Beam member identification method and electronic device

Abstract

The present application relates to a kind of beam component identification method and electronic equipment, comprising: obtaining beam construction drawing subgraph frame;Wherein, beam construction drawing subgraph frame is extracted from the building drawing to be identified, and beam construction drawing subgraph frame includes multiple beam regions and multiple sets of beam annotation information;Multiple sets of beam annotation information are found in beam construction drawing subgraph frame;Multiple sets of beam annotation information are deleted from beam construction drawing subgraph frame, obtain target subgraph frame, and multiple beam regions are detected from target subgraph frame;Multiple beam regions and multiple sets of beam annotation information are matched, to give each set of beam annotation information in multiple sets of beam annotation information to corresponding beam region, so as to improve identification efficiency.

Description

Technical Field

[0001] This invention relates to the field of architectural drawing review technology, and in particular to a method and electronic device for identifying beam components. Background Technology

[0002] Building structure refers to the system in a building, composed of various components (such as roof trusses, beams, slabs, and columns) capable of withstanding various forces. Forces refer to various factors that can cause internal forces and deformations in the system, such as loads, earthquakes, temperature changes, and foundation settlement. Structural drawings graphically display the structural form, component arrangement, and dimensions of a building or other engineering project, including the building's floor plan, the location and connection methods of structural components, and the load transfer paths.

[0003] In structural engineering, beams are a crucial component, primarily used to bear vertical gravity, but also capable of withstanding horizontal forces (such as earthquakes or wind). Beams transfer loads to columns, walls, or other beams, which in turn transfer the loads to compression members. Beams are widely designed and used, found in various building and engineering structures. They are typically connected to columns or walls, transferring loads to these supporting structures. In buildings, the main function of beams is to support floors, roofs, or bridge decks, bearing the loads on them and safely transferring these loads to the supporting structures or foundations.

[0004] Furthermore, intelligent CAD drawing review is a relatively new technological field. It primarily utilizes image processing, machine learning, and deep learning technologies to automate the review of architectural construction drawings. This not only effectively improves review efficiency but also reduces oversights or misunderstandings that may occur during manual review. In this process, the identification method for beam components in structural drawings is a crucial step. Accurately identifying the corresponding areas of beams ensures the accuracy of matching between subsequent structural drawings and calculations and provides boundary references for the later identification of slabs.

[0005] Currently, beam component identification is mainly achieved through layer recognition methods. However, beams that do not meet the layer criteria are often misidentified or discarded, and some algorithm upgrades require manual annotation to add possible beam layers. This means that fully automated image recognition is not yet possible. Therefore, existing technologies suffer from relatively low image recognition efficiency. Summary of the Invention

[0006] (a) Technical problems to be solved

[0007] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a beam component identification method and electronic device, which solves the technical problem of low drawing recognition efficiency in the prior art.

[0008] (II) Technical Solution

[0009] To achieve the above objectives, the main technical solutions adopted by the present invention include:

[0010] In a first aspect, embodiments of the present invention provide a beam component identification method, which is applied to an electronic device. The beam component identification method includes: obtaining a beam construction drawing sub-frame; wherein the beam construction drawing sub-frame is extracted from the architectural drawing to be identified, and the beam construction drawing sub-frame includes multiple beam regions and multiple sets of beam annotation information; searching for multiple sets of beam annotation information from the beam construction drawing sub-frame; deleting the multiple sets of beam annotation information from the beam construction drawing sub-frame to obtain a target sub-frame, and detecting multiple beam regions from the target sub-frame; matching the multiple beam regions and the multiple sets of beam annotation information to assign each set of beam annotation information in the multiple sets of beam annotation information to the corresponding beam region.

[0011] In one possible embodiment, each set of beam labeling information includes beam label lines and beam labeling content, and the beam labeling content includes beam number information, beam section information, and reinforcement information in the beam.

[0012] In one possible embodiment, searching for multiple sets of beam annotation information from the beam construction drawing sub-frame includes: classifying the text elements in the beam construction drawing sub-frame to obtain text element classification results; wherein, the text element classification results include beam number information classification results, beam section information classification results, and beam reinforcement information classification results; a candidate beam annotation line determination step, based on the beam number information classification results, selecting the line that is perpendicular to the direction of each beam number information and is closest to it as the corresponding candidate beam annotation line; and based on the position of each candidate beam annotation line, determining the beam section information and beam reinforcement information belonging to the same set of beam annotation information as each candidate beam annotation line.

[0013] In one possible embodiment, after determining all candidate beam annotation lines, the beam component identification method further includes: analyzing all determined candidate beam annotation lines to determine whether there are outlier candidate beam annotation lines among all candidate beam annotation lines; if there are outlier candidate beam annotation lines, deleting both non-outlier candidate beam annotation lines and outlier candidate beam annotation lines from the beam construction drawing sub-frame, and re-executing the candidate beam annotation line determination step to determine new beam annotation lines in the beam annotation information of the group to which the outlier candidate beam annotation lines belong; and using the new beam annotation lines as the final beam annotation lines in the beam annotation information of the group to which the outlier candidate beam annotation lines belong.

[0014] In one possible embodiment, outlier candidate beam annotation lines are beam annotation lines with different layer information and different line colors.

[0015] In one possible embodiment, detecting multiple beam regions from a target sub-frame includes: inputting the target sub-frame into a pre-trained beam region detection model to detect multiple beam regions; wherein the beam region detection model includes a detection head with an attention mechanism NAM module added.

[0016] In one possible embodiment, the detection head includes n convolutional layers arranged in parallel, and each of the n convolutional layers is connected in series with a NAM module; where n is a positive integer greater than 3.

[0017] In one possible embodiment, multiple beam regions and multiple sets of beam annotation information are matched to assign each set of beam annotation information to the corresponding beam region, including: determining the beam region where the endpoint of the current beam annotation line is located as the beam region corresponding to the current beam annotation line, and assigning the beam annotation information belonging to the same group as the current beam annotation line to the beam region corresponding to the current beam annotation line.

[0018] Secondly, embodiments of this application provide a storage medium storing a computer program, which, when executed by a processor, performs the method described in the first aspect or any optional implementation thereof.

[0019] Thirdly, embodiments of this application provide an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the method described in the first aspect or any optional implementation of the first aspect.

[0020] Fourthly, this application provides a computer program product that, when run on a computer, causes the computer to perform the method in the first aspect or any possible implementation thereof.

[0021] (III) Beneficial Effects

[0022] The beneficial effects of this invention are:

[0023] This application provides a beam component identification method and electronic device. By acquiring a sub-frame of a beam construction drawing, searching for multiple sets of beam annotation information from the sub-frame, and deleting the multiple sets of beam annotation information from the sub-frame, a target sub-frame is obtained. Multiple beam regions are detected from the target sub-frame, and the multiple beam regions are matched with the multiple sets of beam annotation information to assign each set of beam annotation information to the corresponding beam region. Compared with existing beam component identification methods, this method can not only accurately identify beam regions but also match them with the corresponding annotation information, greatly improving work efficiency and accuracy.

[0024] To make the above-mentioned objectives, features and advantages to be achieved by the embodiments of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A flowchart of a beam member identification method provided in an embodiment of this application is shown;

[0027] Figure 2 This paper shows a structural schematic diagram of a beam region detection model provided in an embodiment of this application;

[0028] Figure 3 This paper shows a schematic diagram of the structure of a CBL_NAM module provided in an embodiment of this application;

[0029] Figure 4 This paper shows a schematic diagram of the structure of a Res Unit module provided in an embodiment of this application;

[0030] Figure 5 This paper shows a schematic diagram of the structure of a CSPNAM1_X module provided in an embodiment of this application;

[0031] Figure 6 This paper shows a schematic diagram of the structure of a CSPNAM2_X module provided in an embodiment of this application;

[0032] Figure 7 This paper shows a schematic diagram of the structure of a Focus module provided in an embodiment of this application;

[0033] Figure 8A schematic diagram of the structure of an SPP module provided in an embodiment of this application is shown. Detailed Implementation

[0034] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] In the architectural design industry, the review of CAD drawings is a crucial step, ensuring that the designs comply with national and industry regulations and standards. Furthermore, the review process involves multiple professional fields, such as architecture, structure, interior design, water supply and drainage, gas supply, electrical systems (both high and low voltage), HVAC, municipal engineering, and landscaping.

[0036] Furthermore, the common practice during the drawing review process is to rely on experienced engineers for manual review. However, this method is inefficient and labor-intensive. Therefore, it is necessary to adopt software-assisted drawing review, such as using AI image recognition technology for automated review, which has become an industry trend.

[0037] However, the main technical challenge currently faced lies in the diversity and complexity of the drawings. For example, structural drawings are usually provided by different design institutes, and due to the lack of a unified drafting standard, each design institute and its designers will draw drawings in different formats according to their own habits.

[0038] Based on this, this application provides a beam component identification method. This method obtains a beam construction drawing sub-frame, searches for multiple sets of beam annotation information within the sub-frame, deletes these sets of annotation information from the sub-frame, obtains a target sub-frame, detects multiple beam regions within the target sub-frame, and matches these beam regions with the multiple sets of beam annotation information. Each set of beam annotation information is then assigned to its corresponding beam region. Compared to existing beam component identification methods, this method not only accurately identifies beam regions but also matches them with corresponding annotation information, significantly improving work efficiency and accuracy.

[0039] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0040] Please see Figure 1 , Figure 1 A flowchart illustrating a beam member identification method provided in an embodiment of this application is shown. Figure 1As shown, the beam component identification method can be executed by an electronic device, and the specific device of the electronic device can be configured according to actual needs; the embodiments of this application are not limited thereto. For example, the electronic device can be a computer or a server, etc. Specifically, the beam component identification method includes:

[0041] Step S110: Obtain the beam construction drawing sub-frame. The beam construction drawing sub-frame is extracted from the architectural drawing to be identified, and the beam construction drawing sub-frame includes multiple beam areas and multiple sets of beam annotation information. Each set of beam annotation information includes beam annotation content and a beam annotation line. The beam annotation content includes beam number information (e.g., it can be KL6(4) etc.), beam section information (e.g., it can be 200×450 etc.), and reinforcement information in the beam. The reinforcement information in the beam can include at least one type of reinforcement information, such as longitudinal reinforcement, stirrups, and web reinforcement.

[0042] It should be understood that the architectural drawings can be electronic drawings such as CAD drawings.

[0043] It should also be understood that the method for obtaining the sub-frame of the beam construction drawing can be set according to actual needs, and the embodiments of this application are not limited thereto.

[0044] Optionally, the sub-frame of the beam construction drawing can be obtained using the detection method and electronic equipment for the sub-frame in architectural drawings disclosed in application number 202311096933.0.

[0045] For example, the architectural drawings to be identified can first be converted from CAD drawings to PNG drawings. Then, the PNG drawings can be processed using the detection method of sub-frames in the architectural drawings to obtain the sub-frames of the beam construction drawings.

[0046] It should be noted that the sub-frame of this beam construction drawing includes beam annotation information and lines other than the beam annotation information, and these lines can include walls, slabs, and beams. Specifically, beams are supported by walls, and slabs are supported by both walls and beams.

[0047] Step S120: Locate multiple sets of beam annotation information in the beam construction drawing sub-frame.

[0048] It should be understood that the specific process of searching for multiple sets of beam annotation information from the beam construction drawing sub-frame can also be set according to actual needs, and the embodiments of this application are not limited to this.

[0049] Optionally, the text elements in the sub-frame of the beam construction drawing are classified to obtain text element classification results; wherein, the text element classification results include beam number information classification results, beam section information classification results, and beam reinforcement information classification results; the candidate beam annotation line determination step, based on the beam number information classification results, takes the line that is perpendicular to the direction of each beam number information and is closest to it as the corresponding candidate beam annotation line; based on the position of each candidate beam annotation line, the beam section information and beam reinforcement information belonging to the same group of beam annotation information as each candidate beam annotation line are determined.

[0050] Specifically, by analyzing the architectural drawings, it can be determined that there are various types of annotations, but the number is an essential piece of information for each annotation. Furthermore, the extension direction of each annotation line is perpendicular to the text direction where the beam number is located. Therefore, this application can use the strong correlation (or arrangement) between the beam annotation lines and the beam numbers as an identification feature.

[0051] Therefore, the text elements in the beam construction drawing sub-frame (e.g., beam number information, beam section information, and beam reinforcement information) can be classified first, thereby determining the classification result of each text element. For example, if the text element in the beam construction drawing sub-frame includes KL6(4), KL6(4) can be classified into the beam number information classification result; as another example, if the text element in the beam construction drawing sub-frame includes 200×450, 200×450 can be classified into the beam section information classification result.

[0052] Subsequently, based on the position of each beam number in the beam numbering information classification results (i.e., based on the strong correlation between beam label lines and beam numbers), the line perpendicular to the direction of each beam number and closest to it is determined as the corresponding candidate beam label line. For example, if the current beam number is recorded as horizontal, the line closest to that current beam number and with a vertical direction can be used as the candidate beam label line corresponding to the current beam number. Here, the candidate beam label line is defined based on the premise that it cannot be guaranteed to be the final correct beam label line at this point; the candidate beam label line corresponding to the current beam number refers to the label line used to label the beam label information of the group to which the current beam number belongs.

[0053] Finally, beam section information and beam reinforcement information that are on the same side of the candidate beam label line, near the corresponding beam number (e.g., the beam section can be adjacent to and in the same row as the beam number; the beam reinforcement information can be in the next row after the beam number; the beam reinforcement information can be in the row after the beam number, etc.), and whose direction is perpendicular to the direction of its corresponding candidate beam label line, are identified as beam label information belonging to the same group. It should be understood that this step can be performed after all the final beam label lines have been determined, thereby improving efficiency.

[0054] It's also important to note that after obtaining all candidate beam marking lines using the methods described above, these lines can be verified to ensure their accuracy. Specifically, all candidate beam marking lines are analyzed to determine if any are outliers. For example, it can be determined whether all beam marking lines share the same layer information and whether their line colors are identical. If a candidate beam marking line has an inconsistent line color and different layer information from other candidate beam marking lines, it can be considered an outlier.

[0055] If outlier candidate beam annotation lines exist, all non-outlier candidate beam annotation lines and outlier candidate beam annotation lines are deleted from the beam construction drawing sub-frame, resulting in an updated beam construction drawing sub-frame. The candidate beam annotation line determination step is then re-executed on the updated beam construction drawing sub-frame to determine the new beam annotation lines in the beam annotation information of the group to which the outlier candidate beam annotation lines belong. Furthermore, the new beam annotation lines can be used as the final beam annotation lines in the beam annotation information of the group to which the outlier beam annotation lines belong (for example, after determining the new beam annotation lines corresponding to the current text element classification result, they can be used as the final beam annotation lines corresponding to that current text element classification result).

[0056] If there are no outliers in the beam labeling information, each candidate beam labeling line can be used as the final beam labeling line for its corresponding group of beam labeling information.

[0057] Step S130: Delete multiple sets of beam annotation information from the beam construction drawing sub-frame to obtain the target sub-frame, and detect multiple beam regions from the target sub-frame. Each beam region refers to the area containing beam lines.

[0058] It should be understood that the target sub-frame is a frame that only contains lines of walls, beams, and slabs, and does not contain beam annotation lines or beam annotation content.

[0059] It should also be understood that the specific process of detecting multiple beam regions from the target sub-frame can be set according to actual needs, and the embodiments of this application are not limited thereto.

[0060] Optionally, the target sub-boundary can be input into a pre-trained beam region detection model to detect multiple beam regions. For example, the target sub-boundary can be input into a pre-trained beam region detection model to obtain multiple output results, and the output results can be processed by non-maximum suppression (NMS) to obtain multiple beam regions.

[0061] The beam region detection model includes a detection head with an added attention mechanism (NAM) module. This detection head may comprise n convolutional layers (conv) arranged in parallel, and each of these n convolutional layers is cascaded with a NAM module. Here, n is a positive integer greater than 3. For example, n equals 4.

[0062] To facilitate understanding of the model structure of the beam region detection model in this application, specific embodiments are described below.

[0063] Specifically, since the beam region is a narrow, closed region, in order to increase the discriminative power of the beam region detection model, the YOLOv5 model can be improved using the NAM attention module, and its model structure is as follows: Figure 2 As shown. The beam region detection model includes a backbone module, a neck module, and a detection head module.

[0064] Furthermore, the backbone part can be implemented using CSPDarkNet53, and the backbone part is mainly responsible for feature extraction from the input image; the neck part is implemented using the FPN+PAN structure, which is responsible for multi-scale feature fusion of the feature map and passing these features to the prediction layer; the head is the object detection head, and the detection head module is mainly responsible for multi-scale object detection on the feature map extracted by the backbone network.

[0065] Furthermore, the network input of this beam region detection model is the target sub-boundary, and features are extracted through the backbone. Unlike the original YOLOv5 model, this application adds a NAM module (for example, see [link to relevant documentation]). Figures 3 to 8 This is a normalized attention module that can suppress less significant weights. It applies a weight sparsity penalty to attention modules, thereby making them more computationally efficient while maintaining similar performance.

[0066] NAM is a lightweight and efficient attention mechanism that adopts a modular integration approach similar to CBAM. It redesigns the channel attention and spatial attention sub-modules, so that NAM can be embedded at the end of each network block.

[0067] Furthermore, in the Head stage, due to the diverse types and sizes of beams (for example, large ones can reach tens of thousands of meters in length, while small ones are only a few hundred meters long), compared to the existing YOLOV5 model's Head detection head with its three convolutional layers, this application adds a new convolutional layer to the Head detection head. Each of the four convolutional layers is connected in series with a NAM module, which also provides good detection results for smaller objects (e.g., 4×4).

[0068] It should be noted that although the above description uses a detection head with four convolutional layers arranged in parallel as an example, those skilled in the art should understand that the number of convolutional layers arranged in parallel can be set according to actual needs, and the embodiments of this application are not limited to this.

[0069] In addition, those skilled in the art can also Figure 2 The model structure of the beam region detection model shown may be modified or adjusted, but the embodiments of this application are not limited to this.

[0070] For example, for Figure 2 The beam region detection model shown may also include an input module, which is connected to the first Focus module.

[0071] Step S140: Match multiple beam regions and multiple sets of beam annotation information to assign each set of beam annotation information in the multiple sets of beam annotation information to the corresponding beam region.

[0072] Specifically, it is determined whether both ends of the current beam marking line are within the current beam region. If one end of the current beam marking line is within the current beam region, the beam region where the end of the current beam marking line is located can be determined as the beam region corresponding to the current beam marking line. The beam marking information belonging to the same group as the current beam marking line is assigned to the beam region corresponding to the current beam marking line, so that each beam region can determine its corresponding set of beam marking information.

[0073] Therefore, by utilizing the above technical solution, this embodiment of the application obtains a sub-frame of the beam construction drawing, searches for multiple sets of beam annotation information from the sub-frame, deletes multiple sets of beam annotation information from the sub-frame, obtains a target sub-frame, detects multiple beam regions from the target sub-frame, and matches multiple beam regions with multiple sets of beam annotation information to assign each set of beam annotation information to the corresponding beam region. Compared with existing beam component recognition methods, this not only achieves full automation of beam recognition in structural drawings but also does not rely on error-prone feature information such as layers, greatly improving work efficiency and accuracy.

[0074] It should be understood that the above-described beam component identification method is merely exemplary, and those skilled in the art can make various modifications based on the above method, and the modified solutions also fall within the protection scope of this application.

[0075] This application provides a storage medium storing a computer program, which is executed by a processor to perform the methods described in the embodiments.

[0076] This application also provides a computer program product that, when run on a computer, causes the computer to perform the method described in the method embodiment.

[0077] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0078] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0079] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0080] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0081] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0082] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A beam member identification method characterized by, The beam component recognition method is applied to an electronic device, and the beam component recognition method comprises: obtaining a beam construction drawing sub-frame; wherein the beam construction drawing sub-frame is extracted from a building drawing to be recognized, and the beam construction drawing sub-frame comprises a plurality of beam regions and a plurality of sets of beam annotation information; finding the plurality of sets of beam annotation information from the beam construction drawing sub-frame; deleting the plurality of sets of beam annotation information from the beam construction drawing sub-frame to obtain a target sub-frame, and detecting the plurality of beam regions from the target sub-frame; matching the plurality of beam regions and the plurality of sets of beam annotation information to assign each set of beam annotation information in the plurality of sets of beam annotation information to a corresponding beam region.

2. The beam member identification method according to claim 1, characterized by, Each set of beam annotation information comprises a beam annotation line and beam annotation content, and the beam annotation content comprises beam number information, beam cross-section information, and beam reinforcement information.

3. The beam member identification method according to claim 2, characterized by, The finding of the plurality of sets of beam annotation information from the beam construction drawing sub-frame comprises: classifying text graph elements in the beam construction drawing sub-frame to obtain a text graph element classification result; wherein the text graph element classification result comprises a beam number information classification result, a beam cross-section information classification result, and a beam reinforcement information classification result; a candidate beam annotation line determination step, based on the beam number information classification result, determining a line closest to and perpendicular to a direction in which each beam number information is located as a corresponding candidate beam annotation line; based on a position of each candidate beam annotation line, determining beam cross-section information and beam reinforcement information belonging to a same set of beam annotation information as each candidate beam annotation line.

4. The beam member identification method according to claim 3, characterized by, After determining all the candidate beam annotation lines, the beam component recognition method further comprises: analyzing the determination of all the candidate beam annotation lines to determine whether there is an outlier candidate beam annotation line among all the candidate beam annotation lines; if there is the outlier candidate beam annotation line, deleting the non-outlier candidate beam annotation line and the outlier candidate beam annotation line among all the candidate beam annotation lines from the beam construction drawing sub-frame, and re-executing the candidate beam annotation line determination step to determine a new beam annotation line in the set of beam annotation information to which the outlier candidate beam annotation line belongs; taking the new beam annotation line as a final beam annotation line in the set of beam annotation information to which the outlier candidate beam annotation line belongs.

5. The beam member identification method according to claim 4, characterized by, The outlier candidate beam annotation line is a beam annotation line with different layer information and different line color.

6. The beam member identification method according to claim 1, characterized by, The detection of the plurality of beam regions from the target sub-frame comprises: inputting the target sub-frame into a pre-trained beam region detection model for beam region detection to detect the plurality of beam regions; wherein the beam region detection model comprises a detection head added with a NAM module.

7. The beam member identification method according to claim 6, characterized by, The detection head comprises n convolution layers arranged in parallel, and each convolution layer in the n convolution layers is connected in series with a NAM module; wherein n is a positive integer greater than 3.

8. The beam member identification method according to claim 1, characterized by, The matching of the plurality of beam regions and the plurality of sets of beam annotation information to assign each set of beam annotation information in the plurality of sets of beam annotation information to a corresponding beam region comprises: The beam region where the end point of the current beam annotation line is located is determined as the beam region corresponding to the current beam annotation line, and the beam annotation information belonging to the same group as the current beam annotation line is assigned to the beam region corresponding to the current beam annotation line.

9. An electronic device, comprising: The electronic device comprises a processor, a memory and a bus, the memory stores machine readable instructions executable by the processor, when the electronic device is running, the processor and the memory communicate through the bus, and the machine readable instructions are executed by the processor to perform the beam member identification method of any one of claims 1-8.

Images