Similar drawing retrieval device, similar drawing retrieval method, similar drawing retrieval program, multimodal retrieval display device, multimodal retrieval display method, and multimodal retrieval display program

Abstract

A similar drawing retrieval device according to an embodiment of the present invention retrieves a drawing similar to a target drawing from a drawing database in which feature information items related to the shapes of components shown in a plurality of drawings and the plurality of drawings are stored in association with each other. The similar drawing retrieval device according to an embodiment of the present invention comprises a feature information acquisition unit, a similar drawing retrieval unit, and a display unit. The feature information acquisition unit acquires feature information related to the shape of a target component shown in the target drawing as feature information on the target drawing. The similar drawing retrieval unit retrieves, from the drawing database and on the basis of the feature information on the target drawing, a plurality of first similar drawings which show components in prescribed dimensions and a plurality of second similar drawings which show the components in dimensions different from the prescribed dimensions in the first similar drawing. The display unit displays the plurality of first similar drawings and / or the plurality of second similar drawings.

Description

Similar drawing search device, similar drawing search method, similar drawing search program, multimodal search display device, multimodal search display method, and multimodal search display program

[0001] An embodiment of the present invention relates to a similar drawing search device, a similar drawing search method, a similar drawing search program, a multimodal search and display device, a multimodal search and display method, and a multimodal search and display program.

[0002] In recent years, the use of 3D and 2D computer-aided design (CAD) has been increasing in the manufacturing industry, particularly in design work. For example, there are three situations in which CAD is used, mainly in the process industry: 1. CAD-based design during the development and design stage. 2. After CAD design, 2D drawings are issued. Additional information such as welding instructions, precision information, and whether or not painting is required is added to these 2D drawings. 3. Drawings are used as the primary reference in the processes and value chain following design, such as procurement, manufacturing, quality inspection, and sales and after-sales service.

[0003] Currently, there is a trend to promote the addition of such supplementary information to 3D CAD data. However, the addition of supplementary information to 3D CAD data is still in its infancy. In addition to this trend, there is also a trend to carry out business operations centered on 3D CAD data in processes after design.

[0004] On the other hand, in the case of users who place large orders for products in design or procurement, the process of ordering drawings for those products is often carried out frequently. As a result, a large amount of drawing data is created. In order to utilize the large amount of drawing data created during product ordering and to improve the efficiency of the ordering and receiving process, for example, there are cases where users want to refer to similar drawings from the past.

[0005] Furthermore, in the manufacturing industry, design data representing any given product is not managed in an integrated manner in a single format, and various design data for the same product exist in different dimensions and file types (hereinafter referred to as "modes"). Examples include: 1. 3D-based CAD data, which is mainstream in development and design. This mode is excellent for grasping the shape of the product. 2. 2D-based CAD data, used in procurement and manufacturing departments. Various manufacturing support information such as welding instructions, precision information, and painting is attached. 3. PDF data converted from 2D CAD and paper drawings printed from this. This has excellent immediacy of reference and allows additional information to be written by hand.

[0006] JP 2024-000702 A

[0007] Conventionally, in searching for similar drawings, it has sometimes been difficult to easily search between three-dimensional drawings such as three-dimensional CAD data and two-dimensional drawings.

[0008] Furthermore, because information from multiple modes is managed using multiple management methods, the following issues arise: 1. Information needs to be searched for in each mode, which takes time to find. 2. It is difficult to search for information from another mode using information from a different mode. 3. Information from different modes is difficult to share and utilize across different departments.

[0009] As a result, not only is information search wasteful, but problems also arise, such as repeating the same mistakes and designing similar products without reflecting design improvements.Therefore, there is a need for a method that allows users to search for information across multiple modes (multimodal) and view the information in a single opportunity.

[0010] The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a method for searching design data multimodally and for effectively viewing the results of the multimodal search in an integrated manner.

[0011] A similar drawing search device according to an embodiment is a similar drawing search device that searches for drawings similar to a target drawing from a drawing database that stores feature information about the shapes of parts shown in each of a plurality of drawings in association with each of the plurality of drawings. The similar drawing search device according to an embodiment includes a feature information acquisition unit, a similar drawing search unit, and a display unit. The feature information acquisition unit acquires feature information about the shape of a target part shown in the target drawing as feature information for the target drawing. The similar drawing search unit searches the drawing database for a plurality of first similar drawings that show parts in a predetermined dimension and a plurality of second similar drawings that show parts in a dimension different from the predetermined dimension of the first similar drawings, based on the feature information of the target drawing. The display unit displays the plurality of first similar drawings and / or the plurality of second similar drawings.

[0012] In the similar drawing search device according to the embodiment, the target drawing and the first similar drawing are drawings showing parts in the same dimension, and the display unit switches between displaying a plurality of the first similar drawings and a plurality of the second similar drawings in response to a switching instruction.

[0013] In the similar drawing search device of the embodiment, the multiple first similar drawings and the multiple second similar drawings are associated using at least one of drawing information regarding the parts they respectively indicate and the feature information associated with each of them, and the target drawing and the first similar drawings are drawings showing parts in the same dimension, and the similar drawing search unit searches the multiple first similar drawings from the drawing database based on the feature information of the target drawing, and searches the multiple second similar drawings from the drawing database using at least one of the drawing information and feature information of the multiple first similar drawings retrieved.

[0014] In the similar drawing search device according to the embodiment, the target drawings include a two-dimensional target drawing related to the target component and a three-dimensional target drawing related to the target component.

[0015] In the similar drawing search device according to the embodiment, the feature information acquisition unit acquires first feature information from the two-dimensional target drawing and acquires second feature information from the three-dimensional target drawing, and the similar drawing search unit searches the drawing database for a plurality of the two-dimensional first similar drawings and a plurality of the three-dimensional second similar drawings based on the first feature information, and searches the drawing database for a plurality of the three-dimensional second similar drawings and a plurality of the two-dimensional first similar drawings based on the second feature information.

[0016] A similar drawing search method according to an embodiment is a similar drawing search method that searches for drawings similar to a target drawing from a drawing database that stores feature information regarding the shape of parts shown in each of a plurality of drawings in correspondence with each of the plurality of drawings. Feature information regarding the shape of the target part shown in the target drawing is obtained as feature information for the target drawing. Based on the feature information for the target drawing, the drawing database is searched for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings, and the plurality of first similar drawings and / or the plurality of second similar drawings are displayed on a screen.

[0017] A similar drawing search program according to an embodiment is a similar drawing search program that enables a computer to search for drawings similar to a target drawing from a drawing database that stores characteristic information regarding the shape of parts shown in each of a plurality of drawings in correspondence with each of the plurality of drawings.The similar drawing search program obtains characteristic information regarding the shape of the target part shown in the target drawing as characteristic information for the target drawing, and based on the characteristic information for the target drawing, searches the drawing database for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings, and displays the plurality of first similar drawings and / or the plurality of second similar drawings on a display.

[0018] The multimodal search and display device according to the third embodiment is a multimodal search and display device that multimodally searches a database that stores multimodal design data for design data that meets conditions specified by a user and displays the search results on a screen, and is equipped with an analysis unit, a search unit, and a display unit.

[0019] The analysis unit extracts or detects and recognizes information such as characters from the design data, and acquires feature information of the objects in the design data.

[0020] The search unit has a keyword search function that searches the database for design data that matches at least a part of the character information specified by the user based on the character information specified by the user, and a similarity search function that searches the database for design data that is similar to the characteristic information specified by the user based on the characteristic information of an object in the design data, and performs searches multimodally.

[0021] The display unit displays the search results obtained from the search in a manner that allows the mode for each piece of design data to be recognized, and further displays the search results for each mode in response to a display switching instruction from the user.

[0022] The multimodal search and display program according to the third embodiment enables a computer to search a drawing database that stores multimodal design data for design data that meets conditions specified by a user and display the search results on a screen.

[0023] According to one aspect of the present invention, three-dimensional drawings and two-dimensional drawings can be easily searched.

[0024] Furthermore, users of the design data can search and view product design data in a multimodal manner.

[0025] FIG. 1 is a diagram illustrating an example of the hardware configuration of a similar drawing search apparatus 1 according to the first embodiment. FIG. 2 is a diagram illustrating an example of an outline of a similar drawing search process according to the first embodiment when a target drawing and a first similar drawing are two-dimensional drawings. FIG. 3 is a diagram illustrating an example of an outline of a similar drawing search process according to the first embodiment when a target drawing and a first similar drawing are three-dimensional drawings. FIG. 4 is a flowchart illustrating an example of a procedure for a similar drawing search process according to the first embodiment. FIG. 5 is a diagram illustrating an example of a display on which a plurality of first similar drawings are displayed when the target drawing and the first similar drawing are two-dimensional drawings according to the first embodiment. FIG. 6 is a diagram illustrating an example of a display on which a plurality of first similar drawings are displayed when the target drawing and the first similar drawing are three-dimensional drawings according to the first embodiment. FIG. 7 is a diagram illustrating an example of a display on which a plurality of first similar drawings are displayed together with a second similar drawing when the target drawing and the first similar drawing are two-dimensional drawings according to the first embodiment. FIG. 8 is a diagram showing an example of a display on which a plurality of second similar drawings are displayed when the target drawing and the first similar drawing are two-dimensional drawings according to the first embodiment. FIG. 9 is a diagram showing an example of a display on which a plurality of second similar drawings are displayed when the target drawing and the first similar drawing are three-dimensional drawings according to the first embodiment. FIG. 10 is a diagram showing an example of a display on which a plurality of second similar drawings of a second pair are displayed according to a modified example of the first embodiment. FIG. 11 is a diagram showing an example of an overview of a similar drawing search process using a third trained model according to the second embodiment. FIG. 12 is a flowchart showing an example of the procedure of a similar drawing search process according to the second embodiment. FIG. 13 is a diagram showing an example of the hardware configuration of a multimodal search and display device according to the third embodiment. FIG. 14 is a flowchart showing an example of the procedure of a multimodal search and display device according to the third embodiment.

[0026] Hereinafter, the embodiments will be described in detail with reference to the drawings.

[0027] First Embodiment FIG. 1 illustrates an example of the hardware configuration of a similar drawing search device 1 according to this embodiment. The similar drawing search device 1 searches for drawings similar to a target drawing from a drawing database that stores characteristic information about the shapes of components in each of a plurality of drawings, in association with each of the drawings. The similar drawing search device 1 may also be implemented as a server connected to a network 191. In this case, components in a drawing correspond to component information corresponding to, for example, a component drawing, which refers to a drawing limited to one perspective of a metalwork product, an assembly drawing combining the component drawings, a circuit diagram, or the like. Hereinafter, for ease of explanation, the above-mentioned components will be described as component drawings.

[0028] 1, the similar drawing search device 1 includes a processing circuit 11, a memory 13, an input interface 15, a display 17, and a communication interface 19. As shown in Fig. 1, the processing circuit 11, the memory 13, the input interface 15, the display 17, and the communication interface 19 are electrically connected via a bus. Therefore, the processing circuit 11, the memory 13, the input interface 15, the display 17, and the communication interface 19 appropriately transmit and receive data under the control of a system control unit 111.

[0029] The memory 13 may be realized as a storage device corresponding to storage. The memory 13 corresponds to a storage unit. The input interface 15 corresponds to an input unit. The display 17 corresponds to a display unit. The communication interface 19 corresponds to a communication unit.

[0030] The processing circuitry 11 includes, for example, a processor and an internal memory as hardware resources. The processing circuitry 11 may be implemented by a computer or the like. The processor may be implemented by, for example, a central processing unit (CPU), a graphics processing unit (GPU), a micro processing unit (MPU), a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a complex programmable logic device (CPLD), a simple programmable logic device (SPLD), or the like. The internal memory is realized by, for example, a ROM (Read Only Memory) and / or a RAM (Random Access Memory), etc. The hardware that realizes the processor and the internal memory is not limited to the above, and any known hardware may be used as appropriate.

[0031] The processing circuitry 11 includes a system control unit 111, a feature information acquisition unit 113, and a similar drawing search unit 115, depending on the functions to be realized. The division of units by function is not limited to the above and can be set as appropriate. The system control unit 111, the feature information acquisition unit 113, and the similar drawing search unit 115 that realize various functions by the processing circuitry 11 will be described later.

[0032] The memory 13 is a storage device such as a hard disk drive (HDD), a solid state drive (SSD), or an integrated circuit storage device that stores various information. The memory 13 stores various programs according to the present embodiment. For example, the memory 13 stores programs related to the execution of processes in the system control unit 111, the feature information acquisition unit 113, and the similar drawing search unit 115, which are executed by the processing circuitry 11.

[0033] The memory 13 stores, for example, each of the multiple drawings as drawing data. The memory 13 also stores feature information related to the shape of a part shown in each of the multiple drawings, in association with the drawing ID of each of the multiple drawings. The memory 13 may also store drawing information related to the part drawing in association with the drawing ID. The drawing ID, feature information, and drawing information may be referred to as index data. Each of the multiple drawings is stored in the memory 13 as drawing data. The index data and drawing data are stored in the memory 13 as a database related to the multiple drawings (hereinafter referred to as a drawing database). That is, the memory 13 stores a drawing database in which feature information is associated with each of the multiple drawings.

[0034] Hereinafter, each of the multiple drawings stored in the drawing database will be referred to as a "database drawing." In the database drawings, two-dimensional drawings (2D drawings) and three-dimensional drawings (3D drawings) such as 3D CAD data are associated using, for example, at least one of drawing information related to parts and feature information related to shapes. For example, in the drawing database, two-dimensional drawings and three-dimensional drawings are linked by the file name, title, drawing number, and / or part number in the drawing information.

[0035] Drawing information includes at least one of text attached to a component drawing in a drawing and symbol information attached to the component drawing. The text is, for example, character information and numerical information attached to the drawing. Specifically, the text includes the drawing number, drawing number, product name, material, and processing information (e.g., surface treatment, sheet metal processing, machining, and other processing methods) listed in the title block of the drawing, as well as dimensions and the number of holes listed in the component drawing, part number, file name, title, and other information assigned by the user to each drawing. Furthermore, the symbol information attached to the component drawing includes, for example, symbols that abbreviate the text and / or information related to symbols assigned by the user.

[0036] The memory 13 also stores a trained model that receives a drawing (hereinafter referred to as a user surface) to be searched for similar drawings as input and determines (outputs) the shape of the target drawing. The memory 13 stores the feature information output from at least one intermediate layer in the trained model. In other words, the feature information corresponds to the intermediate output of the target drawing output from at least one intermediate layer in the trained model. The at least one intermediate layer is, for example, the intermediate layer immediately before the output layer in the trained model, i.e., the final intermediate layer. The trained model is a model that can output feature information between input and output, and is, for example, a model composed of a multi-layer neural network. The intermediate output is expressed, for example, as a vector (hereinafter referred to as a feature vector) that indicates the features of the shape of the input target drawing.

[0037] The memory 13 may be a drive device that reads and writes various information from and to portable storage media such as a CD (Compact Disc), a DVD (Digital Versatile Disc), a flash memory, etc., or semiconductor memory elements such as a RAM (Random Access Memory), etc. The storage area of ​​the memory 13 may be located within the similar drawing search device 1, or may be provided in an external storage device connected via the network 31.

[0038] The input interface 15 accepts input operations related to various instructions from the user, converts the accepted input operations into electrical signals, and outputs the electrical signals to the processing circuitry 11. Examples of the input interface 15 that can be used as appropriate include a mouse, keyboard, trackball, switch, button, joystick, touchpad, and touch panel display. Note that in this embodiment, the input interface 15 is not limited to those equipped with physical operation components such as a mouse, keyboard, trackball, switch, button, joystick, touchpad, and touch panel display. For example, an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the similar drawing search device 1 and outputs the electrical signal to the processing circuitry 11 is also included as an example of the input interface 15.

[0039] The display 17 displays various information. For example, the display 17 outputs search results for similar drawings by the processing circuit 11 and a GUI (Graphical User Interface) for accepting various operations from the user. The display 17 may be, for example, a liquid crystal display (LCD), a cathode ray tube (CRT), an organic electroluminescence display (OELD), a plasma display, or any other display. The display 17 may be a desktop type or may be a tablet terminal or the like capable of wireless communication with the similar drawing search device 1.

[0040] The communication interface 19 performs data communication with other input / output devices, for example, via a network 191. Any standard may be used for communication by the communication interface 19. The communication interface 19 outputs drawing data acquired through data communication with other input / output devices to the memory 13. The communication interface 19 receives electrical signals representing input operations input to other input interfaces of other input / output devices. The communication interface 19 outputs the received electrical signals to the processing circuit 11.

[0041] The system control unit 111 controls the processing of each unit in the processing circuit 11 based on an electrical signal corresponding to an input operation received from a user via the input interface 15. Specifically, the system control unit 111 reads out a control program stored in the memory 13, expands it on the internal memory of the processing circuit 11, and controls each unit of the similar drawing search device 1 in accordance with the expanded control program.

[0042] The feature information acquisition unit 113 acquires feature information related to the shape of a target part shown in a target drawing as feature information of the target drawing. Specifically, the feature information acquisition unit 113 acquires feature information related to the shape of a part drawing in the target drawing from the target drawing input by the user via the input interface 15 or the communication interface 19. The feature information acquisition unit 113 associates the acquired feature information with the ID of the target drawing and stores it in the memory 13.

[0043] For example, the feature information acquisition unit 113 inputs the target drawing into a trained model and acquires intermediate output from at least one intermediate layer in the trained model as feature information related to the shape of the part drawing. The trained model is realized by a trained CNN (Convolutional Neural Network). As the trained CNN, for example, EfficientNet described in [1905.11946] EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks, or the like, is used.

[0044] The image format input to the trained model is assumed to be any two-dimensional drawing, such as PNG (Portable Network Graphics) format or TIFF (Tagged Image File Format). The target drawing input to the trained model may also be a handwritten drawing by a user. If the target drawing is a three-dimensional CAD drawing (CAD data), the drawing input to the trained model may be an image rendered from the three-dimensional CAD drawing (hereinafter referred to as a rendered image). The rendering process may be performed by the feature information acquisition unit 113, or may be performed by another unit or an external device via the communication interface 19 and the network 191. The rendered image may be, for example, a three-view drawing, such as a plan view, a front view, or a side view (right side view). The image format is not limited to the above and may also be three-dimensional CAD data. At this time, different trained models depending on the dimensions of the input image are stored in advance in the memory 13.

[0045] Furthermore, the target drawing input to the trained model may be a part drawing region extracted from the target drawing. In this case, the feature information acquisition unit 113 extracts the part drawing region by performing a predetermined extraction process on the drawing prior to executing the trained model. This extraction process can utilize various image processing techniques, such as existing segmentation processing, and therefore will not be described further. Next, the feature information acquisition unit 113 inputs the part drawing data into the trained model to acquire an intermediate output related to the target drawing. While the intermediate output may be a vector output from one of multiple intermediate layers in the trained model, it is preferably the intermediate output input to the output layer (the layer that outputs the shape of the target drawing, etc.) in the trained model, i.e., the intermediate output from the intermediate layer immediately preceding the output layer (hereinafter referred to as the final intermediate layer). The intermediate output may be an output from an intermediate layer other than the final intermediate layer, or may be an output from multiple intermediate layers.

[0046] The trained model is generated by inputting each of the multiple database drawings stored in the drawing database into a pre-trained model and training the pre-trained model using each of the multiple shapes corresponding to the multiple database drawings as training data. The trained model is generated, for example, by training a multi-layer neural network that can output from an intermediate layer using the database drawings and shapes related to the database drawings as training data. For example, if the product category of the shape is the training data, the trained model is generated by inputting the drawings as images into a convolutional neural network (CNN) and training the CNN to classify the shape classes into, for example, eight classes. Known methods can be applied to generating the trained model, so a description thereof will be omitted.

[0047] Hereinafter, the trained model into which a two-dimensional drawing is input will be referred to as the first trained model. The trained model into which a three-dimensional drawing is input will be referred to as the second trained model. The first trained model and / or the second trained model are stored in memory 13 in advance.

[0048] Based on the feature information of the target drawing, the similar drawing search unit 115 searches the drawing database for a plurality of first similar drawings showing the component in a predetermined dimension and a plurality of second similar drawings showing the component in a dimension different from the predetermined dimension of the first similar drawings. For example, the target drawing and the first similar drawings are drawings showing the component in the same dimension. Hereinafter, for the sake of concrete explanation, it is assumed that the target drawing and the first similar drawings are two-dimensional drawings. In this case, the predetermined dimension is two-dimensional, and the second similar drawings correspond to three-dimensional drawings such as three-dimensional CAD data.

[0049] For example, the similar drawing search unit 115 searches the drawing database for multiple first similar drawings based on the feature information of the target drawing. That is, the similar drawing search unit 115 searches the drawing database for feature information similar to the feature information of the target drawing. Specifically, the similar drawing search unit 115 calculates the distance between the feature vector of the target drawing (hereinafter referred to as the target feature vector) and each of multiple feature vectors (hereinafter referred to as 2D search feature vectors) corresponding to multiple two-dimensional database drawings (hereinafter referred to as 2D base drawings) stored in the drawing database. The distance is, for example, a Mahalanobis distance, which corresponds to an index (hereinafter referred to as similarity) indicating the similarity between the target drawing and the 2D base drawings. The similar drawing search unit 115 identifies, as multiple first similar drawings, 2D base drawings corresponding to 2D search feature vectors with distances smaller than a predetermined value among the multiple calculated distances. The predetermined value is a value capable of distinguishing the similarity with the shape of the target drawing, and is preset and stored in the memory 13.

[0050] That is, the similar drawing search unit 115 identifies a 2D search feature vector that is equal to or smaller than a predetermined value in a distance space defined by the target feature vector and each of the plurality of 2D search feature vectors. Then, the similar drawing search unit 115 identifies a 2D base drawing corresponding to the identified 2D search feature vector as a first similar drawing.

[0051] Next, the similar drawing search unit 115 searches the drawing database for multiple second similar drawings using the drawing information and feature information of the multiple first similar drawings. For example, the similar drawing search unit 115 searches the drawing database for multiple second similar drawings using at least one of the drawing information and feature information of the retrieved multiple first similar drawings. At this time, the multiple first similar drawings and the multiple second similar drawings are associated with each other using at least one of the drawing information regarding the components represented by each drawing and the feature information associated with each drawing. Specifically, the similar drawing search unit 115 searches the drawing database for multiple three-dimensional database drawings (hereinafter referred to as three-dimensional base drawings) linked to the first similar drawings by drawing information or feature information (feature vectors). The similar drawing search unit 115 identifies the retrieved three-dimensional base drawings as second similar drawings.

[0052] In the above description, the target drawing and the first similar drawing are described as two-dimensional drawings. However, the target drawing and the first similar drawing may also be three-dimensional drawings, such as three-dimensional CAD data. In this case, the second similar drawing corresponds to a two-dimensional drawing. Specifically, the similar drawing search unit 115 calculates the distance between the target feature vector and each of multiple feature vectors (hereinafter referred to as 3D search feature vectors) corresponding to multiple three-dimensional database drawings (hereinafter referred to as 3D base drawings) stored in the drawing database. The distance is, for example, the Mahalanobis distance, which corresponds to an index (similarity) indicating the similarity between the target drawing and the three-dimensional base drawings. The similar drawing search unit 115 identifies, as multiple first similar drawings, 3D base drawings corresponding to the 3D search feature vectors with distances smaller than a predetermined value among the multiple calculated distances. Next, the similar drawing search unit 115 searches the drawing database for multiple 2D base drawings linked to the first similar drawing by drawing information and / or feature information (feature vectors). The similar drawing search unit 115 identifies the searched two-dimensional base drawing as a second similar drawing.

[0053] The above describes the configuration of the similar drawing search device 1. Below, we will explain the process related to the search for similar drawings by the similar drawing search device 1 (hereinafter referred to as the similar drawing search process). The similar drawing search process is to search for and display drawings similar to the target drawing from the drawing database. Below, we will explain the similar drawing search process using Figures 2 to 9.

[0054] FIG. 2 is a diagram illustrating an example of an overview of a similar drawing search process when the target drawing and the first similar drawing are two-dimensional drawings. As shown in FIG. 2, when a two-dimensional drawing (2D drawing) is input as a target drawing to the first trained model, the target drawing is characterized. The characterized drawing corresponds to obtaining feature information from the target drawing. Also, as shown in FIG. 2, a two-dimensional first similar drawing is identified by searching a drawing database using the feature information. In addition, a three-dimensional second similar drawing is identified by searching the drawing database for drawings linked to the first similar drawing. As shown in FIG. 2, the first similar drawing is displayed two-dimensionally on the display 17. Also, as shown in FIG. 2, the second similar drawing is displayed three-dimensionally on the display 17.

[0055] FIG. 3 is a diagram illustrating an example of an overview of a similar drawing search process when the target drawing and the first similar drawing are three-dimensional drawings. As shown in FIG. 2, when a three-dimensional drawing (3D drawing) is input as a target drawing to the second trained model, feature quantification of the target drawing is performed. Also, as shown in FIG. 3, a three-dimensional first similar drawing is identified by searching a drawing database using the feature information. Additionally, a two-dimensional second similar drawing is identified by searching the drawing database for drawings linked to the first similar drawing. As shown in FIG. 3, the first similar drawing is displayed in three dimensions on the display 17. Also, as shown in FIG. 3, the second similar drawing is displayed in two dimensions on the display 17.

[0056] FIG. 4 is a flowchart showing an example of a procedure for similar drawing search processing.

[0057] (Similar drawing search process) (Step S401) The input interface 15 inputs a target drawing in response to a user instruction. The input of the target drawing is not limited to an operation via the input interface 15. For example, the target drawing may be transmitted from a terminal device to the similar drawing search device 1 in response to a user instruction via the terminal device of the user who can use the similar drawing search process. At this time, the communication interface 19 inputs the target drawing to the similar drawing search device 1 via the network 191.

[0058] (Step S402) The feature information acquisition unit 113 inputs the target drawing input by the user into the trained model. Next, the feature information acquisition unit 113 acquires intermediate output output from at least one intermediate layer in the trained model as feature information. The feature information acquisition unit 113 associates the acquired feature information with the ID of the target drawing and stores it in the memory 13.

[0059] (Step S403) The similar drawing search unit 115 searches the drawing database for multiple first similar drawings that show components in the same dimension as the target component shown in the target drawing, based on the feature information of the target drawing. For example, if the target drawing is a two-dimensional drawing as shown in Figure 2, the similar drawing search unit 115 searches the drawing database for two-dimensional base drawings that have feature information similar to the feature information of the target drawing.

[0060] In addition, the similar drawing search unit 115 may perform a narrowed search using search conditions in the search for the first similar drawing, based on a user instruction (setting of search conditions) via the input interface 15 or the user's terminal device. In this case, the similar drawing search unit 115 further uses search conditions specified by the user to search for multiple first similar drawings. The search conditions may be, for example, drawing information (text, symbols) in two-dimensional drawings. Using these search conditions, the similar drawing search unit 115 searches for multiple first similar drawings that are similar in shape to the target drawing.

[0061] Furthermore, when the target drawing is a three-dimensional drawing as shown in Fig. 3, the similar drawing search unit 115 searches the drawing database for three-dimensional base drawings having feature information similar to the feature information of the target drawing. For example, when the target drawing is a three-dimensional drawing as shown in Fig. 3, the similar drawing search unit 115 searches the drawing database for three-dimensional base drawings having feature information similar to the feature information of the target drawing. Note that the similar drawing search unit 115 may narrow down the search for the first similar drawings in response to a user instruction via the input interface 15 or the user's terminal device.

[0062] In this case, the similar drawing search unit 115 searches for multiple third similar drawings by using additional search conditions specified by the user. The search conditions may be, for example, drawing information (text, symbols) in the three-dimensional drawing. Using these search conditions, the similar drawing search unit 115 searches for multiple first similar drawings that are similar in shape to the target drawing.

[0063] (Step S404) The similar drawing search unit 115 searches the drawing database for a second similar drawing associated (linked) with the first similar drawing based on the drawing information or feature information of the first similar drawing. If the first similar drawing is a two-dimensional drawing, the similar drawing search unit 115 searches the drawing database for, for example, a three-dimensional drawing linked to the drawing information or feature information of the first similar drawing as the second similar drawing. Also, if the first similar drawing is a three-dimensional drawing, the similar drawing search unit 115 searches the drawing database for, for example, a two-dimensional drawing linked to the drawing information or feature information of the first similar drawing as the third similar drawing.

[0064] (Step S405) The system control unit 111 causes the display 17 to display the plurality of first similar drawings. The display 17 may further display the target drawing and / or the search criteria. The plurality of first similar drawings on the display 17 may be displayed, for example, in descending order of similarity and / or descending order of match with the search criteria.

[0065] The display 17 may further display a graphical interface (hereinafter referred to as a display dimension switching GUI (Graphical User Interface)) that allows switching between the first similar drawing and the second similar drawing, a GUI (hereinafter referred to as a size adjustment GUI) that adjusts the size of the displayed drawing as a narrowing-down condition, etc. In response to an operation of the size adjustment GUI 173, the similar drawing search unit 115 changes the size of the drawing displayed on the display 17 and searches for similar drawings again.

[0066] Fig. 5 is a diagram showing an example of the display 17 on which a plurality of first similar drawings SF1 are displayed when the target drawing and the first similar drawings are two-dimensional drawings. As shown in Fig. 5, the display 17 displays the plurality of first similar drawings SF1, a display dimension switching GUI 171, and a size adjustment GUI 173. As shown in Fig. 5, when the plurality of first similar drawings SF1 are two-dimensional drawings, the cursor in the display dimension switching GUI 171 is positioned in 2D.

[0067] 6 is a diagram showing an example of the display 17 on which a plurality of first similar drawings SF1 are displayed when the target drawing and the first similar drawings are three-dimensional drawings. As shown in Fig. 6, the display 17 displays the plurality of first similar drawings SF1, a display dimension switching GUI 171, and a size adjustment GUI 173. As shown in Fig. 6, when the plurality of first similar drawings SF1 are three-dimensional drawings, the cursor in the display dimension switching GUI 171 is positioned in 3D.

[0068] 7 is a diagram showing an example of a display 17 in which a plurality of first similar drawings SF1 are displayed together with a second similar drawing SF2 when the target drawing and the first similar drawings are two-dimensional drawings. As shown in FIG. 7, the display 17 displays the plurality of first similar drawings SF1, a display dimension switching GUI 171, and a size adjustment GUI 173. As shown in FIG. 7, when the plurality of first similar drawings SF1 are two-dimensional drawings, the cursor in the display dimension switching GUI 171 is positioned in 2D. As shown in FIG. 7, the plurality of first similar drawings SF1 are displayed larger than the second similar drawings SF2.

[0069] (Step S406) If an instruction to end the display of similar drawings is input by the user via the input interface 15 or the user's terminal device (Yes in step S406), the similar drawing search process ends. If an instruction to end the display of similar drawings is not input by the user via the input interface 15 or the user's terminal device (No in step S406), the process of step S407 is executed.

[0070] (Step S407) If an instruction to change the dimension of the displayed drawing is not input by a user instruction via the input interface 15 or the user's terminal device (No in step S407), the processing from step S405 onwards is executed. An instruction to change the dimension of the displayed drawing corresponds to, for example, the operation of the display dimension switching GUI 171 in Figures 5 to 7. If an instruction to change the dimension of the displayed drawing is input by a user instruction via the input interface 15 or the user's terminal device (Yes in step S407), the processing of step S408 is executed.

[0071] (Step S408) The system control unit 111 causes the display 17 to display the second similar drawing SF2. As a result, the display 17 displays the second similar drawing SF2. At this time, the display 17 may further display the target drawing and / or the search conditions. The display 17 may also display a display dimension switching GUI 171, a size adjustment GUI 173, etc. The second similar drawing SF2 is displayed on the display 17, for example, in descending order of similarity and / or descending order of match with the search conditions. The display 17 switches between displaying the multiple first similar drawings and the multiple second similar drawings in response to a switching instruction (operation of the display dimension switching GUI).

[0072] 8 is a diagram showing an example of the display 17 on which a plurality of second similar drawings SF2 are displayed when the target drawing and the first similar drawing SF1 are two-dimensional drawings. As shown in Fig. 8, the display 17 displays the plurality of second similar drawings SF2, a display dimension switching GUI 171, and a size adjustment GUI 173. As shown in Fig. 8, when the plurality of second similar drawings SF2 are three-dimensional drawings, the cursor in the display dimension switching GUI 171 is positioned in 3D. The plurality of second similar drawings on the display 17 are displayed, for example, in descending order of similarity.

[0073] 9 is a diagram showing an example of the display 17 on which a plurality of second similar drawings SF2 are displayed when the target drawing and the first similar drawings are three-dimensional drawings. As shown in Fig. 9, the display 17 displays the plurality of second similar drawings SF2, a display dimension switching GUI 171, and a size adjustment GUI 173. As shown in Fig. 9, when the plurality of first similar drawings SF1 are three-dimensional drawings, the cursor in the display dimension switching GUI 171 is positioned in 2D.

[0074] In step S405, when the plurality of first similar drawings SF1 are displayed on the display 17 together with the plurality of second similar drawings SF2, in this step the plurality of second similar drawings SF2 are displayed larger than the first similar drawings SF1.

[0075] (Step S409) If an instruction to end the display of similar drawings is input by the user via the input interface 15 or the user's terminal device (Yes in step S409), the similar drawing search process ends. If an instruction to end the display of similar drawings is not input by the user via the input interface 15 or the user's terminal device (No in step S409), the process of step S410 is executed.

[0076] (Step S410) If an instruction to change the dimension of the displayed drawing is not input by a user instruction via the input interface 15 or the user's terminal device (No in step S410), the processing from step S408 onwards is executed. If an instruction to change the dimension of the displayed drawing is input by a user instruction via the input interface 15 or the user's terminal device (Yes in step S410), the processing from step S405 is executed. At this time, the display 17 switches between displaying multiple first similar drawings and multiple second similar drawings in response to a switching instruction (operation of the display dimension switching GUI). Note that search conditions can be changed at any time during the similar drawing search process. If new search conditions are input or set, the processing from step S403 onwards is repeated. As shown in steps S405 to S410, the display 17 displays multiple first similar drawings and / or multiple second similar drawings.

[0077] As described above, the similar drawing search device 1 according to the first embodiment acquires, from a target drawing, feature information relating to the shape of a target part shown in the target drawing as feature information of the target drawing, searches a drawing database for a plurality of first similar drawings SF1 and a plurality of second similar drawings SF2 showing parts in a dimension different from that of the first similar drawings SF1 based on the feature information of the target drawing, and displays the plurality of first similar drawings SF1 and / or the second similar drawings SF2. Furthermore, the target drawing and the first similar drawings SF1 in the similar drawing search device 1 according to the first embodiment are drawings showing parts in the same dimension, and the similar drawing search device 1 according to the first embodiment switches between displaying the plurality of first similar drawings SF1 and the plurality of second similar drawings SF2 in response to a switching instruction.

[0078] Furthermore, in the similar drawing search device 1 of the first embodiment, a plurality of first similar drawings SF1 and a plurality of second similar drawings SF2 are associated using at least one of the drawing information regarding the parts indicated by each of them and the feature information associated with each of them, and the target drawing and the first similar drawing SF1 are drawings showing parts in the same dimension, and the similar drawing search device 1 of the first embodiment searches for a plurality of first similar drawings SF1 from the drawing database based on the feature information of the target drawing, and searches for a plurality of second similar drawings SF2 from the drawing database using at least one of the drawing information and feature information of the searched plurality of first similar drawings SF1.

[0079] For these reasons, the similar drawing search device 1 according to the first embodiment can simultaneously search for similar 2D drawings, which allow detailed confirmation of welding instructions, precision information, additional information such as whether or not painting is performed, and dimensional information for each part, by inputting a target drawing. Also, similar 3D drawings, the shapes of which are easier for design users and those with relatively little industry experience to intuitively grasp, can be searched for simultaneously by inputting a target drawing. Therefore, the similar drawing search device 1 according to the first embodiment can display similar drawings of the target drawing on the display 17 regardless of the dimension of the target drawing. In other words, the similar drawing search device 1 according to the first embodiment can display similar drawings of the target drawing, such as a first similar drawing of the same dimension as the target drawing and a second similar drawing of a different dimension from the target drawing, on the display 17, either switchably or simultaneously.

[0080] As described above, the similar drawing search device 1 according to the first embodiment can simultaneously search for similar drawings of a target drawing, and can switch between two types of similar drawings, including two-dimensional drawings that are easy to grasp and whose supplementary information can be easily confirmed, even for users with extensive industry experience, and three-dimensional drawings whose shapes are easy to intuitively grasp and determine whether they are the target data, based on the input of the target drawing. Therefore, the similar drawing search device 1 according to the first embodiment allows users to easily search for similar drawings between three-dimensional drawings, such as 3D CAD data, and two-dimensional drawings, and can easily switch between the two types of drawings as desired. As a result, the similar drawing search device 1 according to the first embodiment can improve the efficiency of searching for and checking similar drawings of a target drawing.

[0081] (Modification) As a modification of the first embodiment, both the first trained model and the second trained model may be used. In this case, a two-dimensional target drawing (hereinafter referred to as a 2D target drawing) and a three-dimensional target drawing (hereinafter referred to as a 3D target drawing) are used as the target drawings. That is, the target drawings include a two-dimensional target drawing related to the target component and a three-dimensional target drawing related to the target component. In the following, for the sake of specificity, the first similar drawing will be described as a two-dimensional drawing, and the second similar drawing will be described as a three-dimensional drawing.

[0082] As shown in FIG. 2 , the feature information acquisition unit 113 inputs a 2D target drawing into a first trained model to acquire two-dimensional feature information corresponding to the 2D target drawing (hereinafter referred to as 2D feature information). The 2D feature information may be referred to as first feature information. Furthermore, as shown in FIG. 3 , the feature information acquisition unit 113 inputs a 3D target drawing into a second trained model to acquire three-dimensional feature information corresponding to the 3D target drawing (hereinafter referred to as 3D feature information). The 3D feature information may be referred to as second feature information.

[0083] Based on the 2D feature information, the similar drawing search unit 115 searches the drawing database for a first two-dimensional similar drawing and multiple three-dimensional second similar drawings (hereinafter referred to as linked second similar drawings) that are linked to the first similar drawing and have dimensions different from the first two-dimensional similar drawing. That is, based on the first feature information, the similar drawing search unit 115 searches the drawing database for multiple first two-dimensional similar drawings and multiple second three-dimensional similar drawings (linked second similar drawings). In other words, the similar drawing search unit 115 searches for the first similar drawing using a process similar to that of the first embodiment and searches for linked second similar drawings that are associated with the first similar drawing using drawing information, etc., of the retrieved first similar drawing. The first similar drawing and the linked second similar drawing are linked in advance using drawing information, feature information, etc., of the first similar drawing.

[0084] Furthermore, the similar drawing search unit 115 searches the drawing database for a second similar three-dimensional drawing and multiple first similar two-dimensional drawings (hereinafter referred to as linked first similar drawings) that are linked to the second similar drawing but in a dimension different from the second similar three-dimensional drawing, based on the 3D feature information. That is, the similar drawing search unit 115 searches the drawing database for multiple second similar three-dimensional drawings and multiple first similar two-dimensional drawings (linked first similar drawings), based on the second feature information. In other words, the similar drawing search unit 115 searches for the second similar drawings using a process similar to that of the first embodiment, and searches for linked first similar drawings that are associated with the second similar drawing based on the drawing information of the retrieved second similar drawings, etc. The second similar drawings and the linked first similar drawings are linked in advance based on the drawing information or feature information of the second similar drawings, etc.

[0085] As a result of the above, the similar drawing search unit 115 searches for a pair (hereinafter referred to as the first pair) of a first similar drawing and a linked second similar drawing based on the 2D target drawing, and searches for a pair (hereinafter referred to as the second pair) of a second similar drawing and a linked first similar drawing based on the 3D target drawing.

[0086] In response to a user instruction via the input interface 15 or the terminal device, the system control unit 111 causes at least one of the plurality of first similar drawings, the plurality of linked second similar drawings, the plurality of second similar drawings, and the plurality of linked first similar drawings to be displayed on the display 17. That is, the display 17 displays at least one of the plurality of first similar drawings, the plurality of linked second similar drawings, the plurality of second similar drawings, and the plurality of linked first similar drawings in accordance with the user instruction.

[0087] FIG. 10 is a diagram showing an example of the display 17 displaying a plurality of second similar drawings SF22 from the second pair. As shown in FIG. 10 , a graphical interface (hereinafter referred to as a feature switching GUI (Graphical User Interface)) 175 is displayed, allowing switching between the first pair retrieved based on the feature information of the 2D target drawing and the second pair retrieved based on the feature information of the 3D target drawing. It also displays a display dimension switching GUI 171 and a size adjustment GUI 173. As shown in FIG. 10 , the feature switching GUI 175 is a GUI for selecting either the 2D feature information or the 3D feature information, i.e., the first pair or the second pair, in the search for similar drawings. In FIG. 10 , the second pair is selected (Drawing Feature form = 3D), and the display of the second similar drawing SF22 (Display result in: 3D) is selected.

[0088] As described above, the similar drawing search device 1 according to the modification of the first embodiment acquires first feature information from the two-dimensional target drawing, acquires second feature information from the three-dimensional target drawing, searches the drawing database for a plurality of first similar two-dimensional drawings and a plurality of second similar three-dimensional drawings based on the first feature information, and searches the drawing database for a plurality of second similar three-dimensional drawings and a plurality of first similar two-dimensional drawings based on the second feature information. As a result, the similar drawing search device 1 according to the modification of the first embodiment displays on the display 17 at least one of the plurality of first similar drawings and / or the plurality of second similar drawings, i.e., the second similar drawings linked to the first similar drawings and the first similar drawings linked to the second similar drawings.

[0089] For these reasons, the similar drawing search device 1 according to the modification of the first embodiment can search for a first pair of similar drawings when the target drawing is two-dimensional and a second pair of similar drawings when the target drawing is three-dimensional. Therefore, the similar drawing search device 1 according to the modification of the first embodiment can display the search results for drawings similar to the target drawing by switching the feature information used to search for similar drawings through the operation of the feature switching GUI 175 and the display dimension switching GUI 171 displayed on the display 17.

[0090] That is, according to the similar drawing search device 1 according to a modified example of the first embodiment, by switching between shape similarity based on a 2D target drawing and shape similarity based on a 3D target drawing, similar drawings can be displayed in two dimensions and three dimensions according to the shape complexity of parts in the target drawing and the isomorphism (e.g., number of holes) of parts in the target drawing. Shape complexity corresponds to parts that require time for machining due to multiple machining processes, such as NC (numerical control) machine tools and electrical discharge machining. Specifically, a part has a complex shape when it requires two or more machining processes and a machining time of 20 to 30 hours or more per workpiece. Isomorphism corresponds to homeomorphism, which indicates the number of holes, etc. Shape complexity and isomorphism may be set by specifying search criteria such as the number of bends, hole shape, and number of holes. Therefore, according to the similar drawing search device 1 of the modified example of the first embodiment, it is possible to search for and display a first pair of similar drawings and a second pair of similar drawings according to the feature quantity corresponding to the dimension of the target drawing. Other effects are the same as those of the first embodiment, and therefore will not be described here.

[0091] Second Embodiment In this embodiment, two-dimensional drawings and three-dimensional drawings are used as target drawings. That is, in this embodiment, the target drawings include two-dimensional target drawings related to a target component and three-dimensional target drawings related to the target component. Two-dimensional drawings and three-dimensional drawings are input as target drawings to a trained model used in this embodiment (hereinafter referred to as a third trained model). Since the configuration of this embodiment is the same as that of FIG. 1, functions that differ from the first embodiment will be described.

[0092] The memory 13 stores a third trained model that receives a target drawing as input and determines (outputs) the shape of the target drawing, and feature information output from at least one intermediate layer in the third trained model. The feature information corresponds to an intermediate output of the target drawing output from at least one intermediate layer in the third trained model. The at least one intermediate layer is, for example, the intermediate layer immediately before the output layer in the third trained model, i.e., the final intermediate layer. The third trained model is a model that can output feature information between input and output, and is, for example, a model composed of a multi-layer neural network. The intermediate output is represented, for example, by a feature vector that indicates the features of the shape of the input target drawing.

[0093] The feature information acquisition unit 113 acquires feature information about the shape of a component drawing in a target drawing from the target drawing input by the user via the input interface 15 or the communication interface 19. The feature information acquisition unit 113 associates the acquired feature information with the ID of the target drawing and stores it in the memory 13. For example, the feature information acquisition unit 113 inputs the target drawing to a third trained model and acquires an intermediate output output from at least one intermediate layer in the third trained model as feature information about the shape of the component drawing.

[0094] The image format input to the third trained model is, for example, any two-dimensional drawing such as PNG format or TIFF format, or three-dimensional CAD data. The two-dimensional drawing input to the third trained model may also be a handwritten drawing by a user. The image format is not limited to the above and may also be three-dimensional CAD data.

[0095] Furthermore, the target drawing input to the third trained model may be a part drawing area extracted from the target drawing. In this case, the feature information acquisition unit 113 extracts the part drawing area by performing a predetermined extraction process on the drawing prior to executing the third trained model. This extraction process can utilize various image processing methods, such as existing segmentation processing, and therefore will not be described further. Next, the feature information acquisition unit 113 inputs the part drawing data into the third trained model to acquire an intermediate output related to the target drawing. The intermediate output may be a vector output from one of multiple intermediate layers in the third trained model. However, the intermediate output is preferably an intermediate output input to the output layer (a layer that outputs the shape of the target drawing, etc.) in the third trained model, i.e., an intermediate output from the last intermediate layer immediately preceding the output layer. The intermediate output from the third trained model may be an output from an intermediate layer other than the last intermediate layer, or may be an output from multiple intermediate layers.

[0096] The third trained model is generated by inputting each of the multiple two-dimensional and three-dimensional database drawings stored in the drawing database into a pre-trained model and training the pre-trained model using each of the multiple shapes corresponding to the multiple database drawings as training data. The training data may be, for example, output from the first trained model and the second trained model. The third trained model is generated, for example, by training a multi-layer neural network capable of outputting from an intermediate layer using the database drawings and shapes related to the database drawings as training data. For example, if the training data is a product category of shape, the third trained model is generated by inputting the drawings as images into a convolutional neural network (CNN) and training the CNN to classify the shape classes into, for example, eight classes. Known methods can be applied to generating the third trained model, and therefore, description thereof will be omitted.

[0097] If the target drawing is a two-dimensional drawing or a three-dimensional drawing, either the two-dimensional drawing or the three-dimensional drawing is input to the input layer of the third trained model. At this time, zero padding is performed on the input layer corresponding to the other drawing. Also, if the target drawing is a two-dimensional drawing or a three-dimensional drawing, the first trained model or the second trained model may be used. For the sake of specificity, the target drawing will be described below as including a two-dimensional drawing and a three-dimensional drawing related to the two-dimensional drawing.

[0098] The similar drawing search unit 115 searches the drawing database for a plurality of first similar drawings and a plurality of second similar drawings of a dimension different from that of the first similar drawings, based on the feature information of the target drawing. When the first similar drawings are two-dimensional drawings, the second similar drawings correspond to three-dimensional drawings such as three-dimensional CAD data.

[0099] For example, the similar drawing search unit 115 searches the drawing database for multiple first similar drawings and multiple second similar drawings based on the feature information of the target drawing. Specifically, the similar drawing search unit 115 calculates the distance between the target feature vector of the target drawing and each of multiple feature vectors (hereinafter referred to as search feature vectors) corresponding to multiple database drawings (2D drawings and 3D drawings) stored in the drawing database. The distance is, for example, a Mahalanobis distance, which corresponds to an index (similarity) indicating the similarity between the target drawing and the database drawings. The similar drawing search unit 115 identifies, as multiple first similar drawings and multiple second similar drawings, database drawings corresponding to search feature vectors with distances smaller than a predetermined value among the calculated distances.

[0100] That is, the similar drawing search unit 115 identifies a search feature vector that is equal to or smaller than a predetermined value in a distance space defined by the target feature vector and each of the plurality of search feature vectors. Next, the similar drawing search unit 115 identifies the database drawing corresponding to the identified search feature vector as a first similar drawing in the case of two-dimensional data, or as a second similar drawing in the case of three-dimensional data.

[0101] The configuration of the similar drawing search apparatus 1 according to the second embodiment has been described above. The similar drawing search process according to the second embodiment will be described below. The similar drawing search process will be described below with reference to FIGS. 11 and 12 .

[0102] FIG. 11 is a diagram illustrating an example of an overview of a similar drawing search process using a third trained model. As shown in FIG. 11 , when a two-dimensional drawing (2D drawing) and a three-dimensional drawing (3D drawing) are input as target drawings to the second trained model, feature information is acquired from the target drawings to characterize the target drawings. Also, as shown in FIG. 11 , a first similar three-dimensional drawing and a second similar three-dimensional drawing are identified by searching a drawing database using the feature information. As shown in FIG. 11 , the first similar drawing is displayed in two dimensions on the display 17. Also, as shown in FIG. 11 , the second similar drawing is displayed in three dimensions on the display 17.

[0103] FIG. 12 is a flowchart showing an example of a procedure for similar drawing search processing.

[0104] (Similar drawing search process) (Step S121) The input interface 15 inputs the two-dimensional target drawing and the three-dimensional target drawing in response to a user instruction. The input of the two-dimensional target drawing and the three-dimensional target drawing is not limited to an operation via the input interface 15. For example, in response to a user instruction via a terminal device of a user who can use the similar drawing search process, the two-dimensional target drawing and the three-dimensional target drawing may be transmitted from the terminal device to the similar drawing search device 1. At this time, the communication interface 91 inputs the two-dimensional target drawing and the three-dimensional target drawing to the similar drawing search device 1 via the network 191.

[0105] (Step S122) The feature information acquisition unit 113 inputs the two-dimensional target drawing and the three-dimensional target drawing into the trained model. Next, the feature information acquisition unit 113 acquires intermediate output from at least one intermediate layer in the trained model as feature information. The feature information acquisition unit 113 associates the acquired feature information with the ID of the target drawing and stores it in the memory 13.

[0106] (Step S123) The similar drawing search unit 115 searches the drawing database for a first similar two-dimensional drawing and a second similar three-dimensional drawing based on the feature information of the target drawing. That is, the similar drawing search unit 115 can search for the two-dimensional drawings and the three-dimensional drawings registered in the drawing database without relying on the linkage between the two-dimensional drawings and the three-dimensional drawings.

[0107] Furthermore, in response to a user instruction (setting of search conditions) via the input interface 15 or the user's terminal device, the similar drawing search unit 115 may further execute a refined search using search conditions in the search for the first similar drawings and the second similar drawings. In this case, the similar drawing search unit 115 further uses the search conditions specified by the user to search for multiple first similar drawings and multiple second similar drawings. The search conditions may be, for example, drawing information (text, symbols) in two-dimensional drawings and three-dimensional drawings.

[0108] (Step S124) The system control unit 111 causes the display 17 to display one of the multiple first similar drawings and the multiple second similar drawings. As a result, the display 17 displays one of the multiple first similar drawings and the multiple second similar drawings. At this time, the display 17 may further display the target drawing and / or the search criteria. Furthermore, the multiple first similar drawings and one of the multiple second similar drawings on the display 17 may be displayed on the display 17 in descending order of similarity and / or in descending order of match with the search criteria, for example.

[0109] (Step S125) If an instruction to end the display of similar drawings is input by the user via the input interface 15 or the user's terminal device (Yes in step S125), the similar drawing search process ends. If an instruction to end the display of similar drawings is not input by the user via the input interface 15 or the user's terminal device (No in step S125), the process of step S126 is executed.

[0110] (Step S126) If an instruction to change the display drawing is not input by a user instruction via the input interface 15 or the user's terminal device (No in step S126), the processing from step S125 onwards is executed. An instruction to change the display drawing corresponds to, for example, the operation of the display dimension switching GUI 171 in Figures 5 to 7. If an instruction to change the display drawing is input by a user instruction via the input interface 15 or the user's terminal device (Yes in step S126), the processing of step S127 is executed.

[0111] (Step S127) The system control unit 111 causes the display 17 to display the other of the first similar drawing and the second similar drawing. This causes the display 17 to display the other of the first similar drawing and the second similar drawing. At this time, the display 17 may further display the target drawing and / or the search conditions. The display 17 may also display a display dimension switching GUI 171, a size adjustment GUI 173, and the like.

[0112] (Step S128) If an instruction to end the display of similar drawings is input by the user via the input interface 15 or the user's terminal device (Yes in step S128), the similar drawing search process ends. If an instruction to end the display of similar drawings is not input by the user via the input interface 15 or the user's terminal device (No in step S128), the process of step S129 is executed.

[0113] (Step S129) If an instruction to change the displayed drawing is not input by a user instruction via the input interface 15 or the user's terminal device (No in step S129), the processing from step S128 onwards is executed. If an instruction to change the displayed drawing is input by a user instruction via the input interface 15 or the user's terminal device (Yes in step S129), the processing from step S124 is executed. Note that changes to the search conditions can be input and set at any time during the similar drawing search process. If new search conditions are input and set, the processing from step S123 onwards is repeated.

[0114] From the above, the similar drawing search device 1 according to the second embodiment can search for similar drawings (first similar drawings and second similar drawings) that are similar to two-dimensional and three-dimensional target drawings using a single trained model. As a result, the similar drawing search device 1 according to the second embodiment eliminates the need to link two-dimensional drawings with three-dimensional drawings in the drawing database, thereby improving the versatility of similar drawings. Other effects are similar to those of the first embodiment, and therefore will not be described further.

[0115] When the technical idea according to the first embodiment is realized by a similar drawing search method, the similar drawing search method searches for drawings similar to a target drawing from a drawing database that stores feature information about the shapes of parts shown in each of a plurality of drawings in association with each of the plurality of drawings, and obtains feature information about the shape of the target part shown in the target drawing as feature information for the target drawing. Based on the feature information for the target drawing, the drawing database is searched for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings, and displays the plurality of first similar drawings and / or the plurality of second similar drawings on the display 17. The processing content and effects of each step of the similar drawing search process realized by the similar drawing search method are the same as those in the first embodiment, and therefore description thereof will be omitted.

[0116] When the technical concept of the first embodiment is realized by a similar drawing search program, the similar drawing search program causes a computer to search for drawings similar to a target drawing from a drawing database that stores feature information about the shapes of components shown in each of a plurality of drawings in association with each of the plurality of drawings. The program acquires feature information about the shape of the target component shown in the target drawing as the feature information for the target drawing, searches the drawing database for a plurality of first similar drawings showing the component in a predetermined dimension and a plurality of second similar drawings showing the component in a dimension different from the predetermined dimension of the first similar drawings based on the feature information for the target drawing, and displays the first similar drawings and / or the second similar drawings on a display 17. The program that causes a computer to execute the method can also be stored and distributed on a storage medium such as a magnetic disk (e.g., a hard disk), an optical disk (e.g., a CD-ROM or DVD), or a semiconductor memory. The processing content and effects of each step of the similar drawing search process realized by the similar drawing search program are similar to those of the first embodiment, and therefore will not be described here.

[0117] According to at least one of the embodiments described above, in searching for similar drawings, it is possible to easily search for three-dimensional drawings such as three-dimensional CAD data and two-dimensional drawings.

[0118] (Third Embodiment) This embodiment extends the first and second embodiments in the following respects. Specifically, in this embodiment, the search target is not limited to drawings, but refers to general design data including CAD models and the like, including PDF, PNG, JPEG, 2D CAD files such as DWF, DXF, and DWG, and 3D CAD files such as IGES and STEP. Furthermore, the search method includes not only a similarity search using feature information of objects, but also a method using information such as text extracted from the design data. The following description will focus on functions that differ from the first and second embodiments.

[0119] FIG. 13 illustrates an example of the hardware configuration of a multimodal search and display device 2 according to this embodiment. The multimodal search and display device 2 searches for design data that meets search criteria from a database that stores a correspondence between character information and object feature information in each piece of design data and each piece of design data. That is, the multimodal search and display device 2 multimodally searches for design data that meets specified criteria from a database that stores a correspondence between character information and object feature information contained in multiple pieces of multimodal design data and each piece of design data. The multimodal search and display device 2 may also be implemented as a server device connected to a network 291. In this case, the product in the design data corresponds to, for example, a parts diagram indicating a limited item, such as a metalwork product, an assembly diagram combining the parts diagrams, a circuit diagram, or other component information corresponding to the item. Hereinafter, for ease of explanation, the above-mentioned components will be described as parts diagrams.

[0120] 13, the multimodal search and display device 2 includes a processing circuitry 21, a memory 23, an input interface 25, a display 27, and a communication interface 29. As shown in Fig. 13, the processing circuitry 21, the memory 23, the input interface 25, the display 27, and the communication interface 29 are electrically connected via a bus. Therefore, data is appropriately transmitted and received between the processing circuitry 21, the memory 23, the input interface 25, the display 27, and the communication interface 29 under the control of the system control unit 211.

[0121] The configuration of the device other than the processing circuit 21 is the same as in the first and second embodiments.

[0122] The internal hardware configuration of the processing circuit 21 is also the same as that of the first and second embodiments.

[0123] The processing circuitry 21 includes, depending on the functions to be realized, a system control unit 211, a character information analysis unit 213, a feature information analysis unit 215, a keyword search unit 217, a similarity search unit 219, and a display unit 221. The division of units by function is not limited to the above and can be set as appropriate. The system control unit 211, character information analysis unit 213, feature information analysis unit 215, keyword search unit 217, similarity search unit 219, and display unit 221 that realize various functions by the processing circuitry 21 will be described later.

[0124] The memory 23 is a storage device such as a hard disk drive (HDD), a solid state drive (SSD), or an integrated circuit storage device that stores various information. The memory 23 stores various programs according to this embodiment. For example, the memory 23 stores programs related to the execution of processes in the system control unit 211, the character information analysis unit 113, the feature information analysis unit 215, the keyword search unit 217, the similarity search unit 219, and the display unit 221, which are executed by the processing circuitry 21.

[0125] The memory 23 stores, for example, a plurality of pieces of design data. The memory 23 also stores character and other information included in the plurality of pieces of design data in association with the ID of each piece of design data. The memory 23 also stores feature information regarding objects shown in each piece of design data in association with the ID of each piece of design data. The memory 23 may also store header information regarding the design data in further association with the ID. The header information here includes, for example, a file name and a file creation date. The ID, header information, character and other information, and feature information may be referred to as index data. The index data and design data are stored in the memory 23 as a database regarding the plurality of pieces of design data. That is, the memory 23 stores a database in which a plurality of pieces of design data, header information, character and other information, and feature information are stored in association with each other.

[0126] The database stores multiple different modes of design data representing the same product. These design data are essentially data that should be used in an integrated manner, and are therefore mutually related by one or more elements such as file names, product names, drawing numbers, and part numbers.

[0127] The character information includes at least one of text attached to a component drawing in the design data and symbol information attached to the component drawing. The text is, for example, character information and numerical information attached to the design data. Specifically, the text includes information such as the drawing number, drawing number, product name, material, and processing information (e.g., surface treatment, processing methods such as sheet metal processing, and machining) listed in the title column of the drawing, dimensions and number of holes listed in the component drawing, part number, file name, title, and other information assigned by the user to each drawing. The symbol information attached to the component drawing includes, for example, symbols that abbreviate the text and / or information related to symbols assigned by the user.

[0128] The memory 23 also stores a model for keyword search, which is executed by the keyword search unit 217 and involves searching by specifying a specific word or phrase. This model may use any known method.

[0129] The memory 23 also stores a trained model that is executed by the similarity search unit 219 to search for other design data based on the similarity of objects in specific design data designated by a user, and that receives design data (hereinafter referred to as a target drawing) as an input and determines (outputs) the shape of the target drawing. This trained model uses at least the feature information as input information, and the algorithm may use any known method.

[0130] The hardware configuration of the memory 23 is also the same as in the first and second embodiments.

[0131] The hardware configuration and functions of the input interface 25 are also the same as those in the first and second embodiments.

[0132] The hardware configuration and functions of the display 27 are also the same as those in the first and second embodiments.

[0133] The function of the communication interface 29 is also the same as in the first and second embodiments.

[0134] The function of the system control unit 211 is also the same as in the first and second embodiments.

[0135] The character information analysis unit 213 analyzes character information in the design data to acquire the character information of the design data. For example, the character information analysis unit 213 analyzes the aforementioned text included in the target design data to acquire the character information of the target design data. The method of analyzing the character information varies depending on the mode of the design data. For example, in the case of CAD data, this would involve reading a specific portion of the file contents. In the case of PDF data, this would involve detecting and recognizing characters using a method such as OCR (Optical Character Recognition). Any known method may be used for the OCR. The character information analysis unit 213 associates the acquired character information with the ID of the design data and stores it in the memory 23.

[0136] Furthermore, the character information analyzing unit 213 may structure the analyzed character information and store it in the memory 23 .

[0137] The characteristic information analysis unit 215 analyzes characteristic information of objects in design data to acquire the characteristic information of the design data. For example, the characteristic information analysis unit 215 analyzes characteristic information related to objects in target design data to acquire the characteristic information of the target design data. Specifically, the characteristic information analysis unit 215 acquires characteristic information related to objects in the target design data from the target design data input by the user via the input interface 25 or the communication interface 29. The characteristic information analysis unit 215 associates the acquired characteristic information with the ID of the design data and stores it in the memory 23.

[0138] The mode of data input to the trained model may be, for example, any two-dimensional drawing such as PNG (Portable Network Graphics) format or TIFF (Tagged Image File Format), a hand-drawn drawing by a user, or an image rendered from a three-dimensional CAD drawing (hereinafter referred to as a rendered image). The rendering process may be performed by the feature information analysis unit 215, or may be performed by another unit or an external device via the communication interface 29 and the network 291. The rendered image may be, for example, a three-view drawing. The three-view drawing may be, for example, a plan view, a front view, a side view (right side view), etc. The mode of input data is not limited to the above and may also be three-dimensional CAD data, etc. In this case, different trained models may be prepared depending on the mode of input data, and these are respectively stored in advance in the memory 23.

[0139] Furthermore, the target data input to the trained model may be a part drawing area extracted from the target design data. In this case, the feature information analysis unit 215 extracts the part drawing area by performing a predetermined extraction process on the drawing prior to executing the trained model. Since various types of image processing such as existing segmentation processing can be used for this extraction process, a description thereof will be omitted. Next, the feature information analysis unit 215 inputs the part drawing data into the trained model, analyzes the feature information, and acquires it as feature information of the target data. The selection of the trained model is as described above.

[0140] The operations of the keyword search section 217 and the similarity search section 219 will be described below.

[0141] The keyword search unit 217 performs a multimodal keyword search from the database based on keywords specified by the user. For example, the keyword search unit 217 multimodally searches the database based on a phrase or phrase specified by the user for design data that includes character or other information that at least partially matches the specified phrase or phrase.

[0142] The keyword search may be performed using either a search condition in which multiple keywords are linked with "and" or a search condition in which multiple keywords are linked with "or." For example, when the search conditions "xxxx," "yyyy," and "zzzz" are set, the keyword search unit 217 performs a multimodal search for design data that meets these conditions. Furthermore, when the structured character information described above is stored in the memory 23, a keyword search may be performed by specifying attribute information related to the structuring. For example, when the user sets search conditions such as "SS400" for the material and "within the past year" for the transaction history, the keyword search unit 217 searches the database for design data that meets these conditions.

[0143] As one aspect of keyword search, header information such as an ID may be searched.

[0144] The above keyword search methods may be combined simultaneously or sequentially. For example, an example of the former is a search for "xxx" and transaction history "within the past year," and an example of the latter is a search in which the search results for a keyword search for "xxx" are narrowed down to those with attribute information "SS400."

[0145] The similarity search unit 219 searches the database for similar parts based on the characteristic information of the target design data specified by the user. For example, the similarity search unit 219 searches the database for design data with similar characteristic information in a multimodal manner based on the characteristic information of the design data specified by the user.

[0146] The similarity evaluation method involves calculating the distance between the feature information of the target design data and multiple feature information corresponding to multiple design data stored in the database, and identifying, as multiple similar parts, components in the design data corresponding to feature information for which the distance is smaller than a predetermined criterion. Here, any known algorithm may be selected as the method for calculating the distance between feature information. The predetermined criterion is stored in advance in memory 23.

[0147] The method for implementing multimodal similarity search may be either a stepwise or simultaneous implementation method.

[0148] The stepwise approach uses a trained model trained to perform similarity searches accurately within the same mode or a small number of modes, and a trained model trained to perform similarity searches across multiple modes. For example, a trained model for similarity searches of two-dimensional data such as PDF and DWF (hereinafter referred to as two-dimensional mode), a trained model for similarity searches of three-dimensional data such as STEP (hereinafter referred to as three-dimensional mode), and a trained model for cross-modal searches across all modes are prepared in advance. When a user specifies, for example, PDF design data as target design data, the similarity search unit 219 first performs a similarity search using the trained model for two-dimensional mode. Then, for each of the searched design data, a similarity search of three-dimensional data is performed using a trained model for multimodal search. Conversely, when a user specifies STEP design data as target design data, the similarity search unit 219 first performs a similarity search using the trained model for three-dimensional modal search, and then performs a similarity search of two-dimensional data using the trained model for multimodal search.

[0149] The simultaneous approach uses a trained model for multimodal search to perform a multimodal similarity search of design data, regardless of the mode of design data specified by the user.

[0150] Although the keyword search unit 217 and the similarity search unit 219 have been described separately above, they may be configured to be performed on the same occasion. For example, a refined search may be performed on the results of the similarity search to limit the material to "SS400", or conversely, a similarity search may be performed on the design data resulting from the keyword search.

[0151] The present invention is a device that contributes to user convenience by supporting the multimodal search of design data using a single user interface, and does not exclude implementation means other than those described above that also serve this purpose.

[0152] Next, a method for displaying search results will be described.

[0153] Among the above-described searches, when a similarity search is performed by a stepwise method, the display unit 221 displays search results according to the design data mode specified by the user, and when the user instructs a multimodal search, the display unit 221 displays search results in a mode other than the design data mode specified by the user. For example, when the user specifies design data in two-dimensional data mode, the display unit 221 displays the search results in the two-dimensional mode together with thumbnail images as the first similarity search results, as shown in Fig. 9. Next, when the user performs a screen operation related to the multimodal search, the similarity search unit 219 performs a multimodal search, and the display unit 221 displays the search results in the three-dimensional mode together with thumbnail images, as shown in Fig. 10.

[0154] On the other hand, when a similarity search is performed by a simultaneous means or when a keyword search is performed, the display unit 221 displays thumbnail images of the design data as search results in a multimodal manner, as shown in FIG.

[0155] Furthermore, since the present invention is a device that contributes to user convenience by allowing multimodal searches for design data representing the same product using a single user interface, it is desirable to be able to visually confirm, if there are different modes of design data for each of the search results.

[0156] When a user performs a mode switching operation on at least one thumbnail image on the search result screen, the display unit 221 queries the database for design data of a different mode associated with the design data corresponding to the thumbnail image. If design data of a different mode is available, the thumbnail image switched to the design data of the different mode is displayed on the search result screen. For example, if design data of multiple modes stored in the database indicate the same object, the multiple design data are associated with each other by a unique ID or the like indicating that they are the same object. At this time, the display unit 221 switches and displays each of the search result modes on the search result screen in response to the switching instruction.

[0157] Furthermore, when the user performs an operation to select a specific thumbnail image on the search result screen, the display unit 221 displays an enlarged image of the thumbnail. When the user performs an operation related to mode switching for this image, the display unit 221 queries the database for design data of another mode associated with the design data corresponding to the thumbnail image. If there is design data of another mode, a thumbnail image switched to the design data of the other mode is displayed instead of the enlarged thumbnail image.

[0158] FIG. 14 is a flowchart showing an example of a procedure for similar drawing search processing.

[0159] (Multimodal Search and Display Processing) (Step S201) Target design data is input in response to a user instruction via the input interface 25. The specific method is the same as in the first embodiment, and therefore will not be described in detail.

[0160] (Step S202) The characteristic information analysis unit 215 inputs the target design data input by the user into the trained model. Next, the characteristic information analysis unit 215 analyzes the target design data, acquires characteristic information, associates the acquired characteristic information with the ID of the target design data, and stores the acquired characteristic information in the memory 23.

[0161] (Step S203) The similarity search unit 219 searches the database for similar design data in a multimodal manner based on the feature information of the target design data.

[0162] (Step S204) The display unit 221 displays the search results on the screen together with thumbnail images of the design data.

[0163] (Step S205) If the search for similar drawings is completed (Yes in step S205), the process proceeds to step S209. If the search for similar drawings is not completed (No in step S205), the process proceeds to step S206.

[0164] (Step S206) The user can further narrow down the search results using a keyword search function. In this case, the keyword search unit 217 further narrows down the search results according to keywords specified by the user.

[0165] (Step S207) The display unit 221 displays the refined search results on the screen together with thumbnail images of the design data.

[0166] (Step S208) If the search for similar drawings is completed (Yes in step S208), the process proceeds to step S209. If the search for similar drawings is not completed (No in step S208), the process proceeds to step S206.

[0167] (Step S209) When the user performs an operation to change the mode of the search results on the screen (Yes in step S209), the display unit 221 queries the database for design data in another mode that shows the same part as the design data related to the search results. If there is no instruction to change the mode of the search results (No in step S209), the multimodal search and display process ends.

[0168] (Step S210) If there is design data in another mode (Yes in step S210), the process of step S211 is executed. If there is no design data in another mode (No in step S210), the multimodal search and display process ends.

[0169] (Step S211) If design data of a different mode showing the same part as the design data related to the search result is found in the database, the display unit 221 switches the thumbnail image to the design data of the different mode and displays it on the screen.

[0170] As described above, the multimodal search and display device 2 according to the third embodiment can search for design data multimodally by using the functions of the similarity search unit 219 and the keyword search unit 217 .

[0171] The above flow is an example of a search procedure, and the similarity search and the keyword search may be performed independently, or the order may be reversed.

[0172] Furthermore, the multimodal search and display device 2 according to the third embodiment can switch the mode of the searched design data and check it on the screen.

[0173] For these reasons, the multimodal search and display device 2 according to the third embodiment can search design data across multiple modes in an integrated manner, thereby improving the efficiency of information search. Furthermore, design data for the same part in multiple modes can be viewed on the same screen at the same time, making it possible to integrate and utilize various types of scattered information.

[0174] Although several embodiments have been described in detail above, the disclosure of the embodiments is not limited to the individual embodiments described above. These embodiments can be implemented in various other forms, and various partial omissions, additions, substitutions, modifications, etc. can be made without departing from the spirit of the invention. Furthermore, these embodiments and their modifications are included within the scope and spirit of the invention, as well as within the scope of the invention and its equivalents as set forth in the claims. Furthermore, the order of each process in the embodiments is shown as an example and is not limited to these.

[0175] With respect to the above embodiments, the following supplementary notes are disclosed as one aspect and optional features of the invention. (Supplementary Note 1) A similar drawing search device that searches for drawings similar to a target drawing from a drawing database that stores feature information about the shape of a part shown in each of a plurality of drawings in association with each of the plurality of drawings, the similar drawing search device comprising: a feature information acquisition unit that acquires feature information about the shape of the target part shown in the target drawing as feature information for the target drawing; a similar drawing search unit that searches the drawing database for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings, based on the feature information of the target drawing; and a display unit that displays the plurality of first similar drawings and / or the plurality of second similar drawings. (Supplementary Note 2) The target drawing and the first similar drawings are drawings that show the part in the same dimension, and the display unit may switch between displaying the plurality of first similar drawings and the plurality of second similar drawings in response to a switching instruction. (Supplementary Note 3) The plurality of first similar drawings and the plurality of second similar drawings are associated using at least one of drawing information on the parts indicated by each of the first similar drawings and the feature information associated with each of the first similar drawings, the target drawing and the first similar drawings are drawings showing parts in the same dimension, and the similar drawing search unit may search the plurality of first similar drawings from the drawing database based on the feature information of the target drawings, and search the plurality of second similar drawings from the drawing database using at least one of the drawing information and feature information of the searched plurality of first similar drawings. (Supplementary Note 4) The target drawings may include a two-dimensional target drawing of the target part and a three-dimensional target drawing of the target part.(Supplementary Note 5) The feature information acquisition unit may acquire first feature information from the two-dimensional target drawing, and acquire second feature information from the three-dimensional target drawing, and the similar drawing search unit may search the drawing database for the plurality of two-dimensional first similar drawings and the plurality of three-dimensional second similar drawings based on the first feature information, and search the drawing database for the plurality of three-dimensional second similar drawings and the plurality of two-dimensional first similar drawings based on the second feature information. (Supplementary Note 6) The feature information acquisition unit may input the target drawing to a trained model that uses the target drawing as input and determines the shape of the target drawing, and acquire an intermediate output output from at least one intermediate layer in the trained model as the feature information of the target drawing. (Supplementary Note 7) A similar drawing search method for searching for drawings similar to a target drawing from a drawing database that stores feature information regarding the shape of a part shown in each of a plurality of drawings in correspondence with each of the plurality of drawings, the similar drawing search method comprising: acquiring feature information regarding the shape of the target part shown in the target drawing as feature information for the target drawing; searching the drawing database for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings based on the feature information of the target drawing; and displaying the plurality of first similar drawings and / or the plurality of second similar drawings on a display. (Appendix 8) A similar drawing search program that enables a computer to search for drawings similar to a target drawing from a drawing database that stores characteristic information regarding the shape of a part shown in each of a plurality of drawings in correspondence with each of the plurality of drawings, the similar drawing search program performing the following operations: acquiring characteristic information regarding the shape of a target part shown in the target drawing as characteristic information for the target drawing; searching the drawing database for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings based on the characteristic information of the target drawing; and displaying the plurality of first similar drawings and / or the plurality of second similar drawings on a display.(Supplementary Note 9) A multimodal search device that searches for design data that meets specified conditions in a multimodal manner from a database that stores character information, etc., included in a plurality of multimodal design data, and feature information of objects in association with each of the plurality of design data, the multimodal search display device comprising: a character information analysis unit that analyzes the character information, etc., in the design data to acquire the character information of the design data; a feature information analysis unit that analyzes feature information of objects in the design data to acquire the feature information of the design data; a keyword search unit that multimodally searches, based on a term specified by a user, from the database for design data that includes character information that at least partially matches the specified term; a similarity search unit that multimodally searches, based on the feature information of the design data specified by the user, for design data that has similar feature information; and a display unit that displays a list of the search results in thumbnail format. (Supplementary Note 10) When multiple modes of design data stored in the database indicate the same object, the multiple design data may be associated with each other by a unique ID or the like indicating that they are the same object, and the display unit may switch between the modes of the search results and display them on the search result screen in response to a switching instruction.(Supplementary Note 11) A multimodal search and display method for multimodally searching for design data that meets specified conditions from a database in which character information, etc., included in a plurality of multimodal design data and feature information of objects are stored in association with each of the plurality of design data, the multimodal search and display method comprising: analyzing the character information, etc., in the design data to acquire the character information, etc., of the design data; analyzing feature information of the objects in the design data to acquire the feature information of the design data; multimodally searching, based on a term specified by a user, from the database for design data that includes character information that at least partially matches the specified term; multimodally searching, based on the feature information of the design data specified by the user, for design data that has similar feature information; displaying a list of the search results in thumbnail format; and switching the mode of displaying the search results. (Supplementary Note 12) A multimodal search and display program that runs on a computer and multimodally searches for design data that meets specified conditions from a database that stores character information, etc., included in a plurality of multimodal design data, and feature information of objects, in association with each of the plurality of design data, the multimodal search and display program performing the following operations: analyzing the character information, etc., in the design data to acquire the character information, etc., of the design data; analyzing feature information of objects in the design data to acquire the feature information of the design data; multimodally searching, based on a term specified by a user, from the database for design data that includes character information, etc., at least a portion of which matches the term specified; multimodally searching, based on the feature information of the design data specified by the user, for design data that has similar feature information; displaying a list of the search results in thumbnail format; and switching the mode of displaying the search results.

[0176] 1 Similar drawing search device 11 Processing circuit 13 Memory 15 Input interface 17 Display 19 Communication interface 111 System control unit 113 Feature information acquisition unit 115 Similar drawing search unit 171 Display dimension switching GUI (Graphical User Interface) 173 Size adjustment GUI (Graphical User Interface) 175 Feature switching GUI (Graphical User Interface) 191 Network 2 Multimodal search and display device 21 Processing circuit 23 Memory 25 Input interface 27 Display 29 Communication interface 211 System control unit 213 Character information analysis unit 215 Feature information analysis unit 217 Keyword search unit 219 Similarity search unit 221 Display unit 291 Network

Claims

1. A similar drawing search device that searches for drawings similar to a target drawing from a drawing database that stores characteristic information regarding the shape of a part shown in each of a plurality of drawings in correspondence with each of the plurality of drawings, the similar drawing search device comprising: a characteristic information acquisition unit that acquires characteristic information regarding the shape of the target part shown in the target drawing as characteristic information for the target drawing; a similar drawing search unit that searches the drawing database for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings, based on the characteristic information of the target drawing; and a display unit that displays a plurality of the first similar drawings and / or a plurality of the second similar drawings.

2. The similar drawing search device according to claim 1, wherein the target drawing and the first similar drawing are drawings showing parts in the same dimension, and the display unit switches between displaying a plurality of the first similar drawings and a plurality of the second similar drawings in response to a switching instruction.

3. The similar drawing search device described in claim 1, wherein the plurality of first similar drawings and the plurality of second similar drawings are associated using at least one of drawing information regarding the parts shown in each drawing and the feature information associated with each drawing, the target drawing and the first similar drawing are drawings showing parts in the same dimension, and the similar drawing search unit searches the plurality of first similar drawings from the drawing database based on the feature information of the target drawing, and searches the plurality of second similar drawings from the drawing database using at least one of the drawing information and feature information of the plurality of first similar drawings found.

4. The similar drawing search device according to claim 1, wherein the target drawings include a two-dimensional target drawing related to the target component and a three-dimensional target drawing related to the target component.

5. The similar drawing search device described in claim 4, wherein the feature information acquisition unit acquires first feature information from the two-dimensional target drawing, and acquires second feature information from the three-dimensional target drawing, and the similar drawing search unit searches the drawing database for the multiple two-dimensional first similar drawings and the multiple three-dimensional second similar drawings based on the first feature information, and searches the drawing database for the multiple three-dimensional second similar drawings and the multiple two-dimensional first similar drawings based on the second feature information.

6. A similar drawing search device as described in any one of claims 1 to 5, wherein the feature information acquisition unit inputs the target drawing into a trained model that uses the target drawing as input to determine the shape of the target drawing, and acquires intermediate output output from at least one intermediate layer in the trained model as feature information of the target drawing.

7. A similar drawing search method for searching for drawings similar to a target drawing from a drawing database that stores characteristic information about the shape of a part shown in each of a plurality of drawings in association with each of the plurality of drawings, the similar drawing search method comprising: acquiring characteristic information about the shape of the target part shown in the target drawing as characteristic information for the target drawing; searching the drawing database for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings based on the characteristic information of the target drawing; and displaying the plurality of first similar drawings and / or the plurality of second similar drawings on a display.

8. A similar drawing search program that enables a computer to search for drawings similar to a target drawing from a drawing database that stores characteristic information about the shape of a part shown in each of a plurality of drawings in association with each of the plurality of drawings, the similar drawing search program performing the following operations: acquiring characteristic information about the shape of the target part shown in the target drawing as characteristic information for the target drawing; searching the drawing database for a plurality of first similar drawings that show the part in a predetermined dimension and a plurality of second similar drawings that show the part in a dimension different from the predetermined dimension of the first similar drawings based on the characteristic information of the target drawing; and displaying the plurality of first similar drawings and / or the plurality of second similar drawings on a display.

9. A multimodal search device that searches for design data that meets specified conditions in a multimodal manner from a database that stores character information, etc., contained in a plurality of multimodal design data, and feature information of objects in association with each of the plurality of design data, the multimodal search device comprising: a character information analysis unit that analyzes the character information, etc., in the design data to acquire the character information of the design data; a feature information analysis unit that analyzes feature information of objects in the design data to acquire the feature information of the design data; a keyword search unit that multimodally searches, based on a term specified by a user, from the database for design data that includes character information that at least partially matches the specified term; a similarity search unit that multimodally searches, based on the feature information of the design data specified by the user, for design data that has similar feature information from the database; and a display unit that displays a list of the search results in thumbnail format.

10. A multimodal search and display device as described in claim 9, wherein, when design data of multiple modes stored in the database indicate the same object, the multiple design data are associated with each other by a unique ID or the like indicating that they are the same object, and the display unit switches and displays the modes of the search results on the search result screen in response to a switching instruction.

11. A multimodal search and display method for multimodally searching for design data that meets specified conditions from a database in which character information, etc., and object feature information contained in a plurality of multimodal design data are stored in association with each of the plurality of design data, the method comprising: analyzing the character information, etc., in the design data to acquire the character information, analyzing feature information of the object in the design data to acquire the feature information of the design data; multimodally searching, based on a term specified by a user, from the database for design data that includes character information that at least partially matches the specified term; multimodally searching, based on the feature information of the design data specified by the user, for design data that has similar feature information; displaying the search results in a list in thumbnail format; and switching the display mode of the search results.

12. A multimodal search and display program that runs on a computer and performs a multimodal search for design data that meets specified conditions from a database that stores character information, etc., contained in multiple multimodal design data and feature information of objects in association with each of the multiple pieces of design data, the program performing the following operations: analyzing the character information, etc., in the design data to obtain the character information, analyzing feature information of objects in the design data to obtain the feature information of the design data; multimodally searching, based on a term specified by a user, from the database for design data that includes character information that at least partially matches the specified term; multimodally searching, based on the feature information of the design data specified by the user, for design data with similar feature information; displaying the search results in a list in thumbnail format; and switching the display mode of the search results.

Images