Information processing device, information processing method, and program

Abstract

The information processing device comprises at least one processor. The processor receives a design drawing as input, analyzes the design drawing to estimate a three-dimensional model of the structure shown in the design drawing, and calculates at least one of the volume and surface area of ​​the structure from the estimated three-dimensional model.

Description

[Technical Field]

[0001] The disclosed technology relates to information processing devices, information processing methods, and programs. [Background technology]

[0002] The following technologies are known to support the calculation of the amount of regulated substances used or emitted: Japanese Patent Publication No. 2025-043220 describes a method of receiving image data of invoice information from a business terminal, analyzing the image data using machine learning to extract information on energy consumption, and calculating greenhouse gas emissions based on the information on energy consumption.

[0003] European Patent Application Publication No. 4209975 describes a method for estimating the carbon footprint of a vehicle, comprising the steps of constructing a carbon footprint model of the vehicle, dividing the carbon footprint model into a first module and a second module, obtaining the carbon footprint of the first module from a database, and estimating the carbon footprint of the second module based on its dimensional parameters and materials used. [Overview of the project] [Problems that the invention aims to solve]

[0004] Regulations imposed on products may include requirements that the amount of regulated substances used or emitted be below a specified level, or that the amount of regulated substances used or emitted be declared. There are usually three main methods for calculating the amount of regulated substances used or emitted: (1) direct measurement through inspections, etc., (2) calculation using design data, etc., and (3) estimation using a simplified method. Accurately calculating the amount of regulated substances used or emitted is desirable for achieving the objectives of the regulation and for the regulated entity, and it goes without saying that choosing method (1) is the best option. However, for products whose production has already ended, direct measurement through inspections, etc., is not practical, and one is forced to choose method (2) or (3).

[0005] Even for products whose production has ended, if detailed 3D-CAD (3D-Computer-Aided Design) data is available, it is possible to check whether regulated substances were used by reading the structured data held within the 3D-CAD. Furthermore, it is possible to accurately calculate the volume and surface area of ​​the product from the topological and geometric data within the 3D-CAD, and to accurately calculate the amount of regulated substances used or emitted based on the volume and surface area. For example, if the surface area of ​​a product can be estimated, it becomes possible to estimate the amount of regulated substances used in surface treatment.

[0006] For example, when calculating Scope 3 greenhouse gas emissions, there is a significant difference in the calculation results depending on whether method (2) or method (3) is used. In such areas, the benefits of adopting method (2) are particularly great. More specifically, by using the weight that can be calculated from the volume and density of the product estimated from the design data, it is possible to calculate greenhouse gas emissions during transportation and delivery with high accuracy.

[0007] However, 3D-CAD data is not available for all products, and often the available materials are in the form of paper documents or image data such as PDFs. Due to the nature of this information, it is not machine-readable, so the desired information cannot be obtained by reading the image data on a computer.

[0008] On the other hand, the method of manually reading shapes and dimensions from numerous drawings and calculating volume and surface area has the problem of being overly burdensome to operate. Therefore, when production has ended and 3D-CAD data no longer exists, the simplified method in (3) must be chosen. Efficiently calculating approximate values ​​of the volume and surface area of ​​structures illustrated in drawings is considered useful not only for the purpose of calculating the amount of regulated substances used or emitted, but also for other purposes.

[0009] The disclosed technology was developed in view of the above points and aims to enable the efficient calculation of approximate values ​​for at least one of volume and surface area for structures for which 3D-CAD data does not exist. [Means for solving the problem]

[0010] The information processing device relating to the disclosed technology comprises at least one processor. The processor receives input of a design drawing, analyzes the design drawing to estimate a three-dimensional model of a structure illustrated in the design drawing, and calculates at least one of the volume and surface area of ​​the structure from the estimated three-dimensional model.

[0011] The processor may recognize strings of characters included in the design drawing, recognize the shape elements constituting the structure, and estimate a three-dimensional model of the structure by associating strings of characters indicating dimensional values ​​with the recognized shape elements.

[0012] The information processing device may have a storage medium that stores a parametric model for each of a plurality of shape elements, which associates a three-dimensional model of the shape element with a calculation formula for calculating at least one of the volume and surface area of ​​the shape element. The processor may select from the parametric models stored in the storage medium that correspond to the shape elements constituting the structure, and calculate at least one of the volume and surface area of ​​the structure based on the calculated value obtained by substituting the dimensional value of the shape element constituting the structure into the calculation formula in the selected parametric model.

[0013] If the processor recognizes a plurality of shape elements constituting the structure, it may calculate at least one of the volume and surface area of ​​the structure by integrating the calculated values ​​for each shape element obtained by substituting the dimensional value of the shape element corresponding to the calculation formula in the selected parametric model for each of the recognized shape elements.

[0014] The processor may output a confidence score indicating the confidence level for at least one of the volume and surface area calculated for the structure.

[0015] The processor may display on a display the calculated value of at least one of the volume and surface area of ​​the structure, and the confidence score.

[0016] The processor may perform the process of recognizing the shape elements constituting the structure and the process of associating the recognized shape elements with the strings indicating the dimension values ​​among the recognized strings, using a recognition model constructed by machine learning, and may perform machine learning of the recognition model using data based on CAD data showing the three-dimensional shape of the structure as training data.

[0017] The information processing method according to the disclosed technology receives an input of a design drawing, estimates a three-dimensional model of a structure illustrated in the design drawing by analyzing the design drawing, and calculates at least one of the volume and surface area of the structure from the estimated three-dimensional model. This is an information processing method executed by at least one processor included in an information processing device.

[0018] The program according to the disclosed technology is a program for causing at least one processor included in an information processing device to execute a process of receiving an input of a design drawing, estimating a three-dimensional model of a structure illustrated in the design drawing by analyzing the design drawing, and calculating at least one of the volume and surface area of the structure from the estimated three-dimensional model.

Advantages of the Invention

[0019] For a structure without 3D-CAD data, it becomes possible to efficiently calculate an approximate value of at least one of the volume and surface area.

Brief Description of the Drawings

[0020] [Figure 1] FIG. 1 is a diagram showing an example of the configuration of an information processing system according to an embodiment of the disclosed technology. [Figure 2] FIG. 2 is a diagram showing an example of a design drawing to be analyzed by an information processing device according to an embodiment of the disclosed technology. [Figure 3] FIG. 3 is a diagram showing an example of the hardware configuration of an information processing device according to an embodiment of the disclosed technology. [Figure 4] FIG. 4 is a functional block diagram showing an example of the functional configuration of an information processing device according to an embodiment of the disclosed technology. [Figure 5] FIG. 5 is a diagram showing the concept of a parametric model according to an embodiment of the disclosed technology. [Figure 6] FIG. 6 is a diagram conceptually showing an example of the input and output of a shape element recognition model according to an embodiment of the disclosed technology. [Figure 7] FIG. 7 is a diagram conceptually showing an example of the input and output of a dimension recognition model according to an embodiment of the disclosed technology. [Figure 8] This figure shows an example of how analysis results are displayed by an information processing device according to an embodiment of the disclosed technology. [Figure 9] This flowchart shows an example of the processing flow performed by the CPU of an information processing device according to an embodiment of the disclosed technology when it executes an analysis program. [Figure 10] This figure shows an example of a machine learning method for improving the recognition accuracy of a shape element recognition model and a dimension recognition model according to an embodiment of the disclosed technology. [Modes for carrying out the invention]

[0021] Hereinafter, an example of an embodiment of the disclosed technology will be described with reference to the drawings. In each drawing, identical or equivalent components and parts will be given the same reference numerals, and redundant descriptions will be omitted.

[0022] Figure 1 shows an example of the configuration of an information processing system 1 according to an embodiment of the disclosed technology. The information processing system 1 includes an information processing device 10 and a plurality of terminal devices 20. Each of the information processing device 10 and the plurality of terminal devices 20 is connected to communicate via a network 30 such as the Internet.

[0023] The terminal device 20 can access the information processing device 10, for example, through a website or a dedicated application. The terminal device 20 can send a request for analysis of design drawings to the information processing device 10 along with the design drawings to be analyzed.

[0024] The information processing device 10 has a function to analyze design drawings in response to requests from the terminal device 20. Specifically, the information processing device 10 calculates the volume and surface area of ​​the structure shown in the design drawing by analyzing the design drawing. In the following description, an example is given in which the information processing device 10 calculates both the volume and area of ​​the structure shown in the design drawing, but it may also calculate only one of the volume or area.

[0025] Figure 2 shows an example of a design drawing 200 that is the target of analysis by the information processing device 10. Typically, the design drawing 200 is a two-dimensional drawing of a structure such as a product or part, and is a complex document consisting of multiple elements. The structure 201 is illustrated in a certain manner in the design drawing 200. Figure 2 shows an example of a drawing (front view, top view, side view) of a structure 201 that combines a rectangular prism and a cylinder, illustrated using the third-angle projection method. In addition to the outline of the structure 201, the design drawing 200 may include dimensional values, dimension lines, and extension lines. Furthermore, the design drawing 200 may include various graphic symbols such as welding symbols indicating the processing method and geometric tolerance symbols indicating the specifications, as well as strings of characters associated with these symbols. In addition, the design drawing 200 may include a title block 202 that describes the name, material, and surface treatment method of the structure 201, as well as various tables such as a parts list and a dimension table.

[0026] Image data of design drawings 200 having file formats such as JPEG, PNG, GIF, or PDF is processed by the information processing device 10. Due to their nature, the contents of image data having the above file formats have low machine readability, and therefore, even if the image data is read by a computer, the desired information cannot be obtained. In other words, CAD data is data that has abundant machine-readable structured data so that it can be easily processed by a computer, and it is possible to read the accurate three-dimensional shape and dimensions of the model from the data. In contrast, image data having file formats such as PDF or JPEG is created for the purpose of visual interpretation by humans, and elements such as text, dimensions, lines, and shapes in the drawing are visual information and have low machine readability. In other words, image data consisting of visual information does not contain information that represents the logical relationships between elements such as lines, text, and dimensions in the drawing or the three-dimensional structure. Even if the above individual elements can be recognized by character recognition or image recognition, the information that represents the logical relationships between these elements or the three-dimensional structure is not included in the image data, and therefore, a general computer cannot recognize and interpret the three-dimensional shape and dimensions of a structure expressed as visual information from the image data. The image data of the design drawing 200, which is the target of analysis by the information processing device 10 according to this embodiment, is also not structured, machine-readable data like CAD data. Generally, a computer cannot recognize and interpret the three-dimensional shape and dimensions of the structure illustrated in the design drawing 200.

[0027] Figure 3 shows an example of the hardware configuration of the information processing device 10. The information processing device 10 is a computer and includes a CPU (Central Processing Unit) 101, RAM (Random Access Memory) 102, non-volatile memory 103, input / output interface (I / F) 104, network interface (I / F) 105, and read / write unit 106. These hardware components are interconnected via a bus 107.

[0028] The non-volatile memory 103 is a non-volatile storage medium such as a hard disk or flash memory. The non-volatile memory 103 stores the analysis program 110, the shape element recognition model 111, the dimension recognition model 112, and the parametric model 113. The RAM 102 is the work memory for the CPU 101 to execute processing. The CPU 101 loads the analysis program 110 into the RAM 102 and executes processing according to the analysis program 110. Input / output devices (not shown) such as a mouse, keyboard, and display are connected to the input / output interface 104. The network interface 105 is an interface for the information processing device 10 to connect to the network 30. The read / write unit 106 controls the reading and writing of data to the non-volatile memory 103.

[0029] Figure 4 is a functional block diagram showing an example of the functional configuration of the information processing device 10. The information processing device 10 includes a reception unit 11, an analysis unit 12, a calculation processing unit 16, a reliability calculation unit 17, and a display processing unit 18. The analysis unit 12 includes a character recognition unit 13, a shape element recognition unit 14, and a dimension recognition unit 15. When the CPU 101 executes the analysis program 110, the information processing device 10 functions as the reception unit 11, the analysis unit 12 (character recognition unit 13, shape element recognition unit 14, dimension recognition unit 15), the calculation processing unit 16, the reliability calculation unit 17, and the display processing unit 18.

[0030] The terminal device 20 transmits the design drawing to be analyzed along with the analysis request to the information processing device 10. The reception unit 11 receives the input of the design drawing and the analysis request transmitted from the terminal device 20.

[0031] The analysis unit 12 analyzes the design drawings received by the reception unit 11 to estimate a three-dimensional model of the structure shown in the design drawings, and calculates the volume and surface area of ​​the structure from the estimated three-dimensional model. As described above, the analysis unit 12 includes a character recognition unit 13, a shape element recognition unit 14, and a dimension recognition unit 15.

[0032] The character recognition unit 13 acquires character string data by recognizing the character strings contained in the design drawing using known OCR (Optical Character Recognition) technology. The character strings contained in the design drawing include the dimensional values ​​of each part of the structure illustrated in the design drawing. The character recognition unit 13 may correct the character recognition result by using an internal dictionary that records industry-specific terminology to which the design drawing belongs. For example, the technology described in Japanese Patent No. 7377565 can be applied to the technique of correcting the character recognition result using an internal dictionary.

[0033] The shape element recognition unit 14 uses the shape element recognition model 111 to recognize the shape elements included in the structure shown in the design drawing. Shape elements are basic or elemental three-dimensional shapes such as cylinders, cuboids, spheres, and cones. The shape element recognition model 111 is a machine learning model that recognizes which combinations of shape elements (cylinders, cuboids, spheres, cones, etc.) make up the structure shown in the design drawing.

[0034] The shape elements that the shape element recognition unit 14 can recognize are defined in the parametric model 113 stored in the non-volatile memory 103. The shape element recognition unit 14 searches for a parametric model 113 that fits part or all of the structure shown in the design drawing. If the shape element recognition unit 14 recognizes multiple shape elements, it identifies the positional relationship of the multiple recognized shape elements.

[0035] Figure 5 shows a diagram illustrating the concept of a parametric model 113. A parametric model 113 is data that associates a three-dimensional shape model of a shape element such as a cylinder, cuboid, sphere, or cone with a calculation formula for calculating the volume and surface area of ​​that shape element. The non-volatile memory 103 stores the parametric model 113 for each of multiple shape elements. Not only basic shapes as exemplified in Figure 5, but also parametric models 113 corresponding to more complex shapes (e.g., flanged cylinder, tapered cylinder, etc.) and industry-specific standard part shapes (square pipe, L-angle, nut, bearing, etc.) may be stored in the non-volatile memory 103.

[0036] Figure 6 is a conceptual diagram showing an example of the input and output of the shape element recognition model 111. The shape element recognition model 111 receives image data of a design drawing and the parametric model 113 as input. For example, when image data of the design drawing 200 exemplified in Figure 2 is input to the shape element recognition model 111, it outputs a recognition result indicating that the structure 201 shown in the design drawing 200 includes cylinders and rectangular parallelepipeds as shape elements as defined in the parametric model 113.

[0037] The dimension recognition unit 15 uses the dimension recognition model 112 to associate the strings representing dimension values ​​recognized by the character recognition unit 13 with the shape elements recognized by the shape element recognition unit 14. In other words, the dimension recognition unit 15 recognizes which part of the shape elements constituting the structure illustrated in the design drawing corresponds to the strings representing dimension values ​​included in the design drawing. The dimension recognition model 112 is a model constructed by machine learning to perform the above processing. For example, the technology described in Japanese Patent No. 7503085 can be applied as a technique to recognize which part of the structure illustrated in the design drawing corresponds to the strings representing dimension values ​​included in the design drawing.

[0038] Figure 7 is a conceptual diagram showing an example of the input and output of the dimension recognition model 112. The dimension recognition model 112 receives image data of a design drawing, a string of characters recognized by the character recognition unit 13 for the design drawing, and shape elements recognized by the shape element recognition unit 14 as input. For example, if the dimension recognition model 112 receives image data of the design drawing 200 exemplified in Figure 2, a string of characters including dimension values ​​(100, 100, 30, 45, Φ75) recognized by the character recognition unit 13 for the design drawing 200, and shape elements (cylinder, rectangular prism) recognized by the shape element recognition unit 14 as input, it outputs a recognition result indicating that the diameter D and height h of the cylindrical part constituting the structure 201 shown in the design drawing 200 are 75 mm and 45 mm, respectively, and the depth L, width W, and height h of the rectangular prism part are 100 mm, 100 mm, and 30 mm, respectively. In other words, the output of the dimension recognition model 112 can be said to be a three-dimensional model that identifies the three-dimensional shape and dimensions of the structure shown in the design drawing.

[0039] The shape element recognition model 111 and the dimension recognition model 112 may be constructed using deep learning architectures such as convolutional neural networks (CNNs) and transformers. Alternatively, these recognition models may be implemented using other more analytical machine learning algorithms such as support vector machines, random forests, and gradient boosting, or they may use generative models such as LLMs (Large Language Models).

[0040] The character recognition unit 13, shape element recognition unit 14, and dimension recognition unit 15, which constitute the analysis unit 12, perform the above-mentioned processes to identify the three-dimensional shape and dimensions of each part of the structure shown in the design drawing, thereby estimating a three-dimensional shape model of the structure shown in the design drawing.

[0041] The calculation processing unit 16 calculates the volume and surface area of ​​the structure shown in the design drawing from the three-dimensional model estimated by the analysis unit 12. Specifically, the calculation processing unit 16 selects the parametric model 113 stored in the non-volatile memory 103 that corresponds to the shape elements recognized by the shape element recognition unit 14. The calculation processing unit 16 calculates the volume and surface area of ​​the structure shown in the design drawing based on the calculated values ​​obtained by substituting the dimensional values ​​of each part of the shape element recognized by the dimension recognition unit 15 into the calculation formula defined in the selected parametric model 113.

[0042] If the structure shown in the design drawing includes multiple geometric elements, the calculation processing unit 16 selects a corresponding parametric model 113 for each of the recognized geometric elements. The calculation processing unit 16 performs a process for each geometric element to obtain calculated values ​​by substituting the corresponding dimensional values ​​into the calculation formula defined in the selected parametric model 113. The calculation processing unit 16 calculates the volume and surface area of ​​the structure shown in the design drawing by integrating the calculated values ​​for each geometric element. The integration of calculated values ​​includes, for example, the sum of the volumes of each geometric element, the sum of the surface areas, and the subtraction of the surface area of ​​the joints between geometric elements.

[0043] For example, in response to an analysis request for the design drawing 200 illustrated in Figure 2, the calculation processing unit 16 performs the following processing: The shape element recognition unit 14 recognizes that the structure 201 shown in the design drawing 200 includes cylinders and rectangular parallelepipeds as shape elements. The calculation processing unit 16 selects a parametric model 113 from the parametric models 113 stored in the non-volatile memory 103 that corresponds to the recognized shape elements, which are cylinders and rectangular parallelepipeds.

[0044] The calculation processing unit 16 substitutes the dimension values ​​D=75mm and h=45mm recognized by the dimension recognition unit 15 into the calculation formula defined in the parametric model 113 of the cylinder, thereby calculating the volume of the cylinder V=π×(75mm / 2) 2 ×45mm = 198943.7mm 3 And the surface area S = π(75 mm / 2)2 +π×75mm×45mm = 19439.1mm 2 Calculate it.

[0045] Also, the calculation processing unit 16 substitutes the dimension values L = 100mm, W = 100mm, and h = 30mm recognized by the dimension recognition unit 15 into the calculation formula defined in the rectangular parallelepiped parametric model 113, so that the volume V of the rectangular parallelepiped is 100mm × 100mm × 30mm = 300000mm 3 and the surface area S = 2×(100mm × 100mm) + 2×(100mm × 30mm) + 2×(100mm × 30mm) = 32000mm 2 Calculate it.

[0046] The calculation processing unit 16 calculates the calculated value V = 198943.7mm of the volume of the cylinder 3 and the calculated value V = 300000mm of the volume of the rectangular parallelepiped 3 The combined value 498943.7mm 3 is calculated as the volume of the structure 201 shown in the design drawing 200. Also, the calculation processing unit 16 calculates the calculated value S = 仅19439.1mm of the surface area of the cylinder 2 and the calculated value S = 32000mm of the surface area of the rectangular parallelepiped 2 The value 42602.5mm obtained by subtracting the surface area of the joint part between the cylinder and the rectangular parallelepiped from the combined value 2 is calculated as the surface area of the structure 201 shown in the design drawing 200.

[0047] The reliability calculation unit 17 calculates a reliability score indicating the reliability for the volume and surface area of the structure calculated by the calculation processing unit 16. The reliability calculation unit 17 may calculate the reliability score based on, for example, the probability that the shape element output from the output layer of the shape element recognition model 111 fits the structure shown in the design drawing and the probability that the character string output from the output layer of the dimension recognition model 112 is the dimension value of the corresponding part of the structure shown in the design drawing.

[0048] It should be noted that there seems to be an error in the text you provided. In line , the surface area value of the cylinder is written as "仅19439.1mm", which is likely a typo. It should be "19439.1mm" as in line . This note is for your reference only and not part of the translation.The display processing unit 18 displays the volume and surface area of ​​the structure shown in the design drawing, calculated by the calculation processing unit 16, and the reliability score calculated by the reliability calculation unit 17, on the display of the terminal device 20 that issued the analysis request. Figure 8 is a diagram showing an example of how the analysis results by the information processing device 10 are displayed. The display processing unit 18 may also display the volume and surface area of ​​the structure shown in the design drawing and the reliability score together with the design drawing to be analyzed on the display 21 of the terminal device 20. In addition, the display processing unit 18 may further display the recognition results by the shape element recognition unit 14 and the dimension recognition unit 15, as well as the calculation process of volume and surface area in the calculation processing unit 16, on the display 21 of the terminal device 20 in response to a request from the terminal device 20. Furthermore, the display processing unit 18 may also display the above various information on the display provided by the information processing device 10.

[0049] Figure 9 is a flowchart illustrating an example of the processing flow performed when the CPU 101 of the information processing device 10 executes the analysis program 110. For example, the analysis program 110 is executed when a user operates the terminal device 20 and sends a request for analysis of a design drawing along with the design drawing to be analyzed to the information processing device 10.

[0050] In step S1, the reception unit 11 receives input and analysis requests for design drawings transmitted from the terminal device 20.

[0051] In step S2, the character recognition unit 13 acquires character data by recognizing the character string contained in the design drawing using known OCR technology.

[0052] In step S3, the shape element recognition unit 14 uses the shape element recognition model 111 to recognize the shape elements included in the structure shown in the design drawing. Specifically, the shape element recognition unit 14 searches for a parametric model 113 that fits part or all of the structure shown in the design drawing.

[0053] In step S4, the dimension recognition unit 15 uses the dimension recognition model 112 to associate the strings representing dimension values ​​recognized in step S2 with the shape elements recognized in step S3. That is, the dimension recognition unit 15 recognizes which part of the shape elements constituting the structure illustrated in the design drawing corresponds to the strings representing dimension values ​​included in the design drawing.

[0054] In step S5, the calculation processing unit 16 selects a parametric model 113 stored in the non-volatile memory 103 that corresponds to the shape element recognized in step S3. Based on the calculated values ​​obtained by substituting the dimensional values ​​of each part of the shape element recognized in step S4 into the calculation formula defined in the selected parametric model 113, the calculation processing unit 16 calculates the volume and surface area of ​​the structure shown in the design drawing.

[0055] In step S6, the reliability calculation unit 17 calculates a reliability score indicating the reliability of the volume and surface area of ​​the structure calculated in step S5.

[0056] In step S7, the display processing unit 18 displays the volume and surface area of ​​the structure calculated in step S5, as well as the confidence score calculated in step S6, on the display of the terminal device 20 that issued the analysis request.

[0057] As described above, the information processing device 10 according to the embodiment of the disclosed technology receives input of a design drawing, analyzes the design drawing to estimate a three-dimensional model of the structure shown in the design drawing, and calculates the volume and surface area of ​​the structure from the estimated three-dimensional model. According to the information processing device 10 according to the embodiment of the disclosed technology, it is possible to efficiently calculate approximate values ​​of volume and surface area for structures for which 3D-CAD data does not exist. As a result, it is possible to accurately calculate the amount of regulated substances used and emitted, as well as Scope 3 emissions of greenhouse gases, compared to when calculations are made using a simplified method.

[0058] The information processing device 10 may further have a function to perform machine learning to improve the recognition accuracy of the shape element recognition model 111 and the dimension recognition model 112. Figure 10 is a diagram showing an example of a machine learning method for improving the recognition accuracy of the shape element recognition model 111 and the dimension recognition model 112. The information processing device 10 has a learning processing unit 19 and a database 50.

[0059] Database 50 stores geometric and topological data extracted from 3D-CAD data 51. The learning processing unit 19 uses the geometric and topological data stored in database 50 to generate training data pairs of image data of design drawings and ground truth data of the volume and surface area of ​​structures illustrated in the design drawings. The learning processing unit 19 uses this training data to perform machine learning on the shape element recognition model 111 and the dimension recognition model 112. When the learning processing unit 19 provides the design drawings for training to the shape element recognition model 111 and the dimension recognition model 112, it sets the parameters of the shape element recognition model 111 and the dimension recognition model 112 so that the loss function corresponding to the difference between the volume and area of ​​the structures output from the calculation processing unit 16 and the ground truth data is minimized. A new parametric model may also be created using the geometric and topological data stored in database 50.

[0060] The following additional information is disclosed regarding the embodiments described above. (Note 1) An information processing device comprising at least one processor, The aforementioned processor, Accepting input of design drawings, By analyzing the aforementioned design drawings, a three-dimensional model of the structure illustrated in the design drawings is estimated. From the estimated three-dimensional model, at least one of the volume and surface area of ​​the structure is calculated. Information processing device.

[0061] (Note 2) The aforementioned processor, Recognizing the string of characters contained in the aforementioned design drawing, Recognizing the shape elements that constitute the aforementioned structure, A three-dimensional model of the structure is estimated by associating the recognized strings representing dimensional values ​​with the recognized shape elements. The information processing device described in Appendix 1.

[0062] (Note 3) The system has a storage medium that stores a parametric model for each of multiple shape elements, which associates a three-dimensional model of the shape element with a calculation formula for determining at least one of the volume and surface area of ​​the shape element. The processor selects from the parametric models stored in the storage medium those corresponding to the shape elements constituting the structure, The volume and surface area of ​​the structure are calculated based on the calculated values ​​obtained by substituting the dimensional values ​​of the shape elements constituting the structure into the calculation formula in the selected parametric model. The information processing device described in Appendix 1 or Appendix 2.

[0063] (Note 4) When the processor recognizes multiple shape elements constituting the structure, it calculates at least one of the volume and surface area of ​​the structure by integrating the calculated values ​​for each shape element obtained by substituting the dimensional values ​​of the shape element corresponding to the calculation formula in the selected parametric model for each of the recognized shape elements. The information processing device described in Appendix 3.

[0064] (Note 5) The aforementioned processor, Output a confidence score indicating the confidence level for at least one of the volume and surface area calculated for the aforementioned structure. An information processing device as described in any one of the appendices 1 through 4.

[0065] (Note 6) The processor displays on the display the calculated value of at least one of the volume and surface area of ​​the structure, and the confidence score. The information processing device described in Appendix 5.

[0066] (Note 7) The aforementioned processor, The process of recognizing the shape elements constituting the aforementioned structure, and the process of associating the recognized strings indicating dimension values ​​with the recognized shape elements, are performed using a recognition model constructed by machine learning. The machine learning of the recognition model is performed using data based on CAD data showing the three-dimensional shape of the structure as training data. An information processing device as described in any one of the appendices 1 through 6.

[0067] (Note 8) Accepting input of design drawings, By analyzing the aforementioned design drawings, a three-dimensional model of the structure illustrated in the design drawings is estimated. From the estimated three-dimensional model, at least one of the volume and surface area of ​​the structure is calculated. An information processing method in which processing is performed by at least one processor provided in an information processing device.

[0068] (Note 9) Accepting input of design drawings, By analyzing the aforementioned design drawings, a three-dimensional model of the structure illustrated in the design drawings is estimated. From the estimated three-dimensional model, at least one of the volume and surface area of ​​the structure is calculated. A program that causes at least one processor in an information processing device to perform a task.

Claims

1. An information processing device comprising at least one processor, The aforementioned processor, Accepting input of design drawings, Recognizing the string of characters contained in the aforementioned design drawing, Recognizing the shape elements that constitute the structure shown in the aforementioned design drawing, By associating the recognized strings indicating dimensional values ​​with the recognized shape elements, a three-dimensional model of the structure is estimated. A parametric model is created by associating a three-dimensional model of a shape element with a calculation formula for calculating at least one of the volume and surface area of ​​the shape element. From among the parametric models stored in a storage medium that stores each of a plurality of shape elements, the one corresponding to the shape element constituting the structure is selected. The volume and surface area of ​​the structure are calculated based on the calculated values ​​obtained by substituting the dimensional values ​​of the shape elements constituting the structure into the calculation formula in the selected parametric model. Information processing device.

2. (delete)

3. (delete)

4. When the processor recognizes multiple shape elements constituting the structure, it calculates at least one of the volume and surface area of ​​the structure by integrating the calculated values ​​for each shape element obtained by substituting the dimensional values ​​of the shape elements corresponding to the calculation formula in the selected parametric model for each of the recognized shape elements. The information processing apparatus according to claim 1.

5. The aforementioned processor, Output a confidence score indicating the confidence level for at least one of the volume and surface area calculated for the aforementioned structure. The information processing apparatus according to claim 1.

6. The processor displays on the display the calculated value of at least one of the volume and surface area of ​​the structure, and the confidence score. The information processing apparatus according to claim 5.

7. The aforementioned processor, The process of recognizing the shape elements constituting the aforementioned structure, and the process of associating the recognized strings indicating dimension values ​​with the recognized shape elements, are performed using a recognition model constructed by machine learning. The machine learning of the recognition model is performed using data based on CAD data showing the three-dimensional shape of the structure as training data. The information processing apparatus according to claim 1.

8. Accepting input of design drawings, Recognizing the string of characters contained in the aforementioned design drawing, Recognizing the shape elements that constitute the structure shown in the aforementioned design drawing, By associating the recognized strings indicating dimensional values ​​with the recognized shape elements, a three-dimensional model of the structure is estimated. A parametric model is created by associating a three-dimensional model of a shape element with a calculation formula for calculating at least one of the volume and surface area of ​​the shape element. From among the parametric models stored in a storage medium that stores each of a plurality of shape elements, the one corresponding to the shape element constituting the structure is selected. The volume and surface area of ​​the structure are calculated based on the calculated values ​​obtained by substituting the dimensional values ​​of the shape elements constituting the structure into the calculation formula in the selected parametric model. An information processing method in which processing is performed by at least one processor provided in an information processing device.

9. Accepting input of design drawings, Recognizing the string of characters contained in the aforementioned design drawing, Recognizing the shape elements that constitute the structure shown in the aforementioned design drawing, By associating the recognized strings indicating dimensional values ​​with the recognized shape elements, a three-dimensional model of the structure is estimated. A parametric model is created by associating a three-dimensional model of a shape element with a calculation formula for calculating at least one of the volume and surface area of ​​the shape element. From among the parametric models stored in a storage medium that stores each of a plurality of shape elements, the one corresponding to the shape element constituting the structure is selected. The volume and surface area of ​​the structure are calculated based on the calculated values ​​obtained by substituting the dimensional values ​​of the shape elements constituting the structure into the calculation formula in the selected parametric model. A program that causes at least one processor in an information processing device to perform a process.

Images