Design document check system and method

Abstract

To provide a design document check system and method which is more user-friendly.SOLUTION: A design document check system 1 for checking a design document by a computer includes a recognition part 30 for recognizing an area of a prescribed color included in a design document 22 acquired as digital data and characters included in the design document, a correspondence determination part 70 for determining a correspondence relation between the recognized area of the prescribed color and the recognized characters, and a result display part 80 for outputting result information including the determined correspondence relation.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a system and method for checking design documents. [Background technology]

[0002] In design work, there is a demand for efficiency in design work by reducing the human load and man-hours required for drawing processing, and for eliminating human error. Therefore, Patent Document 1 describes a "drawing recognition device that makes it possible to determine the correspondence between characters and symbols without requiring knowledge of placement." [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-59670 Summary of the Invention [Problem to be solved by the invention]

[0004] In Patent Document 1, simple symbols and surrounding printed characters are recognized, and the connection relationships between symbols and the correspondence relationships between symbols and characters are estimated based on pre-registered drawing information and the correspondence relationships between simple symbols and characters. Therefore, with the technology in Patent Document 1, it is difficult to check the work done by a designer using paper drawings when the design drawings include complex shapes and handwritten characters in multiple colors.

[0005] Therefore, an object of the present invention is to provide a more user-friendly system and method for checking design documents. [Means for solving the problem]

[0006] In order to solve the above problem, a design document checking system according to one aspect of the present invention is a design document checking system that checks design documents using a computer, and is equipped with a recognition unit that recognizes areas of a predetermined color contained in the design documents acquired as digital data and characters contained in the design documents, a correspondence determination unit that determines the correspondence between the recognized areas of the predetermined color and the recognized characters, and a result display unit that outputs result information including the determined correspondence. [Effects of the Invention]

[0007] According to the present invention, it is possible to determine and output the correspondence between a region of a predetermined color contained in a design document and characters. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a diagram showing the overall configuration of a digital check system for design documents according to this embodiment. [Figure 2] A diagram showing an example of a design document. [Figure 3] FIG. 10 is a diagram showing an example of design document parameters. [Figure 4] FIG. [Figure 5] FIG. 10 is a diagram illustrating attribute region extraction. [Figure 6] FIG. 4 is a diagram illustrating an example of attribute definition information. [Figure 7] FIG. 1 is a diagram illustrating character recognition. [Figure 8] FIG. 10 is a diagram showing an example of color definition information. [Figure 9] FIG. [Figure 10] FIG. 10 is a diagram showing an example of color information estimation. [Figure 11] A diagram showing the training data used in the first machine learning model. [Figure 12] A diagram showing a drawing configuration correspondence table used in the first machine learning model. [Figure 13] FIG. 10 is a diagram illustrating an example of a result display screen. [Figure 14] FIG. 10 is a diagram showing an example of a result display screen displayed on a mobile terminal. [Figure 15]10 is a flowchart showing the processing of the design document checking system. [Figure 16] 10 is a flowchart of a process for detecting a shape. [Figure 17] 10 is a flowchart of a process for recognizing a color. [Figure 18] 10 is a flowchart of a process for estimating character information. [Figure 19] 10 is a flowchart of a process for estimating color information. [Figure 20] This is an overall diagram of the digital checking system for design documents in Example 2. [Figure 21] This is an overall diagram of the digital checking system for design documents in Example 3. [Figure 22] This is an overall diagram of the digital checking system for design documents, relating to Example 4. [Figure 23] 10 is a flowchart showing the processing of the design document checking system according to the fifth embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] The following describes an embodiment of the present invention with reference to the drawings. The design document checking system according to this embodiment is implemented on a computer, and as will be described later, when a user manually checks design documents output on paper, the user's manual check can be confirmed in digital form.

[0010] The design document checking system of this embodiment acquires design documents containing check details manually written by a user (such as marks indicating that numerical values ​​such as dimensions have been checked) as digital data, rechecks the check details within the checking system, and notifies the user of the results. A user is someone who uses the design document checking system, such as a designer. The user who creates the design documents, the user who manually checks the created design documents, and the user who receives the results of the automatic check of the design documents that have been manually checked and entered as digital data may be the same person or may be different people.

[0011] Paper-based design documents that have been manually checked can be digitized using, for example, a scanner, digital camera, or digital video camera, and the digital data can be input into a design document checking system.

[0012] The design document check system 1 of this embodiment can also be expressed as follows, for example.

[0013] An acquisition unit 20 having a digitization unit 21 that digitizes a paper-based design document 111, a reading unit 23 that processes the design document 111 into a recognizable format, and an input unit 24 that accepts design document parameters 12 input by a user (designer); a recognition unit 30 having a figure detection unit 31 that detects a figure area in the design document 111 by receiving image data from the reading unit 23 and design document parameters 12 from the input unit 24 as input; an attribute area extraction unit 32 that extracts an attribute area including attributes that exist around the area detected by the figure detection unit 31; a character recognition unit 34 that recognizes character information of the attribute; a color definition unit 35 that defines arbitrary color information; a color recognition unit 36 ​​that extracts only a specific color portion from the character information recognized by the character recognition unit 34 by receiving image data from the reading unit 23 and design document parameters 12 from the input unit 24 as input; and an information estimation unit 37 that estimates the information recognized by the character recognition unit 34 and the color recognition unit 36; A digital checking system for design documents, comprising: a result display unit 80 that determines whether the check by the user is correct based on the information determined by the information estimation unit 37, and displays the result.

[0014] A machine learning model 62 may be used to estimate color information.

[0015] A machine learning model 63 may be used to determine correspondence.

[0016] Machine learning models 62 and 63 may be used to estimate color information and determine correspondence.

[0017] A function 310 may be provided in which, after the user makes corrections on the result display unit 80, the user inputs the corrections into the check system 1 again to check whether there are any omissions.

[0018] A function 310 may be provided to determine whether the image has been appropriately corrected by extracting the difference between the image before and after the correction.

[0019] A function of notifying the relevant parties of whether or not there have been any revisions and an outline of the revisions (step S16) may also be provided. [Example]

[0020] A first embodiment will be described with reference to Figures 1 to 19. Figure 1 shows the overall configuration of a digital check system 1 for design documents. This system 1 is constructed using computer resources (hardware resources and software resources) of a computer, and includes, for example, an acquisition unit 20 that acquires design document image data 22, a recognition unit 30 that performs processing to recognize information in the design documents, a data unit 40 that stores data used in the system 1, and an inference device storage unit 60 that stores a machine learning model 61 used in machine learning. Furthermore, as will be described later, this system 1 can also include a correspondence determination unit 70 and a result display unit 80.

[0021] The acquisition unit 20 includes, for example, a digitization unit 21 that digitizes paper-based design documents 11, a reading unit 23 that processes the digitized design document image data 22 into a format that can be recognized by a computer, and an input unit 24 that accepts design document parameters 12 input by a designer as a user.

[0022] A scanner may be used for the digitization unit 21. The input information for the system 1 may be a paper design document 11, which is a paper drawing, or design document image data 22 that is stored as electronic data (digital data) from the beginning. The design document image data 22 is stored in advance as electronic data in input image data 41 of the data unit 40.

[0023] The reading unit 23 converts the design document image data 22, which is obtained by digitizing the paper-based design document 1 by the digitization unit 21, or the design document image data 22 stored in the input image data 41, into a format that can be processed on a computer. For this conversion process, a general computer vision library such as OpenCV or Pillow may be used.

[0024] The input unit 24 receives design document parameters 12 input by the designer to the system 1. Examples of the design document parameters 12 will be described later.

[0025] The recognition unit 30 includes, for example, a figure detection unit 31, an attribute region extraction unit 32, an attribute definition unit 33, a character recognition unit 34, a color definition unit 35, a color recognition unit 36, and an information estimation unit 37.

[0026] The figure detection unit 31 has the function of detecting figure areas within the design document image data 22 using the design document image data 22 from the reading unit 23 and the design document parameters 12 from the input unit 24 as inputs. The attribute area extraction unit 32 has the function of extracting attribute areas containing attributes that exist around the areas detected by the figure detection unit 31. The attribute definition unit 33 has the function of defining attribute information. The character recognition unit 34 has the function of recognizing character information within the attribute areas extracted by the attribute area extraction unit 32 by referring to the attribute definition unit 33. The color definition unit 35 has the function of defining arbitrary color information. The color recognition unit 36 ​​has the function of extracting only specific color portions using the design document image data 22 (hereinafter sometimes abbreviated as image data 22) from the reading unit 23 and the design document parameters 12 from the input unit 24 as inputs by referring to the color definition unit 35. The information estimation unit 37 has the function of estimating the information recognized by the character recognition unit 34 and the color recognition unit 36. The information estimation unit 37 includes a color information estimation unit 38 that estimates color information, and a character information estimation unit 39 that estimates character information.

[0027] As described above, the figure detection unit 31 receives the image data 22 from the reading unit 23 and the design document parameters 12 from the input unit 24 as input, and uses the first machine learning model 61 described later to recognize figures in the image data 22 of the design document and detect figure areas.

[0028] The attribute area extraction unit 32 cuts out a rectangular area by multiplying each of the long and short sides of the rectangle of the figure area detected by the figure detection unit 31 by a certain coefficient while keeping the center position unchanged. This coefficient uses statistical values ​​obtained by collecting design document image data in advance.

[0029] The character recognition unit 34 recognizes characters in the region extracted by the attribute region extraction unit 32. If the attribute is typed, it applies printed character recognition, and if it is handwritten, it applies handwritten character recognition, thereby recognizing the content written in the image data 22. For character recognition, an AI-OCR tool such as LINE Clova or Smart Read can be used, for example.

[0030] The color definition unit 35 defines any color information in advance. For the colors used in the image data 22 of the design document, the colors are defined by numerical values ​​that can be recognized by a computer vision library, such as the RBG color space or the HSV color space.

[0031] The color recognition unit 36 ​​receives the image data 22 from the reading unit 23 and the design document parameters 12 from the input unit 24 and detects areas where any color defined by the color definition unit 35 exists. Any one color, multiple colors, or all colors included in the any color are considered to be "predetermined colors." For color recognition, a computer vision library such as OpenCV or Pillow may be used.

[0032] The information estimation unit 37 can include a color information estimation unit 38 that estimates color information from the recognition result of the color recognition unit 36, and a character information estimation unit 39 that estimates character information from the recognition result of the character recognition unit 34.

[0033] The color information estimation unit 38 refers to the color definition unit 35 for the area recognized by the color recognition unit 36, and estimates what information the area of ​​that color represents.

[0034] The character information estimation unit 39 refers to the attribute definition unit 33 for the character information recognized by the character recognition unit 34, and estimates what attribute the character information has.

[0035] The data unit 40 includes input image data 41 and learning image data 42. The input image data 41 stores design document image data 22 that has already been digitized or created digitally. The learning image data 42 stores image data of design documents for learning to be used in machine learning performed by the learning unit 50, which will be described later.

[0036] The inference unit storage unit 60 includes a first machine learning model 61. As will be described in an embodiment below, the inference unit storage unit 60 can store at least one of a second machine learning model 62 and a third machine learning model 63 in addition to the first machine learning model 61. By using these multiple machine learning models, it is possible to improve the recognition accuracy or estimation accuracy in the attribute region extraction unit 32, the color information estimation unit 38, the character information estimation unit 39, etc.

[0037] The first machine learning model 61 is a machine learning model for recognizing a graphic area in the image data 22 of a design document, such as a projection drawing of a 3D CAD model. To apply such a technique called object detection, for example, a library such as OpenCV or Yolo may be used.

[0038] The correspondence determination unit 70 determines the correspondence between the attribute areas and their contents recognized as attributes by the information estimation unit 37 and the areas and information extracted by color. For example, the correspondence determination unit 70 determines whether any checks have been missed based on the overlap rate and coordinate distance between dimension areas, which are attributes in design drawings, and marker check areas, where dimensions are hand-marked using a yellow marker, for example. This determination can be made, for example, by using a statistical method to determine a threshold value or by creating a model of the relationship between each area using machine learning. Note that the correspondence between dimension areas and check areas may vary depending on the designer, so a machine learning model may be created for each designer.

[0039] The result display unit 80 displays the results determined by the correspondence determination unit 70 to the designer. Here, for example, if there is an oversight in the check, a function may be considered in which the designer makes corrections and then re-enters the data into the system to check whether there are any oversights, and a function 310 may be considered in which the designer extracts the difference between the original image and the corrected image, checks whether any parts that should not be changed have been changed, and determines whether the corrections have been made appropriately. Furthermore, as in other embodiments described later, it may be considered to notify the relevant parties when a correction instruction is given by the person in charge of checking, and when the correction has been completed.

[0040] As described above, the correction part determination unit 310 has the function of comparing the design document before correction with the design document after correction and making a determination. The user checks the design document output on paper, and the checked design document is read by the acquisition unit 20 and input into the recognition unit 30. The correction part determination unit 310 of the recognition unit 30 compares the design document before correction with the checked design document (the corrected design document) based on information (ID) that uniquely identifies the design document, and determines whether any checks have been missed. The determination result is output from the result display unit 80.

[0041] An example of the target design document 11 will be described using FIG. 2. Here, as an example of the design document, a design drawing for a component in a structural design is shown. The design document 11 includes, as shown in FIG. 2, a 3D CAD diagram 111 of the structure, a table 112, typed characters 113 or handwritten characters 114 for corrections and communication, and handwritten symbols 115. The characters 113, 114, and symbols 115 may include colors other than black. The shaded area 116 represents a color marker, such as yellow. A user reviewing the design document visually checks each character (e.g., dimensions) and indicates that the character has been checked by using the shaded area 116 of the color marker. As will be described later, the meanings of the colors used in the design document can be defined in advance. In the figure, the shaded area 116 and the character 113 are shown separated from each other, but in reality, the shaded area 116 is often written overlapping the character 113. The design document (design drawing) 11 shown in FIG. 2 will be described below with reference to FIGS. 3 to 12.

[0042] An example of the design document parameters 12 input by the designer will be described with reference to Fig. 3. The design document parameters 12 include, for example, a drawing number 121, a drawing category 122, a drawing size 123, a color ID 124, and a color use 125.

[0043] The drawing number 121 is information that uniquely identifies a drawing as the design document 11. The drawing category 122 indicates the type of drawing, such as a front view, parts drawing, wiring diagram, or assembly drawing. The color ID 124 is information that identifies the color used in the drawing. The color use 125 indicates the use of the identified color. Examples of color uses include checking, correcting, and confirming correspondence.

[0044] The design document parameters 12 are tabular data with the drawing number as the primary key so that they can be applied to multiple sheets of design document data (design document image data) by simply entering them once. The design document parameters 12 can be, for example, spreadsheet software, RPA (Robotic Process Automation) tools, GUI (Graphical User Interface) applications, etc.

[0045] An example of detecting a figure will be described using Figure 4. For a complex figure 111, such as a projection of a 3D CAD model, included in a design document, the coordinates and class (component name) of a rectangle 311 are estimated by the first machine learning model 61 shown in Figure 1. In other words, the first machine learning model 61 detects the rectangle 311 that circumscribes the figure 111.

[0046] An example of extracting an attribute region will be described using FIG. 5. For each region such as rectangle 311 detected by figure detection unit 31, attribute region extraction unit 32 extracts rectangle 321 larger than rectangle 311 by multiplying the short and long sides by a certain coefficient. Rectangle 321 indicating an attribute region is a figure region containing attributes. Here, attributes refer to typed characters 113 corresponding to dimensions in design documents, etc. Values ​​calculated using statistical techniques or machine learning can be used as the coefficients by which the short and long sides of rectangle 311 are multiplied.

[0047] An example of attribute definition information 330 will be described using FIG. 6. Most of the type characters in design documents are dimensions, but they may also contain specific symbol characters other than dimensions. For this reason, the attribute definition unit 31 prepares two definition tables in advance: a dimension definition table 331 and a specific symbol character definition table 332. Dimensions consist of only numeric values, or specific initial characters and numeric values. Specific symbol characters refer to character strings defined, for example, by JIS or the department or organization to which the user belongs.

[0048] The dimension definition table 331 includes, for example, a dimension ID 3311, a content 3312, a data type 3313, a definition by regular expression 3314, and an example 3315. For example, the dimension, diameter, R chamfer, screw diameter, etc. are set in the content 3312. The specific symbol character definition table 332 includes, for example, a character ID 3321, a name 3322, a data type 3323, and a value 3324. For example, a scale, a disconnector, a circuit breaker, a relay, etc. are set in the name 3322. The content defined in the definition tables 331 and 332 varies depending on the type of design document, etc.

[0049] An example of character recognition will be described using Figure 7. Character recognition unit 34 performs character recognition processing on rectangular area 321 (Figure 5) extracted by attribute area extraction unit 32, and recognizes the typed characters included in rectangular area 321. In general character recognition, in order to detect what is written in which area, rectangular area 341 surrounding the characters and text information are recognized. In order to recognize all characters within specified area 341, character recognition unit 34 recognizes not only the dimensions of typed characters 113 surrounded by rectangle 341, but also information other than attributes, such as character strings within rectangle 342 that surrounds the dimensions.

[0050] An example of color definition information 351 will be described using FIG. 8. Color definition information 351 defines a color using, for example, color ID 3511, color name 3512, and numerical data 3513 to 3518. Numerical data 3513 to 3518 is expressed as three-dimensional data such as RGB or HSV, which is data that can be used in a computer vision library. In color recognition, when an object is handwritten, this numerical data will not be constant depending on the pen used and the shading at the time of scanning, so minimum and maximum values ​​are defined. In the example of FIG. 8, the minimum and maximum values ​​of the color information are defined using HSV values. The process of color recognition will be described using FIG. 9. The color recognition unit 36 ​​uses the color recognition function of the computer vision library to recognize a specific color area 362 in the drawing by referencing the color information defined by the color definition unit 35 for the color ID included in the design document parameters 12 input from the input unit 24. In FIG. 9, the color recognition unit 36 ​​recognizes an area 362 checked with a yellow marker as an example. For a yellow marker area such as the shaded area 361, the color recognition unit 36 ​​recognizes a rectangular checked area 362 according to the definition in the first line of the design document parameters 12 shown in FIG. 3, for example.

[0051] An example of estimating color information will be described using FIG. 10. The color information estimation unit 38 estimates what each color region 362 in the drawing recognized by the color recognition unit 36 ​​represents. In the example shown at the top of FIG. 10, the color information estimation unit 38 detects a yellow marker. The color information estimation unit 38 estimates a group of color regions 381 based on the area calculated by setting a distance at which a point cloud is considered to be a group based on the color recognition results of the computer vision library.

[0052] The dots in the figure are pixels extracted by color recognition. If the same color ID exists for another purpose, such as handwritten text, or if noise such as dust is present during scanning, accuracy may be reduced when judging based on area alone. In such cases, as shown in the lower part of Figure 10, a two-dimensional histogram 382 can be created from the coordinate data of the area detected by color recognition, and a machine learning model can be created for its distribution to make a judgment.

[0053] The first machine learning model 61 will be described using FIG. 11. The first machine learning model 61 is an object detection model that detects specific shapes from image data 22 of design documents. Therefore, image data containing the shapes to be detected is prepared as training data. Here, the configuration of training data 611 when Yolo5, a common object detection algorithm, is used will be described.

[0054] The training data 611 includes, for example, a component element 6111, a training image 6112, and coordinates 6113 to 6116 of a rectangular area circumscribing the image.

[0055] That is, the training data 611 includes image data 6112 containing the desired figure to be detected, and coordinate data 6113 to 6116, each of which indicates by a rectangle where the image data 6112 of each figure is located in the design document image data 22. To improve the accuracy of figure detection, drawing components 6111 are used as class names for the training data 611. Here, the component 6111 refers to the type of figure and refers to a medium classification of the drawing category 122 (FIG. 3), which is a major classification of the design document. This makes it possible to compare the drawing category 122 described in the design document parameters 12 with the figure component 6111 detected by object detection, and to eliminate components that cannot exist in the drawing category.

[0056] 12, a drawing configuration correspondence table 612 is prepared to compare drawing divisions 122 with components 6111. The drawing configuration correspondence table 612 includes, for example, drawing divisions 6121, components 6122, and types 6123.

[0057] An example of the result display screen 81 will be described with reference to Fig. 13. The result display screen 81 provides the determination result of the correspondence determination unit 70 to a user such as a designer. The result display screen 81 may not only be displayed on a monitor display, but may also be printed out on a printer. An email containing the contents of the result display screen 81 may also be sent.

[0058] The result display screen 81 includes, for example, a result detail display section 811 that shows the details of the results, an operation section 812 that is operated by the user, a thumbnail display section 813 that displays a thumbnail of the relevant drawing, and a display section 814 that displays a thumbnail of the relevant location.

[0059] The detailed result display section 811 includes, for example, the judgment result, details of the judgment result, information on the target drawing, and related information. The judgment result may be described as, for example, "missing check." The details of the judgment result may include, for example, the number of judged locations and the judged text. The information on the target drawing may include, for example, the drawing number and the components. The related information may include, for example, the number of the related drawing.

[0060] Reference numeral 815 in Fig. 13 indicates a shape in which a dimension determined to be "omitted from check" was detected. Reference numeral 816 in Fig. 13 indicates a dimension "393.2" determined to be "omitted from check." Because the dimension "393.2" is not marked with a marker of a predetermined color, the correspondence determination unit 70 determines that the value of the dimension "393.2" has not been checked.

[0061] As shown in FIG. 14, the design document checking system 1 can also be completed on a mobile terminal 9. The functional configuration of the design document checking system 1 shown in FIG. 1 can be realized using the computer resources (hardware resources and software resources) of the mobile terminal 9. The acquisition unit 20 can capture design document image data 22 by photographing a paper-based design document with a camera (not shown) provided in the mobile terminal 9. The result display unit 80 can display the check result 81A on the screen of the mobile terminal 9, as shown in FIG. 14. The user can also save the check result (for example, areas determined to be errors, etc.).

[0062] The mobile terminal 9 may use external computational resources and external storage resources. For example, the data unit 40, the learning unit 50, and the inference unit storage unit 60 may be provided in a computer (not shown) external to the mobile terminal 9, and data may be transmitted and received by communication between the mobile terminal 9 and the external computer.

[0063] The overall processing of the design document digital check system 1 will be described with reference to Fig. 15. The system 1 acquires design document image data 22 by the acquisition unit 22 (S1), and further acquires design document parameters by the acquisition unit 22 (S2).

[0064] The system 1 detects figures based on the two pieces of acquired data (S3), and performs the following for each detected figure (S4 to S7).

[0065] System 1 extracts attribute regions for the detected figures (S5) and recognizes characters within the extracted attribute regions (S6). After completing character recognition for all extracted figures, System 1 estimates character information (S8). System 1 branches the subsequent processing depending on the presence or absence of attribute information as a result of estimating the character information (S9).

[0066] If there is no attribute information (S9: NO), the system 1 displays the determination result and accepts corrections (S15).

[0067] If there is one or more pieces of attribute information (S9: YES), the system 1 reads color definition information (S10) and recognizes the colors in the design document image data 22 (S11). The system 1 estimates color information (S12) and branches the next process depending on whether or not a color area exists (S13). If the target color area is not found (S13: NO), the system 1 proceeds to displaying the determination result and accepting corrections (S15). If a color area exists (S13: YES), the system 1 determines the correspondence between the character information and color information (S14) and proceeds to step S15.

[0068] The processing executed by the figure detection unit 31 will be described with reference to Fig. 16. The figure detection unit 31 acquires the design document image data 22 and the design document parameters 12 (S301, S302). The figure detection unit 31 detects figures from the design document image data 22 (S303). The figure detection unit 31 checks the validity of the detected figures by referring to the design document parameters 12. Then, the figure detection unit 31 calculates a rectangular area circumscribing the detected figure (S304) and outputs the coordinates of the rectangular area (S305).

[0069] 17, the color recognition process executed by the color recognition unit 36 ​​will be described. The color recognition unit 36 ​​acquires the design document image data 22 and the design document parameters 12 (S111, S112). The color recognition unit 36 ​​acquires color definition information 351 and acquires numerical range information of the color to be recognized (S113).

[0070] The color recognition unit 36 ​​sets the minimum value of the area of ​​the recognized region (S114). The color recognition unit 36 ​​recognizes any color selected as a "predetermined color" from the design document image data 22 and creates a mask of the recognized color region (S115). The color recognition unit 36 ​​calculates the coordinates of the created mask region (S116).

[0071] 18, the character information estimation process executed by character information estimation unit 39 will be described. Character information estimation unit 39 acquires the character recognition result from character recognition unit 34 (S801), and then reads attribute definition information (S802).

[0072] The character information estimation unit 39 determines whether the character string contains only numerical values ​​from the character recognition result (S803), and if it contains only numerical values ​​(S803: YES), determines that the character string is a dimension (S804). The character information estimation unit 39 records the attribute area and ends the process (S809).

[0073] If the character recognition result includes non-numeric characters (S803: NO), character information estimation unit 39 counts the number of non-numeric characters (S805). Character information estimation unit 39 determines whether the non-numeric character is at the beginning and there is only one (S806). If the condition is met (S806: YES), character information estimation unit 39 refers to dimension definition table 331, searches for a match (S807), records the attribute area, and ends the process (S809).

[0074] If the condition in step S806 is not met (S806: NO), the character information estimation unit 39 searches for a match by referring to the specific symbol character definition table 332 (S808). In either case, if a match is found as a result of the search, the attribute area is recorded (S809) and the process ends.

[0075] The color information estimation process executed by the color information estimation unit 38 will be described with reference to Fig. 19. The color information estimation unit 38 acquires a color recognition result (S121), and further acquires the design document parameters 12 (S122).

[0076] The color information estimation unit 38 determines whether the same color ID exists in multiple columns in the design document parameters (S123). If the same color ID does not exist in multiple columns (S123: NO), the color information estimation unit 38 calculates the rectangle of the area (S126), records the color area, and ends the process (S127).

[0077] If the same color ID exists in multiple columns (S123: YES), the color information estimation unit 38 calculates the area of ​​the recognized color region (S124) and extracts regions whose calculated area is equal to or greater than a threshold (S125). The color information estimation unit 38 calculates rectangular regions for only the extracted regions (S126), records the color regions, and ends the process (S127).

[0078] According to this embodiment configured as described above, when a user manually checks a design document, the user can confirm the manual check on the digital data.

[0079] According to this embodiment, it is possible to detect the shapes contained in the design documents, the dimension values ​​associated with the shapes, and areas of a specified color, and determine whether or not a check is required based on the correspondence between the dimension values ​​and the specified color, thereby reducing the user's workload and improving usability.

[0080] According to this embodiment, the correspondence between color areas and character areas can be provided to the user from the result display unit 80, and the user can recheck the design documents and input them into the design document checking system 1. Therefore, mistakes such as missed checks can be efficiently corrected, improving user usability.

[0081] According to this embodiment, the user can modify the design documents via the result display unit 80, which further improves usability.

[0082] In this embodiment, the correction part determination unit 310 can compare the design documents to be checked with the design documents after digital check by the system 1, making it possible to determine whether any checks have been missed, thereby maintaining the reliability of the design documents. [Example]

[0083] Example 2 will be described with reference to Fig. 20. In the following examples including this example, differences from Example 1 will be mainly described. In the design document checking system 1A of Example 2, the color information estimation unit 38 uses the second machine learning model 62 to estimate recognized color information.

[0084] 20 is an overall view of a design document checking system 1A. An inference unit storage unit 60A includes a first machine learning model 61 and a second machine learning model 62. The second machine learning model 62 is a model for recognizing color information.

[0085] This embodiment configured as described above also achieves the same effects as those of Embodiment 1. Furthermore, in this embodiment, the color information is recognized and estimated using the second machine learning model 62, which can further improve accuracy. [Example]

[0086] Example 3 will be described using the overall system diagram of Fig. 21. In the design document checking system 1B of Example 3, a first machine learning model 61 and a third machine learning model 63 are stored in an inference unit storage unit 60B. A correspondence determination unit 70 determines the correspondence using the third machine learning model 63 for determining the correspondence.

[0087] The present embodiment configured in this manner also achieves the same effects as those of Example 1. Furthermore, in this embodiment, the correspondence between the characters and the predetermined colors is determined using the third machine learning model 63, so that the determination accuracy can be improved compared to Example 1. [Example]

[0088] Example 4 will be described using the overall system diagram of Fig. 22. In a design document checking system 1C according to Example 4, an inference unit storage unit 60C includes a first machine learning model 61, a second machine learning model 62, and a third machine learning model 63. A color information estimation unit 38 uses the second machine learning model 62 for color recognition and estimation. A correspondence determination unit 70 uses the third machine learning model 63 to determine the correspondence between characters and colors.

[0089] The present embodiment configured in this manner also achieves the same effects as those of Embodiment 1. In the present embodiment, the color information estimation unit 38 uses the second machine learning model 62, and the correspondence determination unit 70 uses the third machine learning model, so that the recognition accuracy and determination accuracy can be improved compared to Embodiment 1. [Example]

[0090] A fifth embodiment will be described with reference to Figure 23. Figure 23 shows the processing of the design document checking system 1D. In this processing, after step S15, the system 1D notifies the relevant parties of predetermined information (S16). The system 1D notifies the computer terminals used by the pre-registered relevant parties by e-mail or the like.

[0091] The predetermined information may include, for example, information indicating corrections to the design documents. The information indicating corrections may include, for example, any one or more or all of information identifying the drawings to be checked, information indicating the determination result of a check omission or the like, information indicating that the check omission has been corrected, information identifying the user who performed the check, information indicating that the recheck by the system 1D has been completed, and the like.

[0092] This embodiment configured as described above also has the same effects as those of embodiment 1. Furthermore, in this embodiment, the determination results of the system 1D can be notified to the relevant parties, thereby further improving user convenience.

[0093] It should be noted that the present invention is not limited to the above-described embodiments. Those skilled in the art can make various additions and modifications within the scope of the present invention. The above-described embodiments are not limited to the configuration examples shown in the accompanying drawings. The configurations and processing methods of the embodiments can be modified as appropriate within the scope of achieving the object of the present invention.

[0094] Furthermore, the components of the present invention can be selected arbitrarily, and the invention including the selected components is also included in the present invention. Furthermore, the components described in the claims can be combined in combinations other than those explicitly stated in the claims. [Explanation of symbols]

[0095] 1, 1A, 1B, 1C, 1D: design document checking system, 11: design document, 12: design document parameters, 20: acquisition unit, 21: digitization unit, 22: design document image data, 23: reading unit, 24: input unit, 30: recognition unit, 31: figure detection unit, 32: attribute area extraction unit, 33: attribute definition unit, 34: character recognition unit, 35: color definition unit, 36: color recognition unit, 37: information estimation unit, 38: color information estimation unit, 39: character information estimation unit, 40: data unit, 41: input image data, 32: learning image data, 50: learning unit, 60: inference device storage unit, 61: first machine learning model, 62: second machine learning model, 63: third machine learning model, 70: correspondence determination unit, 80: result display unit, 310: correction location determination unit

Claims

1. A design document checking system that checks design documents using a computer, a recognition unit that recognizes an area of ​​a predetermined color included in a design document acquired as digital data and characters included in the design document; a correspondence determination unit that determines a correspondence relationship between the recognized area of ​​the predetermined color and the recognized character; a result display unit that outputs result information including the determined correspondence relationship; A design document checking system that includes:

2. The recognition unit recognizes a figure included in the design document, extracts a dimension area including a dimension value associated with the recognized figure, and recognizes the area of ​​the predetermined color and the character in the extracted dimension area. The design document checking system according to claim 1.

3. The recognition unit recognizes the area of ​​the predetermined color and the character based on design document parameters including at least the type of the design document and the meaning of the predetermined color. The design document checking system according to claim 2.

4. The result display unit accepts corrections to the design documents. A design document checking system according to any one of claims 1 to 3.

5. The recognition unit includes a correction part determination unit that compares the design document before correction with the design document after correction and makes a determination. The design document checking system according to claim 4.

6. The recognition unit recognizes a figure included in the design document using a first machine learning model for recognizing a figure. The design document checking system according to claim 2.

7. The recognition unit further recognizes the region of the predetermined color using a second machine learning model for recognizing color information. The design document checking system according to claim 6.

8. The correspondence determination unit determines the correspondence using a third machine learning model for determining the correspondence. The design document checking system according to claim 6.

9. the recognition unit further recognizes the region of the predetermined color using a second machine learning model for recognizing color information; The correspondence determination unit determines the correspondence using a third machine learning model for determining the correspondence. The design document checking system according to claim 6.

10. Predetermined information including information indicating the correction to the design document is notified to a pre-registered computer terminal. The design document checking system according to claim 4.

11. The design document includes at least one of color information and text added by a user, and is acquired by the recognition unit as the digital data. A design document checking system according to any one of claims 1 to 3.

12. A method for checking design documents by using a computer, comprising: a recognition step of recognizing an area of ​​a predetermined color included in a design document acquired as digital data and characters included in the design document; a correspondence determination step of determining a correspondence relationship between the recognized predetermined color area and the recognized character; a result display step of outputting result information including the determined correspondence; A method for checking design documents by causing the computer to execute the above.

Images