reference points from the set of data source points; the set of third device points includes at least two third device reference points, and each third device reference point is a device reference point that has established a binding relationship with a device detection point.

[0117] Step A2: Construct a first closed figure based on the second device reference point and each set of third device reference points, and determine the target third device reference point set from the third device reference point set based on the area of ​​each first closed figure, thereby obtaining at least two target third device reference points.

[0118] The third device reference point is a point that has already established a one-to-one binding relationship with the device detection point. Taking a triangular face or triangle as an example, two third device reference points need to be obtained each time as a set of third device reference points.

[0119] Simultaneously, any second device reference point that has not yet established a binding relationship can be obtained from the set of device reference points. Then, the points in each set of third device reference points are used to construct a closed figure with the second device reference point, which is denoted as the first closed figure. Taking a triangle as an example, the first closed figure constructed here can be a triangle, that is, three points form a triangle.

[0120] Each set of third device reference points will construct a first closed shape with the second device points, resulting in multiple first closed shapes for the second device reference points. The area of ​​each first closed shape can then be calculated, and the largest area can be selected. Simultaneously, the three points corresponding to the first closed shape with the largest area are obtained; one of these points is the second device reference point, and the other two points can be designated as the target third device reference point.

[0121] For example, taking a triangular facet or triangle as an example, referring to Figure 6, which shows a schematic diagram of a triangular facet formed by device reference points provided in certain embodiments of this disclosure, it can be seen from P'( x,y,z,t Select any point in the list that has not yet been bound, denoted as p'3 (the second device reference point), and iterate through P' ( x,y,z,t For points whose binding relationship has been confirmed, find two points, calculate the area of ​​the triangle they form, and sort them by size. Assume the two points (the third device reference point) are p' 11 ,p' 12 The area of ​​the triangle formed by these y = f(x) (the dashed triangle in the figure) is S. The formula for calculating the area is as follows:

[0122] Wherein, the cross product p'3p' 11 ×p'3p' 12 The modulus is the area of ​​the triangle;

[0123] By calculation, the coordinates of the three vertices with the largest area can be obtained, assumed to be: p'3(x′3,y′3,z′3),p' 11 (x′ 11 ,y′ 11, z′ 11 ),p' 12 (x′ 12 ,y′ 12 ,z′ 12 ).

[0124] Among them, due to p' 11 ,p' 12 Since a binding relationship has been established, it is known that the two points correspond to device detection points in the 3D scene model, assumed to be: p 11 p 12 .

[0125] S404, construct each second closed pattern based on each second candidate detection point and the device detection point corresponding to each target third device reference point.

[0126] In this step, after identifying at least two target third device reference points, since each target third device reference point has already established a one-to-one binding relationship with its corresponding device detection point, the device detection points bound to each target third device reference point can be obtained. Then, for each second candidate detection point in the candidate detection point set corresponding to the second device reference point, a closed graph can be constructed with the device detection points bound to each target third device reference point, denoted as the second closed graph, ultimately resulting in multiple second closed graphs.

[0127] Here, the number of second closed shapes is the same as the number of second candidate detection points; that is, a second closed shape is constructed for each second candidate detection point. It can be seen that the second closed shape here is a shape constructed between device detection points, while the first closed shape mentioned above is a shape constructed between device reference points. Subsequently, the similarity between these two shapes / triangles, but not limited to the principle of triangle similarity, can be used to select device detection points that match the second device reference points, thereby improving the accuracy of the matching.

[0128] S406, based on the similarity between each second closed pattern and the target first closed pattern corresponding to at least two target third device reference points, determine the target detection point that matches the second device reference point from each second candidate detection point.

[0129] In this step, as mentioned above, after determining at least two target third device reference points, the first closed pattern formed by them can be obtained simultaneously, denoted as the target first closed pattern. Then, the similarity between each second closed pattern and the target first closed pattern can be compared. As an optional embodiment, this step can select matching device detection points based on similarity through the following steps:

[0130] Step B1: Calculate the ratio of the side length of each second closed figure to the side length of the target first closed figure.

[0131] Step B2: Calculate the angle difference between the interior angle of each second closed figure and the interior angle of the target first closed figure.

[0132] Step B3: Determine the similarity between each second closed figure and the target first closed figure based on the side length ratio and angle difference value of each second closed figure.

[0133] Step B4: Determine the target detection point from each second candidate detection point based on each similarity.

[0134] Continuing with the selection of p'3 (second device reference point) and p' in Figure 6 above... 11 ,p' 12 For example, referring to Figure 7, a schematic diagram of graphic similarity comparison provided in certain embodiments of this disclosure, the dashed triangle formed by three black dots is the first closed shape (i.e., the triangle formed by the device reference points), and the solid triangle formed by three hollow circles is the second closed shape (i.e., the triangle formed by the candidate detection points of the device). The multiple second candidate detection points p3 corresponding to p'3 are traversed and compared with p... 11 p 12 Constructing triangular faces / triangles yields multiple second closed figures, i.e., multiple triangular faces / triangles. To find the best matching point, a similarity evaluation value (or similarity score) S is introduced. s S s The similarity score is represented by a weighted average of the side length ratio and angle differences. The following is the method for calculating the similarity score:

[0135] 1. Assume the second closed figure is Δp3p 11 p 12 With the first closed figure Δp'3p' 11 p' 12 If the lengths of the three sides are a, b, c and a', b', c', then the ratio of the side lengths R' can be expressed by the following formula:

[0136] The average side length ratio is expressed as:

[0137] The average ratio of the side length of each second closed figure to the side length of the target first closed figure can be calculated using the above formula.

[0138] 2. Regarding the above angular difference values, assume the second closed figure is Δp3p 11 p 12 With the first closed figure Δp'3p' 11 p' 12 The edge vectors are respectively and Each interior angle can be obtained using the dot product of vectors and the cosine formula:

[0139] Where i represents the label of any interior angle in the first closed figure / second closed figure, and any interior angle θ in the first closed figure / second closed figure can be obtained by solving the arccosine.

[0140] Angular difference value A v The following formula can be used for calculation:

[0141] The above formula can be used to calculate the angular difference between the interior angle of each second closed figure and the interior angle of the target first closed figure (specifically, the average angular difference).

[0142] 3. Calculate the similarity score / similarity evaluation value S s The following formula can be used to calculate S: s =ω1·R v +ω2·A v

[0143] Among them, R v Average side length ratio, A v The average angle difference value is represented by ω1 and ω2, which are weight parameters. Generally, the sum of ω1 and ω2 is 1. Usually, the value of ω1 can be greater than the value of ω2, that is, the proportion of the side length ratio is larger, which can further improve the accuracy of point matching.

[0144] Furthermore, to improve the accuracy of point calibration, a value of ω is added here, which represents the weight calculated based on distance, as shown in the following formula:

[0145] in, Δd i Let be the great circle distance between the three corresponding matching points of the two triangular faces / triangles; σ is a parameter controlling the weight decay rate, which is a known quantity; N represents the number of edges of the closed figure; d I This represents the average distance.

[0146] The weighted value ω and the similarity evaluation value S s After normalization, the final similarity evaluation value S is obtained. s 'For: S s '=ω·S s .

[0147] The similarity / similarity evaluation value between each second closed shape and the target first closed shape can be calculated using the above method. Generally, the smaller the similarity evaluation value, the higher the similarity. Then, by comparing the similarity evaluation values ​​corresponding to each second closed shape, the smallest similarity evaluation value is selected, and the second closed shape corresponding to this smallest similarity evaluation value is obtained. This leads to the three device detection points constituting the second closed shape, two of which are already bound to each other, and the remaining point is the device detection point with the highest matching degree to the second device reference point, denoted as the target detection point.

[0148] S408, establish a binding relationship between the second device reference point and the device detection point corresponding to the target detection point, and mark the information corresponding to the second device reference point on the device detection point corresponding to the target detection point in the 3D scene model.

[0149] In this step, after determining the target detection point that uniquely matches the second device reference point, the device detection point corresponding to the second device reference point can be marked in the 3D scene model. Here, the latitude, longitude, and altitude of the device point in the 3D scene model are used for marking / annotating. This operation is mainly for situations where the point needs to be displayed on the interface via an icon when marking points on the map (generally, it is sufficient to directly establish a binding relationship, but in some cases, an additional icon needs to be added to the corresponding device point in the scene). This can optimize the problem of the difference between the position of the point icon in the scene and the actual marked point position.

[0150] In other words, the name, IP address, and other information of the second device reference point can be assigned to the device detection point corresponding to the matching target detection point. However, the position of the target detection point can be the position detected by the above target detection. This allows the point to be accurately marked on the 3D scene model, making the positions of the two more visually matched.

[0151] For the remaining unlabeled / uncalibrated device reference points in the above data source point set, the labeling can be repeated according to the steps S402-S408 above, thereby completing the labeling of all device reference points.

[0152] In this embodiment, a unique matching point is determined based on the similarity between the closed shape formed by device reference points and the closed shape formed by device detection points. This method of finding matching points through the relationships between multiple points significantly reduces the error in determining matching points, improves the stability and accuracy of the determined matching points, and thus greatly improves the accuracy of device point labeling. Furthermore, the area of ​​the closed shape formed by the device reference points is used to determine the device reference point that forms the closed shape with the unmatched device reference points from among the multiple device reference points. This improves the accuracy of the determined device reference points, thereby improving the accuracy of the reference data used in the graphic similarity comparison, and further improving the accuracy of the determined matching points. Additionally, obtaining similarity using multiple parameters such as the side length ratio and angle difference value of the graphics can improve the precision of the obtained similarity, thus improving the accuracy of the results obtained through similarity comparison.

[0153] Furthermore, in the above process of comparing the similarity of triangular faces / triangles, the result of matching the most similar triangular face will usually only yield one most similar triangular face, meaning that only one set of points that simultaneously satisfies a certain rule will appear. However, it is also possible to match two or more similar triangular faces. When this happens, we can add adjacent points to form a polygon, and then find the corresponding matching points by judging the similarity relationship of the polygon. Since a polygon can be regarded as being composed of multiple triangular faces, the above process of calculating the similarity of triangles can still be used. When the corresponding triangular faces / triangles that make up the polygon simultaneously satisfy the similarity condition, it can be considered that two or more sets of points have been found and a matching relationship exists.

[0154] The following is a detailed embodiment. Please refer to Figure 8 for a detailed flowchart of some embodiments of this disclosure. This solution can first obtain the metadata of the 3D scene model (i.e., map data metadata) and the corresponding data source point set (if the device is a camera, it can be the camera list data source P1). Then, the target in the metadata can be detected and its features extracted by a deep learning model to obtain the device detection point set (i.e., device list P2) in the 3D scene model. After that, for each point in the data source point set, the candidate detection point set that meets the threshold condition in the device detection point set (i.e., the list of each point that meets the threshold condition) can be found. Then, triangles or triangular faces are introduced, and a similarity evaluation method is used to find the points that match each other and mark or label them in the 3D scene model.

[0155] In this embodiment, by acquiring the map metadata corresponding to the 3D scene model, and extracting the devices in the scene through feature extraction, the data source of the devices is calculated. Then, device points that meet the threshold conditions are filtered out, and triangular faces are established for the selected points. A similarity evaluation value is introduced, and the relationship between the points is determined through the relationship between the faces, thereby determining the relationship between the points and binding them. At the same time, the position of the device points is corrected (that is, the position information during the above calibration uses the detected position information). This can reduce the manpower and time spent on calibration, and can also make the points accurately calibrated on the 3D scene model, making the positions of the two more visually matched.

[0156] The equipment point marking device provided in this disclosure is described below. The equipment point marking device described below and the equipment point marking method described above can be referred to in correspondence.

[0157] Figure 9 is a schematic diagram of the structure of a device dot marking apparatus provided in some embodiments of this disclosure. Referring to Figure 9, the apparatus may include:

[0158] The determination module 510 is configured to determine the set of device detection points corresponding to the 3D scene model based on the metadata corresponding to the 3D scene model; the set of device detection points includes multiple device detection points.

[0159] The matching and annotation module 520 is configured to match the device detection points in the device detection point set with the device reference points in the preset data source point set, and to annotate the device points in the 3D scene model according to the matching results; wherein, the aforementioned data source point set is a set of device points in the real scene corresponding to the 3D scene model, including multiple device reference points.

[0160] In one exemplary embodiment, the matching and annotation module 520 described above may include:

[0161] The first calculation unit is configured to traverse each device reference point in the data source point set, calculate the first distance between each device reference point and each device detection point, and obtain the first distance set corresponding to each device reference point;

[0162] The candidate point determination unit is configured to determine candidate detection points that match each device reference point from each device detection point based on a first distance set for each device reference point;

[0163] The candidate point set determination unit is configured to determine a corresponding candidate detection point set based on the candidate detection points corresponding to each device reference point; the candidate detection point set includes at least one candidate detection point.

[0164] In an exemplary embodiment, if the candidate detection point set corresponding to the first device reference point includes a first candidate detection point, the matching and annotation module 520 is specifically configured to establish a binding relationship between the first device reference point and the device detection point corresponding to the first candidate detection point, and to annotate the information corresponding to the first device reference point on the device detection point corresponding to the first candidate detection point in the three-dimensional scene model.

[0165] In an exemplary embodiment, if the candidate detection point set corresponding to the second device reference point includes multiple second candidate detection points, and the second device reference point is a device reference point without an established binding relationship, the matching and labeling module 520 may further include:

[0166] The target reference point determination unit is configured to determine at least two target third device reference points from the data source point set; each target third device reference point is a device reference point that has a binding relationship with a device detection point.

[0167] The graph construction unit is configured to construct each second closed graph based on each second candidate detection point and the device detection point corresponding to each target third device reference point;

[0168] The target detection point determination unit is configured to determine a target detection point matching the second device reference point from each second candidate detection point based on the similarity between each second closed shape and the target first closed shape corresponding to at least two target third device reference points;

[0169] The annotation unit is configured to establish a binding relationship between the second device reference point and the device detection point corresponding to the target detection point, and to annotate the information corresponding to the second device reference point on the device detection point corresponding to the target detection point in the 3D scene model.

[0170] Optionally, the target reference point determination unit is specifically configured to determine at least one set of third device reference points from the data source point set; the third device point set includes at least two third device reference points, each of which is a device reference point that has established a binding relationship with a device detection point; construct a first closed shape based on the second device reference point and each set of third device reference points, and determine a target third device reference point set from the set of third device reference points based on the area of ​​each first closed shape, thereby obtaining at least two target third device reference points.

[0171] Optionally, the target detection point determination unit is specifically configured to: calculate the ratio of the side length of each second closed figure to the side length of the target first closed figure; calculate the angle difference between the interior angle of each second closed figure and the interior angle of the target first closed figure; determine the similarity between each second closed figure and the target first closed figure based on the side length ratio and angle difference; and determine the target detection point from each second candidate detection point based on each similarity.

[0172] In an exemplary embodiment, the determining module 510 is specifically configured to acquire at least one set of metadata corresponding to the three-dimensional scene model, and perform target detection and feature extraction on each set of metadata to determine the initial device detection point set corresponding to each set of metadata; the initial device detection point set includes initial detection points; and the initial detection points belonging to the same device in each initial detection point are removed to determine the device detection point set corresponding to the three-dimensional scene model.

[0173] Figure 10 illustrates a schematic diagram of the physical structure of an electronic device. As shown in Figure 10, the electronic device may include: a processor 610, a communication interface 620, a memory 630, and a communication bus 640. The processor 610, communication interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a device point annotation method. This method includes: determining a set of device detection points corresponding to the 3D scene model based on the metadata corresponding to the 3D scene model; the set of device detection points includes multiple device detection points; matching the device detection points in the set of device detection points with device reference points in a preset set of data source points, and annotating the device points in the 3D scene model according to the matching results; wherein the set of data source points is a set of device points in the real scene corresponding to the 3D scene model, including multiple device reference points.

[0174] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a part 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 each embodiment of this disclosure. The aforementioned storage medium includes: a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc., each of these media capable of storing program code.

[0175] On the other hand, this disclosure also provides a computer program product, which includes a computer program that can be stored on a computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the device point annotation method provided by each of the above methods. The method includes: determining a set of device detection points corresponding to the three-dimensional scene model based on the metadata corresponding to the three-dimensional scene model; the set of device detection points includes multiple device detection points; matching the device detection points in the set of device detection points with device reference points in a preset set of data source points, and annotating the device points in the three-dimensional scene model according to the matching result; wherein the set of data source points is a set of device points in the real scene corresponding to the three-dimensional scene model, including multiple device reference points.

[0176] In another aspect, this disclosure also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a device point annotation method provided by each of the above methods. The method includes: determining a set of device detection points corresponding to the 3D scene model based on metadata corresponding to the 3D scene model; the set of device detection points includes multiple device detection points; matching the device detection points in the set of device detection points with device reference points in a preset set of data source points, and annotating the device points in the 3D scene model according to the matching result; wherein the set of data source points is a set of device points in the real scene corresponding to the 3D scene model, including multiple device reference points.

[0177] The device embodiments described above are merely illustrative. 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 modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0178] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0179] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit them. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in each of the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of each embodiment of this disclosure.

Claims

1. A method for marking equipment points, comprising: Based on the metadata corresponding to the 3D scene model, determine the set of device detection points corresponding to the 3D scene model; The set of equipment detection points includes multiple equipment detection points; The device detection points in the device detection point set are matched with the device reference points in the preset data source point set, and the device points in the three-dimensional scene model are labeled according to the matching results. The data source point set is a set of device points in the real scene corresponding to the 3D scene model, including multiple device reference points.

2. The equipment point marking method according to claim 1, wherein, The step of matching the device detection points in the device detection point set with the device reference points in the preset data source point set includes: Traverse each device reference point in the data source point set, calculate the first distance between each device reference point and each device detection point, and obtain the first distance set corresponding to each device reference point; Based on a first distance set for each of the device reference points, candidate detection points matching each of the device detection points are determined from each of the device detection points; Based on the candidate detection points corresponding to each device reference point, a corresponding set of candidate detection points is determined; the set of candidate detection points includes at least one candidate detection point.

3. The equipment point marking method according to claim 2, wherein, If the set of candidate detection points corresponding to the first device reference point includes a first candidate detection point, then the step of labeling the device points in the 3D scene model according to the matching result includes: Establish a binding relationship between the first device reference point and the device detection point corresponding to the first candidate detection point, and mark the information corresponding to the first device reference point on the device detection point corresponding to the first candidate detection point in the three-dimensional scene model.

4. The equipment point marking method according to claim 2 or 3, wherein, If the candidate detection point set corresponding to the second device reference point includes multiple second candidate detection points, and the second device reference point is a device reference point without an established binding relationship, then the step of labeling the device points in the 3D scene model according to the matching result includes: At least two target third device reference points are determined from the set of data source points; each target third device reference point is a device reference point that has a binding relationship with a device detection point; Each second closed pattern is constructed based on each second candidate detection point and the device detection point corresponding to each target third device reference point; Based on the similarity between each second closed shape and the target first closed shape corresponding to the at least two target third device reference points, a target detection point matching the second device reference point is determined from each second candidate detection point; Establish a binding relationship between the second device reference point and the device detection point corresponding to the target detection point, and mark the information corresponding to the second device reference point on the device detection point corresponding to the target detection point in the three-dimensional scene model.

5. The equipment point marking method according to claim 4, wherein, Determining at least two target third device reference points from the data source point set includes: At least one set of third device reference points is determined from the set of data source points; the set of third device points includes at least two third device reference points, and each third device reference point is a device reference point that has a binding relationship with a device detection point; A first closed figure is constructed based on the second device reference point and each set of the third device reference points, and a target third device reference point set is determined from the set of the third device reference points based on the area of ​​each first closed figure, thereby obtaining at least two target third device reference points.

6. The equipment point marking method according to claim 4, wherein, The step of determining a target detection point matching the second device reference point from each of the second candidate detection points based on the similarity between each second closed shape and the target first closed shape corresponding to the at least two target third device reference points includes: Calculate the ratio of the side length of each second closed figure to the side length of the target first closed figure; Calculate the angle difference between the interior angle of each of the second closed shapes and the interior angle of the target first closed shape; Based on the side length ratio of each second closed figure and the angle difference value, the similarity between each second closed figure and the target first closed figure is determined; The target detection point is determined from each of the second candidate detection points based on each similarity.

7. The equipment point marking method according to any one of claims 1-3, wherein, The step of determining the set of device detection points corresponding to the 3D scene model based on the metadata corresponding to the 3D scene model includes: At least one set of metadata corresponding to a 3D scene model is obtained, and target detection and feature extraction are performed on each set of metadata to determine an initial device detection point set corresponding to each set of metadata; the initial device detection point set includes initial detection points; Each initial detection point belonging to the same device is removed from the initial detection points to determine the set of device detection points corresponding to the three-dimensional scene model.

8. A device for marking equipment points, comprising: The determination module is configured to determine the set of device detection points corresponding to the 3D scene model based on the metadata corresponding to the 3D scene model. The set of equipment detection points includes multiple equipment detection points; The matching and annotation module is configured to match the device detection points in the device detection point set with the device reference points in the preset data source point set, and to annotate the device points in the 3D scene model according to the matching results; wherein, the data source point set is a set of device points in the real scene corresponding to the 3D scene model, including multiple device reference points.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the device dot annotation method as claimed in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the device point annotation method as described in any one of claims 1 to 7.

Images