Blueprint building component automatic identification method, device and equipment and storage medium

Abstract

The application relates to a blueprint building component automatic identification method, device and equipment and a storage medium. The method comprises the following steps: acquiring a plan image, an elevation image and / or a section image of a building component to be identified, inputting the plan image, the elevation image and / or the section image as input; and automatically classifying and detecting an output to obtain a component type of the building component to be identified by using a preset classification detection model. Different characteristics of the building component to be identified in different perspectives of a plan, a section and an elevation are combined, the preset classification detection model is used to automatically identify the component type of the building component, more useful classification identification information can be obtained by the preset classification detection model, and therefore the classification identification accuracy of the building component type can be improved.

Description

Technical Field

[0001] This application relates to the field of drawing information recognition and processing, and in particular to methods, devices, equipment and storage media for automatic recognition of architectural components in drawings. Background Technology

[0002] In the field of engineering construction, architectural drawings are not used directly for construction after they are completed. They need to be submitted to relevant review departments for review to check whether the design drawings meet the key points of relevant building industry specifications or other relevant requirements. Only architectural drawings that pass the review will be used for production and construction.

[0003] Drawing review can be divided into manual review, automated computer review, or a combination of both. Compared to manual review, automated computer review is more efficient and can greatly reduce the workload of reviewers; therefore, automated review is currently the preferred method. Relevant review agencies and departments have a high market demand for efficient, accurate, and comprehensive automated review technology.

[0004] Drawing review primarily involves analyzing and identifying architectural components in architectural drawings to determine their compliance with architectural design codes and requirements. Different architectural components have different design codes or requirements; therefore, accurate detection and classification of architectural components in architectural drawings are crucial for accurate drawing review. Especially for architectural components with similar geometric shapes in floor plans, current automated drawing review technology is prone to errors and cannot accurately identify component types. For example, it may identify a connecting corridor as a window, or a window as a railing, resulting in inconsistencies between the parsed information and the intended meaning, thus affecting the accuracy of the drawing review. Summary of the Invention

[0005] To improve the current problem of poor accuracy in identifying architectural components on drawings, this application provides a method, apparatus, equipment, and storage medium for automatic identification of architectural components on drawings.

[0006] Firstly, the automatic identification method for architectural components in drawings provided in this application adopts the following technical solution:

[0007] A method for automatic identification of architectural components in drawings, comprising:

[0008] Acquire plan images, elevation images, and / or sectional images of the building component to be identified; the plan images include images of the building component to be identified as corresponding cropped images in the building's plan view; the elevation images include images of the building component to be identified as corresponding cropped images in the building's elevation view; the sectional images include images of the building component to be identified as corresponding cropped images in the building's sectional view.

[0009] The planar image, the elevation image, and / or the cross-sectional image are used as input; using a preset classification and detection model, the component type of the building component to be identified is automatically classified and detected.

[0010] By adopting the above technical solution, and by combining the different features displayed by the building components to be identified from different perspectives such as plan view, section view, and elevation view, the component type of the building components can be automatically identified by using a preset classification detection model. This allows the preset classification detection model to obtain more useful classification and identification information, thereby improving the accuracy of classification and identification of building component types.

[0011] Optionally, the preset classification and detection model includes a convolutional neural network, an encoder, a sequence mechanism module, and a decoder connected in sequence.

[0012] By adopting the above technical solution, firstly, a convolutional neural network is used to extract geometric shape, lines, and color features from the image of the building component to be identified. Then, the feature dimensions are enriched by an encoder. Next, a sequence mechanism module is used to integrate and sort the multi-dimensional feature information. Finally, a decoder is used to achieve similarity matching and classification recognition output, thereby improving the accuracy of building component type detection and recognition.

[0013] Optionally, the convolutional neural network consists of 4 to 10 convolutional kernels (conv) connected in series; the encoder consists of 3 to 5 variable convolutional kernels (DCN) connected in series; the sequence mechanism module includes one of recurrent neural networks (RNN), long short-term memory networks (LSTM), and transformer neural networks; the decoder consists of a convolutional layer (conv1*1), an activation layer (RELU), and a fully connected layer (FC) connected in series.

[0014] By adopting the above technical solution, the test results obtained based on measured data show that the accuracy of the preset classification detection model in detecting and identifying building components can reach 100%.

[0015] Optionally, acquiring the plan image, elevation image, and / or section image of the building component to be identified includes:

[0016] Obtain a planar image of the building component to be identified, and determine the position information of the planar image in the floor plan of the building to which it belongs;

[0017] In the building elevation drawing, a first drawing area corresponding to the location information is determined, and the first drawing area is cropped as the elevation image of the building component to be identified.

[0018] And / or, determine a second drawing area corresponding to the location information in the corresponding building section drawing, and extract the second drawing area as the section image of the building component to be identified.

[0019] By adopting the above technical solution, the elevation image and / or section image corresponding to the planar image position of the building component to be identified can be accurately mapped and obtained. Then, the elevation image and / or section image of the building component to be identified can be combined as the input of the preset classification and detection model to improve the accuracy of classification, detection and recognition.

[0020] Optionally, determining the location information of the planar image in the corresponding building plan includes:

[0021] Obtain the target axis number corresponding to the planar image in the building plan, and establish the axis grid coordinates of the planar image;

[0022] Obtain the floor information of the building's floor plan;

[0023] Based on the grid coordinates and the floor information, the position information of the plan image in the floor plan of the building to which it belongs is obtained.

[0024] By adopting the above technical solution, the location information of the building component to be identified can be uniquely and accurately determined, thereby enabling accurate extraction of images from building elevation drawings and building section drawings.

[0025] Optionally, obtaining the target axis number corresponding to the planar image in the building plan includes:

[0026] In the building plan, obtain two longitudinal axes adjacent to the left and right sides of the outline of the plan image, and two transverse axes adjacent to the top and bottom sides of the outline of the plan image; use the axis numbers of the transverse axes and the axis numbers of the longitudinal axes as the axis numbers of the target axis.

[0027] The process of obtaining the floor information of the building's floor plan includes:

[0028] The floor plan name of the building is extracted using OCR (optical character recognition), and the floor information is determined based on the floor plan name.

[0029] By adopting the above technical solution, the axis grid coordinates of the building component to be identified are established using the axis numbers of four adjacent axes. At the same time, the floor information is extracted by OCR to obtain the relative elevation of the building component to be identified. Thus, the elevation and section images of the building component to be identified can be accurately mapped and extracted from the elevation and section views.

[0030] Optionally, determining the first drawing area corresponding to the location information in the building elevation drawing includes:

[0031] In the building elevation drawing, a first axis line associated with the grid coordinates is determined;

[0032] Determine the target floor based on the floor information;

[0033] The drawing area defined by the first axis is determined within the target floor and designated as the first drawing area.

[0034] Determining the second drawing area corresponding to the location information in the building section view includes:

[0035] In the architectural cross-section drawing, a second axis line is determined that is associated with the grid coordinates.

[0036] Determine the target floor based on the floor information;

[0037] The drawing area defined by the second axis within the target floor is determined as the second drawing area.

[0038] By adopting the above technical solution, the mapping and extraction of facade and section images of the building components to be identified were realized.

[0039] Secondly, the automatic identification device for architectural components in drawings provided in this application adopts the following technical solution:

[0040] An automatic identification device for architectural components on blueprints, comprising:

[0041] The acquisition module is used to acquire plan images, elevation images, and / or sectional images of the building component to be identified; the plan images include images of the building component to be identified as corresponding cropped images in the building's plan view; the elevation images include images of the building component to be identified as corresponding cropped images in the building's elevation view; and the sectional images include images of the building component to be identified as corresponding cropped images in the building's sectional view.

[0042] A preset classification and detection model is used to automatically classify and detect the component type of the building component to be identified, using the planar image, the elevation image and / or the cross-sectional image as input.

[0043] Thirdly, the automatic identification device for architectural components in drawings provided in this application adopts the following technical solution:

[0044] An automatic identification device for architectural components on drawings includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aforementioned automatic identification method for architectural components on drawings.

[0045] Fourthly, the computer-readable storage medium provided in this application adopts the following technical solution:

[0046] A computer-readable storage medium storing a computer program; when executed by a processor, the computer program implements the above-described method for automatically identifying architectural components in drawings.

[0047] By adopting the above technical solution, a computer program carrier for an automatic identification method of architectural components in drawings is provided.

[0048] In summary, this application includes at least the following beneficial technical effects:

[0049] 1. By combining the different features displayed by the building components to be identified from different perspectives such as plan view, section view, and elevation view, the preset classification detection model can obtain more useful classification and identification information, thereby improving the accuracy of classification and identification of building component types.

[0050] 2. The pre-defined classification and detection model first uses a convolutional neural network to extract geometric shape, lines, and color features from the images of the building components to be identified. The encoder enriches the feature dimensions, and then the sequence mechanism module integrates and sorts the multi-dimensional feature information. Finally, the decoder achieves similarity matching and classification recognition output, thereby improving the accuracy of building component type detection and recognition. Attached Figure Description

[0051] Figure 1 This is a flowchart of an automatic identification method for architectural components in drawings according to an embodiment of this application;

[0052] Figure 2 This is a schematic planar image of the building components railings and windows in the embodiments of this application;

[0053] Figure 3 This is a schematic diagram of the elevation images of building components such as railings and windows in the embodiments of this application;

[0054] Figure 4 This is a flowchart of the facade image extraction method in the embodiments of this application;

[0055] Figure 5 This is a schematic diagram of the axis grid in an architectural plan according to an embodiment of this application;

[0056] Figure 6This is a schematic diagram of elevation in a building elevation drawing according to an embodiment of this application;

[0057] Figure 7 This is a structural block diagram of a preset classification and detection model in an embodiment of this application;

[0058] Figure 8 This is a structural block diagram of an automatic identification device for architectural components on drawings, as described in an embodiment of this application.

[0059] Figure 9 This is a structural block diagram of an automatic identification device for architectural components on drawings, as described in an embodiment of this application. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0061] This application discloses an automatic identification method for architectural components in drawings.

[0062] refer to Figure 1 A method for automatic identification of architectural components on drawings, comprising the following steps:

[0063] S101. Obtain a plan view, elevation view and / or section view of the building component to be identified; wherein the plan view includes the image of the building component to be identified as a corresponding crop in the building plan; the elevation view includes the image of the building component to be identified as a corresponding crop in the building elevation; and the section view includes the image of the building component to be identified as a corresponding crop in the building section.

[0064] Building components can refer to the basic building units or elements used to form a complete building. For example, walls, columns, piers, elevators, stairs, balconies, canopies, steps, ramps, flues, ventilation ducts, pipe shafts, fire escape ladders, rainwater pipes, drainage ditches, flower beds, pits, trenches, reserved holes in walls, doors, windows, railings, etc., in engineering drawings can all be called building components.

[0065] In engineering drawings, different architectural components are represented by different graphic elements. However, some architectural components have very similar geometric shapes when represented by graphic elements on a floor plan. This makes it difficult for computers to accurately detect and classify architectural components in floor plans during the automatic drawing review process. In the case of incorrect identification of architectural component types, errors in the drawing review conclusions are inevitable.

[0066] For example, building components such as rainwater pipes, risers, and air conditioning condensate pipes can all be represented by circles "⊕"; similarly, building components such as railings and windows can be represented by rectangles. (See reference...) Figure 2 Therefore, it is easy for detection and classification errors to occur. To address this, embodiments of this application combine images of the building component to be identified from different perspectives, referencing... Figure 3 ,in Figure 3 (a) represents the graphic image of the railing on the elevation view. Figure 3 (b) represents the primitive image of the window on the elevation view, which enables the preset classification detection model to obtain more dimensions of information that helps in detection and classification, thereby improving the accuracy of detection and classification.

[0067] The planar image of the building component to be identified refers to the image cropped from the corresponding position in the building plan. The cropped image must contain graphic information of the building component to be identified, including but not limited to the geometric shape, color, line thickness / type, etc. of the building component to be identified.

[0068] It should be understood that an architectural floor plan refers to a horizontal sectional view obtained by cutting a building along a horizontal plane slightly above the windowsill, removing the upper part, and projecting the remaining part onto a horizontal plane. An architectural floor plan reflects the plan shape of a new building, the location, size, and interrelationship of rooms, the location, thickness, and material of walls, the cross-sectional shape and size of columns, and the location and type of doors and windows. Therefore, an architectural floor plan contains several building components. During the review process, each building component can be identified as a component to be identified. The component type is then compared with the corresponding review standards to determine whether it conforms to the building code for that component, thus achieving an automated review process.

[0069] Regarding the acquisition method of the planar images of the building components to be identified, existing related technologies can be used to automatically extract the building components from the building plan and treat them as building components to be identified. For example, an algorithm model consisting of object detection models (YOLO series, SSD network, CenterNet network, YOLOX network), semantic segmentation models (deeplabv3 series, Unet network, Segnet network, etc.), and instance segmentation models (Mask R-CNN network, yolact network, SOLO series) can be used to extract the planar images of building components.

[0070] The planar image of the building component to be identified can be cropped using the smallest rectangular frame, or it can be cropped using other selection methods.

[0071] After obtaining the planar image of the building component to be identified, although it is possible to identify the component type based solely on the planar image, for example, the current deep learning algorithm is mainly used for training and identification. However, due to the limited dimension of the abstract features learned by the model, and the fact that some building components have similar geometric shapes of primitives shown in the planar image, the deep learning algorithm currently used is prone to classification and identification errors.

[0072] In response, this application embodiment provides more effective information that helps to better represent the differences in feature characteristics of building components with similar geometric shapes, thereby better achieving the classification, detection, and recognition of building components.

[0073] Specifically, this application embodiment will also acquire elevation images of the building components to be identified. These elevation images can better display the differences in the building components' characteristics along the building's height. It should be understood that a building elevation drawing refers to an orthographic projection drawn on a vertical projection plane parallel to the building facade, or simply an elevation drawing. Building elevation drawings can reflect the height, appearance, and decoration requirements of various parts of the building.

[0074] How can we obtain the facade image of the building component to be identified? Typically, existing technologies can be used to automatically extract building components from a building facade drawing, and the facade image of the corresponding building component can be obtained by cropping the image of that building component from the facade drawing. However, building facade drawings contain a large number of building components, so finding the accurate mapping of the building component to be identified in the facade drawing is crucial. This application provides a highly efficient and accurate method that requires minimal computing power to accurately extract the facade image corresponding to the building component to be identified.

[0075] For details, please refer to Figure 4 It mainly includes the following steps:

[0076] S401. Obtain the target axis number of the plan image of the building component to be identified in the building plan, and establish the axis grid coordinates of the plan image of the building component to be identified.

[0077] First, determine the outline of the planar image of the building component to be identified, obtain two longitudinal axes adjacent to the left and right sides of the outline of the planar image, and two transverse axes adjacent to the top and bottom sides of the planar image, and determine the axis numbers of the four transverse axes and the longitudinal axes as the target axis numbers.

[0078] It should be understood that "adjacent" includes both cases where they are exactly touching (i.e., in contact) and cases where they are not touching, see reference. Figure 5The left adjacent axis of building component α is longitudinal axis 1 (not in contact), the right adjacent axis is longitudinal axis 2 (not in contact), the upper adjacent axis is transverse axis F (in contact), and the lower adjacent axis is transverse axis E (not in contact).

[0079] In an optional embodiment of this application, the axis numbers of the four horizontal axes and the vertical axes can be combined sequentially in a set order to serve as the grid coordinates of the planar image of the building component to be identified.

[0080] For example, the axis numbers of the two horizontal axes are E and F, and the axis numbers of the two vertical axes are 1 and 2. The axis numbers of the four horizontal axes and the vertical axes are sorted in order from bottom to top and from left to right, respectively, so that the axis grid coordinates of the planar image of the building component to be identified are (E, F, 1, 2).

[0081] In an optional embodiment of this application, the actual distances between the top, bottom, left, and right edges of the building component to be identified and the four adjacent axes can be calculated first, and then the four actual distances can be combined in a set order to serve as the grid coordinates of the planar image of the building component to be identified.

[0082] For example, the distance between building component α and axis E is L. αE The distance from the axis F is L. αF The distance from axis 1 is L. α1 The distance from axis 2 is L. α2 Following the order of the horizontal axis from bottom to top and the vertical axis from left to right, the grid coordinates of the planar image of the building component to be identified are obtained as (L... αE ,L αF ,L α1 ,L α2 ).

[0083] It should be understood that the distance between the planar image of the building component to be identified and the adjacent axis can be obtained by any existing method, such as by calculating it based on the distance and position information marked in the drawing. The specific process will not be elaborated here.

[0084] S402. Obtain the floor information of the building's floor plan. That is, obtain the floor information of the building component to be identified.

[0085] In an optional embodiment of this application, the drawing name of the building floor plan can be extracted using OCR, and the floor information can be determined based on the drawing name. Typically, the number of floor plans corresponds to the number of floors in a building, with the corresponding drawing name noted below each plan. Since the structure and layout of intermediate floors in multi-story (high-rise) buildings are basically the same, only one floor plan is needed for a standard floor.

[0086] For example, the drawing title may be "First Floor Plan", "Ground Floor Plan", "Second Floor Plan", "Standard Floor Plan", "Top Floor Plan", etc. Floor information can be obtained through the drawing title.

[0087] Based on the grid coordinates and floor information, the accurate location information of the planar image can be obtained.

[0088] S403. Determine the first axis line associated with the grid coordinates of the building component to be identified in the building elevation drawing.

[0089] It should be noted that the first axis associated with the grid coordinates refers to the axis with the same axis number as the grid coordinates in the architectural elevation drawing. For example, the grid coordinates of the building component to be identified are (E,F,1,2) or (L... αE ,L αF ,L α1 ,L α2 In architectural elevation drawings, there are usually axes numbered E and F, or axes numbered 1 and 2 (the axis numbers vary depending on the elevation orientation). Axes E and F are the first associated axes, or axes 1 and 2 are the first associated axes. It should be understood that elevation drawings may omit the middle axis numbers, retaining only the axis numbers at the ends, such as only retaining F and A, but the axis numbers of the middle axes can be labeled sequentially.

[0090] S404. Determine the target floor based on the floor information.

[0091] refer to Figure 6 For example, if the drawing is named "Ground Floor Plan", then the target floor can be determined as "Floor 1", which corresponds to the floor with an elevation of ±0.000 to 3.000 meters in the architectural drawing. For another example, if the drawing is named "Top Floor Plan", then the target floor can be determined as "Floor 3", which corresponds to the floor with an elevation of 6.000 to 9.000 meters in the architectural drawing.

[0092] S405. Determine the drawing area bounded by the first axis in the target floor as the first drawing area.

[0093] By limiting the elevation using floor information and restricting the elevation position using two primary axes, a closed drawing area can be obtained. By extracting this closed drawing area, the elevation image of the building component to be identified can be obtained.

[0094] S406. In the first drawing area, extract the first drawing area as the elevation image of the building component to be identified.

[0095] Based on the above technical solution, accurate and efficient mapping and extraction of facade images of building components to be identified were achieved.

[0096] In an optional embodiment of this application, when an elevation image of the building component to be identified is not available, a cross-sectional image of the building component can be obtained. Specifically, a second axis line associated with the grid coordinates is determined in the building cross-sectional drawing; a target floor is determined based on the floor information; a drawing area defined by the second axis line is determined within the target floor as a second drawing area; and the second drawing area is cropped as the cross-sectional image of the building component to be identified. The process of obtaining the cross-sectional image is the same as that of obtaining the elevation image, and will not be described again here.

[0097] It should be understood that a sectional view obtained by cutting a building with one or more vertical cutting planes is called an architectural sectional view, or simply a section view. Architectural sectional views are used to show the building's internal structure, vertical layering, floor levels, roof construction, and related dimensions and elevations. The cutting locations are often stairwells, door and window openings, and other structurally complex areas. The number of sectional views depends on the complexity of the building and the actual construction needs. The name of the sectional view corresponds to the cutting location and viewing direction marked on the ground floor plan.

[0098] It should be understood that when the building component to be identified has corresponding elevation and section images, both elevation and section images can be obtained simultaneously; when only one exists, one of them can be obtained, such as an elevation image or a section image.

[0099] Furthermore, this application embodiment also improves the deep learning model, enabling the improved deep learning model to better extract abstract features of building components, better reflect the feature differences between similar building components, and improve classification and detection accuracy. See the relevant description in step S102.

[0100] S102. Take the planar image, as well as the elevation image and / or section image, as input; use the preset classification and detection model to automatically classify and detect the component type of the building component to be identified.

[0101] refer to Figure 7 The preset classification and detection model includes a convolutional neural network 71, an encoder 72, a sequence mechanism module 73, and a decoder 74 connected in sequence.

[0102] The convolutional neural network 71 is composed of 4 to 10 convolutional kernels conv connected in series; the convolutional neural network 71 is used to extract geometric shape, lines and color features of the building component image to be identified.

[0103] The encoder 72 is composed of 3 to 5 variable convolutional kernels (DCN) connected in series; the encoder 72 enriches the feature dimensions.

[0104] The sequence mechanism module 73 includes one of the following: recurrent neural network (RNN), long short-term memory network (LSTM), and transformer neural network; the sequence mechanism module 73 is used to integrate and sort multidimensional feature information.

[0105] Decoder 74 consists of a convolutional layer (conv1*1), an activation layer (ReLU), and a fully connected layer (FC) connected in series. The ReLU activation layer can use one of the following activation functions: ReLU, Leaky ReLU, Randomized LeakyReLU, PReLU, softmax, etc. Decoder 74 achieves similarity matching and classification recognition output. Test results based on actual measurement data show that the preset classification detection model can achieve 100% accuracy in detecting and recognizing building components.

[0106] Based on the same design concept, this embodiment also discloses an automatic identification device for architectural components in drawings.

[0107] refer to Figure 8 An automatic identification device for architectural components on blueprints, mainly comprising:

[0108] The acquisition module 81 is used to acquire plan images, elevation images and / or section images of the building components to be identified; the plan images include the corresponding cropped images of the building components to be identified in the building's plan view; the elevation images include the corresponding cropped images of the building components to be identified in the building's elevation view; and the section images include the corresponding cropped images of the building components to be identified in the building's section view.

[0109] The preset classification and detection model 82 is used to automatically classify and detect the component type of the building component to be identified by taking planar images, as well as elevation images and / or sectional images as input.

[0110] The various variations and specific examples of the methods provided in the above embodiments are also applicable to the automatic identification device for architectural components in drawings in this embodiment. Through the foregoing detailed description of the automatic identification method for architectural components in drawings, those skilled in the art can clearly understand the implementation method of the automatic identification device for architectural components in drawings in this embodiment. For the sake of brevity, it will not be described in detail here.

[0111] To better implement the above method, embodiments of this application also provide an automatic identification device for architectural components on drawings, such as... Figure 9 As shown, the automatic identification device for architectural components in drawings includes a processor 91 and a memory 92.

[0112] Automatic identification equipment for architectural components in blueprints can be implemented in various forms, including mobile phones, tablets, handheld computers, laptops, desktop computers, servers, and other devices.

[0113] The memory 92 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 92 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as algorithm instructions for acquiring floor plans, elevations, and sections of a building component to be identified), and instructions for implementing the automatic identification method for building components based on drawings provided in the above embodiments. The data storage area may store data involved in the automatic identification method for building components based on drawings provided in the above embodiments.

[0114] Processor 91 may include one or more processing cores. Processor 91 executes instructions, programs, code sets, or instruction sets stored in memory 92, calls data stored in memory 92, and performs various functions and processes data as described in this application. Processor 91 may be at least one of the following: Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), Central Processing Unit (CPU), controller, microcontroller, and microprocessor. It is understood that for different devices, the electronic devices used to implement the above-described processor functions may also be other types, and this application embodiment does not specifically limit the specific implementation.

[0115] This application provides a computer-readable storage medium, including, for example, various media capable of storing program code such as a USB flash drive, portable hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk. This computer-readable storage medium stores a computer program that can be loaded by a processor and executed using the automatic identification method for architectural components based on drawings described in the above embodiments.

[0116] The above description of the embodiments is only used to provide a detailed introduction to the technical solutions of this application. However, the description of the above embodiments is only for the purpose of helping to understand the methods and core ideas of this application, and should not be construed as a limitation of this application. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the protection scope of this application.

Claims

1. A method for automatic identification of architectural components on drawings, characterized in that, The automatic identification method for architectural components in the drawings includes: Acquire plan images, elevation images, and / or sectional images of the building component to be identified; the plan images include images of the building component to be identified as corresponding cropped images in the building's plan view; the elevation images include images of the building component to be identified as corresponding cropped images in the building's elevation view; the sectional images include images of the building component to be identified as corresponding cropped images in the building's sectional view. The plan image, the elevation image and / or the section image are used as input; using a preset classification and detection model, the component type of the building component to be identified is automatically classified and detected. The acquisition of plan images, elevation images, and / or sectional images of the building components to be identified includes: Obtain a planar image of the building component to be identified, and determine the position information of the planar image in the floor plan of the building to which it belongs; In the building elevation drawing, a first drawing area corresponding to the location information is determined, and the first drawing area is cropped as the elevation image of the building component to be identified. And / or, determine a second drawing area corresponding to the location information in the corresponding building section drawing, and extract the second drawing area as the section image of the building component to be identified; The method of determining the location information of the planar image in the corresponding building plan includes: Obtain the target axis number corresponding to the planar image in the building plan, and establish the axis grid coordinates of the planar image; Obtain the floor information of the building's floor plan; Based on the grid coordinates and the floor information, the position information of the plan image in the floor plan of the building to which it belongs is obtained; The step of obtaining the target axis number corresponding to the planar image in the building plan includes: In the building plan, obtain two longitudinal axes adjacent to the left and right sides of the outline of the plan image, and two transverse axes adjacent to the top and bottom sides of the outline of the plan image; use the axis numbers of the transverse axes and the axis numbers of the longitudinal axes as the axis numbers of the target axis. The process of obtaining the floor information of the building's floor plan includes: The name of the building floor plan is extracted by OCR recognition, and the floor information is determined based on the name of the floor plan. Determining the first drawing area corresponding to the location information in the building elevation drawing includes: In the building elevation drawing, a first axis line associated with the grid coordinates is determined; Determine the target floor based on the floor information; The drawing area defined by the first axis is determined within the target floor and designated as the first drawing area. Determining the second drawing area corresponding to the location information in the building section view includes: In the architectural cross-section drawing, a second axis line is determined that is associated with the grid coordinates. Determine the target floor based on the floor information; The drawing area defined by the second axis within the target floor is determined as the second drawing area.

2. The method for automatic identification of architectural components in drawings according to claim 1, characterized in that, The preset classification and detection model includes a convolutional neural network, an encoder, a sequence mechanism module, and a decoder connected in sequence.

3. The automatic identification method for architectural components in drawings according to claim 2, characterized in that, The convolutional neural network consists of 4 to 10 convolutional kernels (conv) connected in series; the encoder consists of 3 to 5 variable convolutional kernels (DCN) connected in series; the sequence mechanism module includes one of recurrent neural network (RNN), long short-term memory network (LSTM), and transformer neural network; the decoder consists of a convolutional layer (conv1*1), an activation layer (RELU), and a fully connected layer (FC) connected in series.

4. An automatic identification device for architectural components on blueprints, characterized in that, include: The acquisition module is used to acquire plan images, elevation images and / or sectional images of the building components to be identified; The plan image includes the image of the building component to be identified as a corresponding cropped image in the building's plan view; the elevation image includes the image of the building component to be identified as a corresponding cropped image in the building's elevation view; the section image includes the image of the building component to be identified as a corresponding cropped image in the building's section view. The acquisition of plan images, elevation images, and / or sectional images of the building components to be identified includes: Obtain a planar image of the building component to be identified, and determine the position information of the planar image in the floor plan of the building to which it belongs; In the building elevation drawing, a first drawing area corresponding to the location information is determined, and the first drawing area is cropped as the elevation image of the building component to be identified. And / or, determine a second drawing area corresponding to the location information in the corresponding building section drawing, and extract the second drawing area as the section image of the building component to be identified; The method of determining the location information of the planar image in the corresponding building plan includes: Obtain the target axis number corresponding to the planar image in the building plan, and establish the axis grid coordinates of the planar image; Obtain the floor information of the building's floor plan; Based on the grid coordinates and the floor information, the position information of the plan image in the floor plan of the building to which it belongs is obtained; The step of obtaining the target axis number corresponding to the planar image in the building plan includes: In the building plan, obtain two longitudinal axes adjacent to the left and right sides of the outline of the plan image, and two transverse axes adjacent to the top and bottom sides of the outline of the plan image; use the axis numbers of the transverse axes and the axis numbers of the longitudinal axes as the axis numbers of the target axis. The process of obtaining the floor information of the building's floor plan includes: The name of the building floor plan is extracted by OCR recognition, and the floor information is determined based on the name of the floor plan. Determining the first drawing area corresponding to the location information in the building elevation drawing includes: In the building elevation drawing, a first axis line associated with the grid coordinates is determined; Determine the target floor based on the floor information; The drawing area defined by the first axis is determined within the target floor and designated as the first drawing area. Determining the second drawing area corresponding to the location information in the building section view includes: In the architectural cross-section drawing, a second axis line is determined that is associated with the grid coordinates. Determine the target floor based on the floor information; The drawing area defined by the second axis within the target floor is determined as the second drawing area; A preset classification and detection model is used to automatically classify and detect the component type of the building component to be identified, using the planar image, the elevation image and / or the cross-sectional image as input.

5. An automatic identification device for architectural components on blueprints, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the automatic identification method for architectural components in drawings as described in any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program; when the computer program is executed by a processor, it implements the automatic identification method for architectural components in drawings as described in any one of claims 1 to 3.

Images