JP2023055219A - Information processing device, information processing method and program - Google Patents

Abstract

To enable an efficient search of a drawing similar to an objective drawing (a predetermined drawing) among already-presenting drawings.SOLUTION: An information processing system that compares an objective drawing which is a drawing to be utilized by a user with a comparison drawing which has one or more regions to be compared with the objective drawing and set as comparison regions in advance includes a region designating unit 42 and a similar drawing searching unit 45. The region designating unit 42 designates, as a first comparison region that is a region (the region of a main line) to be compared with already-presenting drawings, the one ore more comparison regions in the objective drawing. The similar drawing searching unit 45 compares the one or more first comparison regions (the regions of the main line) of the objective drawing with one or more second comparison regions (the regions of the main line) of the already-presenting drawing, respectively, thereby calculating a similarity between the objective drawing and the comparison drawing.SELECTED DRAWING: Figure 4

Description

The present invention relates to an information processing device, an information processing method, and a program.

For example, a company that manufactures industrial machinery and some of its products creates a draft drawing of the product based on the specifications of the product received from the customer (customer), and prepares an estimate for the product before receiving the order. There is work to send to the supplier.

Japanese Patent Application Laid-Open No. 2021-51469

In the process of estimating as described above, the company accumulates tens of thousands of drawings for estimating goods, and it is conceivable that the drawings may be reused for estimating similar goods.
It is difficult even for veteran employees to find the right one among the tens of thousands of drawings. Since veterans make estimates based on past experience, there is concern about personal risk.
In addition, there are cases where the specification sheet for the product arrives from the customer as a 2D drawing (plan view) instead of a 3D drawing (stereoscopic view). There is also a need to match
SUMMARY OF THE INVENTION It is an object of the present invention to efficiently search existing drawings for drawings similar to a predetermined drawing.

In order to achieve the above object, an information processing device according to one aspect of the present invention includes:
In an information processing system for comparing a drawing to be used by a user as a target drawing and comparing the target drawing with one or more comparative drawings in which one or more regions to be compared are set in advance as comparison regions,
area designating means for designating one or more comparison areas in the target drawing as areas to be compared with the comparison drawing as comparison areas;
a similarity calculation means for calculating a similarity between the target drawing and the comparison drawing by comparing the one or more comparison regions of the target drawing and the one or more comparison regions of the comparison drawing;
Prepare.

According to the present invention, it is possible to efficiently search existing drawings for drawings similar to a predetermined drawing.

1 is a diagram showing a schematic configuration of an information processing system to which a server according to an embodiment of an information processing apparatus of the present invention is applied; FIG. 2 is a diagram showing an example of a network configuration of the information processing system of FIG. 1; FIG. 3 is a block diagram showing an example of a hardware configuration of a server in the information processing system shown in FIGS. 1 and 2; FIG. FIG. 4 is a functional block diagram showing an example of a functional configuration for realizing a similar drawing search process among the processes whose execution is controlled by the server of FIG. 3; FIG. 5 is a flow chart showing operation of a process related to similar drawing search of the server of FIG. 4; FIG. FIG. 10 is a diagram showing a state in which a graphic area (comparison area) to be compared is specified in the target drawing; FIG. 10 is a diagram showing how feature points are extracted from a designated comparison region; It is a figure which shows an example of the incidental information added to search conditions. FIG. 10 is a diagram showing a comparison example of histograms generated from feature points and incidental information; FIG. 10 is a diagram showing a list screen of drawings searched according to similarity; FIG. 10 is a diagram showing how an AI model is tuned based on a user's evaluation of search results for similar drawings. It is a figure which shows an example of the CAD drawing as a target drawing. It is a figure which shows the information of an estimate master, and the flow of automatic accumulation. It is a figure which shows an example of the assembly product (apparatus) comprised from several components. FIG. 15 is a diagram showing the search operation for similar drawings in the case where the apparatus is composed of units and parts as in FIG. 14; It is a figure which shows a mode that similar goods are searched for by the part unit which comprises a search device, and an estimated price is output.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the drawings.
A schematic configuration of an information processing system to which a server according to an embodiment of the information processing apparatus of the present invention is applied will be described with reference to FIG.
FIG. 1 is a diagram showing a schematic configuration of an information processing system to which a server according to an embodiment of an information processing apparatus of the present invention is applied.

As shown in FIG. 1, the information processing system of the embodiment creates an estimate sheet using an estimate system 4 and sends it to the customer terminal 6 before accepting an order for an article ordered by the customer.

The estimation system 4 refers to the past results and costs (incidental information such as equipment, tools, subcontractors, etc.) accumulated in the core system 2 and the existing drawings accumulated in the server 1 when creating an estimate, By cooperating with the IoT result collection system 5, an item quotation is created. At this time, based on the order specifications of the goods requested by the customer, drawings similar to the goods described in the order specifications are searched from the existing drawings of the server 1, and a quotation including the searched drawings is obtained. to create

In this information system, when creating an estimate drawing for an item, the server 1 reads a drawing similar to the specification of the order specification for the item from the past estimate drawing (existing drawing), processes the drawing, and processes the item. Create quotation drawings.

At this time, the salesperson operates the salesperson terminal 4a to display the drawing (target drawing) of the article described in the order specification on the screen of the salesperson terminal 4a operated by the salesperson, for example. . Then, when the sales person operates the search button on the screen, a search request is sent to the server 1 .
When the search request is received, the server 1 uses the similar drawing search function to search for existing comparative drawings similar to the target drawing.

When retrieving an existing drawing, the server 1 automatically designates an area (body line region) where the main body of the article to be compared with the existing comparative drawing is drawn, among the regions of the target drawing, and designates the designated region. One or more existing comparison drawings that are stored in advance as comparison regions are compared.

Specifically, when comparing drawings, the server 1 designates one or more comparison areas, with an area (body line area) of a target drawing to be compared with an existing comparison drawing as a first comparison area.

Subsequently, the server 1 compares one or more first comparison areas (body line areas) of the designated target drawing with one or more second comparison areas (body line areas) of the existing comparison drawing. Thus, the degree of similarity between the target drawing and the comparative drawing is calculated (calculated).

As a result of calculating (computing) the degree of similarity, the server 1 selects similar drawing candidates in a predetermined order and displays a list on the salesperson terminal 4a.
The salesperson selects a desired drawing from one or more existing drawings listed on the salesperson terminal 4a, and the existing drawing is displayed in an editable state. By correcting the description in the title block, etc., it can be attached to the estimate sheet as an estimate drawing.

According to the information processing system of the present embodiment, it is possible to efficiently search existing drawings similar to a drawing to be estimated (predetermined drawing) from existing drawings.
By sharing the past knowledge accumulated in the server 1 in this way, it is possible to contribute to improvement in the speed and accuracy of provision of quotations.
It should be noted that the search for existing drawings is not limited to the creation of a written estimate, and can contribute to improving the productivity of design work at the design terminal 3, for example.

FIG. 2 is a diagram showing the configuration of an information processing system including a server according to an embodiment of the information processing apparatus of the present invention.

In the information processing system shown in FIG. 2, a target drawing to be used by a user such as a salesperson is set as a target drawing, and one or more comparison regions are set in advance as regions to be compared with the target drawing. A server 1, a core system 2, a design terminal 3 such as a PDM (Product Data Management) system or a CAD terminal, an estimate system 4 including a salesperson terminal 4a, and an IoT result collection system 5 , and customer terminal 6 .
The server 1, the core system 2, the salesperson terminal 4a, the quotation system 4, the IoT result collection system 5, and the customer terminal 6 are connected so as to be able to communicate with each other via a network NW such as the Internet. . The form of the network NW is not particularly limited to the Internet, and for example, Bluetooth (registered trademark), Wi-Fi, LAN (Local Area Network), etc., may be employed.

The server 1 is an information processing device managed by a service provider, in which existing drawings for quotations created in the past are stored in a searchable manner. In response to a drawing search request from the quotation system 4 , the server 1 searches for similar drawings from existing drawings stored in advance, and returns the search result to the quotation system 4 .

The core system 2 is a system for managing business operations of a company, and stores, for example, past business results and product costs (equipment, tools, subcontractors, etc.), and can read out such information. In response to a request from the estimation system 4, the core system 2 provides information such as the above-mentioned accumulated past results and costs. The core system 2 also provides information to the estimation system 4 in cooperation with the IoT result collection system 5 .

The design terminal 3 is, for example, an information processing device operated by a person in charge of design, and is, for example, a CAD terminal or the like in which CAD software or the like is installed.
The design terminal 3 can access the server 1 and read and process the existing drawing from the server 1 separately from the estimate.
The design terminal 3 is composed of, for example, a personal computer. In this example, one design terminal 3 is illustrated, but there may be more than one.

The estimate system 4 is a system that includes, for example, a salesperson terminal 4a used by a salesperson and assists in creating an estimate.
A person in charge of a customer sends a request for an estimate based on an order specification or the like for an item such as a product or parts to be ordered to the sales person terminal 4a of the estimate system 4 via the customer terminal 6. FIG.
A salesperson operates a salesperson terminal 4a in response to an estimate request from a customer and prepares an estimate with the support of the estimate system 4. - 特許庁When creating an estimate sheet, the estimate system 4 sends a request for similar drawing search to the server 1 for estimate drawings, and the server 1 executes search processing for similar drawings (searched drawing list), and the result (searched drawing list) is sent to the estimate system. 4 is displayed on the salesperson terminal 4a.

The IoT performance collection system 5 collects information (model number, manufacturer, function, performance, dimensions, and other information) about parts and products that this company has a track record of handling, and in response to a request from the estimation system 4 provide information to

The customer terminal 6 is a terminal used by the person in charge of the customer, and transmits the specifications of the article to the order recipient before placing an order, and receives the quotation prepared by the order recipient.

FIG. 3 is a block diagram showing an example of a hardware configuration of a server in the information processing system of FIG. 2. As shown in FIG.

The server 1 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a bus 14, an input/output interface 15, an input section 16, and an output section 17. , a storage unit 18 , a communication unit 19 , and a drive 20 .

The CPU 11 executes various processes according to programs recorded in the ROM 12 or programs loaded from the storage unit 18 to the RAM 13 .
The RAM 13 also stores data necessary for the CPU 11 to execute various processes.

The CPU 11 , ROM 12 and RAM 13 are interconnected via a bus 14 . An input/output interface 15 is also connected to this bus 14 . An input unit 16 , an output unit 17 , a storage unit 18 , a communication unit 19 and a drive 20 are connected to the input/output interface 15 .

The input unit 16 is composed of various hardware buttons and the like, and inputs various kinds of information according to the instruction operation of the operator.
The output unit 17 is configured by a display such as liquid crystal, and displays various images.

The storage unit 18 is composed of a DRAM (Dynamic Random Access Memory) or the like, and stores various data.
The communication unit 19 controls communication with other devices (the design terminal 3, the salesperson terminal 4a, the quotation system 4, etc.) via the network NW including the Internet.

A drive 20 is provided as required. A removable medium 31 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is mounted in the drive 20 as appropriate. A program read from the removable medium 31 by the drive 20 is installed in the storage unit 18 as necessary. The removable medium 31 can also store various data stored in the storage unit 18 in the same manner as the storage unit 18 .

Such cooperation between various hardware and software of the server 1 in FIG. 3 enables the server 1 to execute various processes described later.
Although not shown, among the information processing system in FIG. 2, the core system 2, the design terminal 3, the salesperson terminal 4a, etc. also have the hardware configuration shown in FIG. Therefore, the description of the hardware configuration of the core system 2, the design terminal 3, the salesperson terminal 4a, etc. will be omitted. However, when the design terminal 3 and the salesperson terminal 4a are composed of a smart phone or a tablet, they have touch panels as the input unit 16 and the output unit 17 .

FIG. 4 is a functional block diagram showing an example of a functional configuration for realizing a similar drawing search process among the processes whose execution is controlled by the server in FIG.

As shown in FIG. 4, in one area of the storage unit 18 of the server 1, an area designation AI model 61, a similarity calculation AI model 62, a drawing DB 63, and the like are stored. DB is an abbreviation for database.

The area designation AI model 61 is a machine learning model of an area designation AI (Artificial Intelligence) machine-learned using training data for area designation. Designate (specify a range) a comparison target area (area) in the target drawing for comparison.
The contents of the region-designating AI model 61 may be configured in a database that allows information to be searched according to conditions and rules. It is sufficient to exclude (ignoring) dimension lines and frame lines such as title blocks outside the area and specify only the area of the main body line indicating the shape (figure) of the main body of the article.

The similarity calculation AI model 62 is a machine learning model of a similarity calculation AI that has been machine-learned using teacher data for similarity calculation. The degree of similarity between the region to be compared (area) and the existing region of another drawing is calculated, and the degree of similarity of the calculation result is output.

In the drawing DB 63, drawings (existing CAD drawings, etc.) created when estimating an item before receiving an order are attached with accompanying information (drawing number, drawing name, name of estimated item, customer, revision, material, plate thickness, type of material, thumbnail, creation date, etc.) are stored in association with each other.

As shown in FIG. 4, in the CPU 11 of the server 1, when a process related to searching for a target drawing to be quoted is executed, a target drawing acquiring unit 41, an area designating unit 42, a drawing processing unit 43, A supplementary information acquisition unit 44, a similar drawing search unit 45, a similar drawing presentation unit 46, a learning unit 47, and a feedback information acquisition unit 48 (hereinafter referred to as "FB information acquisition unit 48") function.

The target drawing acquisition unit 41 acquires a target drawing used by a salesperson as a target drawing. Specifically, when there is a drawing acquisition request from the quotation system 4 according to an instruction from the salesperson via the salesperson terminal 4a, the target drawing for the drawing acquisition request is acquired.
In addition, it is also possible to obtain drawings and the like attached to the specifications contained in the request for quotation sent from the customer from the terminal 4a of the person in charge of sales without going through the quotation system 4. FIG. Alternatively, one of the drawings of the existing quotations stored in the sales representative terminal 4a may be read out and used as the target drawing.

The area specifying unit 42 selects an area of the target drawing acquired by the target drawing acquiring unit 41 to be compared with the comparative drawing (area of the body line ignoring lines that are not directly related to the shape of the article, such as dimension lines and title blocks). etc.) as comparison areas, one or more comparison areas are specified.
Specifically, the area specifying unit 42 inputs the target drawing to the area specifying AI model 61, acquires the comparison area information (coordinates of the area on the target drawing) output from the area specifying AI model 61, and A comparison area is superimposed on the drawing (see comparison areas 73 and 74 in FIG. 6).

The drawing processing unit 43 processes a confidential area such as a title block of a CAD drawing in the target drawing. The processing is, for example, mask processing (mask processing) or mosaic processing in the title column, so that the information in the confidential area is not included in the retrieval of similar drawings. Conversely, the information in the title column may be passed to the incidental information acquisition unit 44 and used as incidental information.

The supplementary information acquisition unit 44 acquires information about the estimate target (item) or the user as supplementary information. Supplementary information is information about the object of quotation (goods), for example, information such as customer, revision, material, plate thickness, etc. shown in FIG. In addition, the incidental information includes information obtained from the core system 2 and the like.

The similar drawing search unit 45 compares one or more comparison areas of the target drawing and one or more comparison areas of the comparison drawing to calculate (operate) the degree of similarity between the target drawing and the comparison drawing.
Specifically, the similar drawing search unit 45 generates histograms (see histograms 82 to 84 in FIG. 9) each showing the frequency of a plurality of elements including at least feature points extracted from figures included in the comparison area of the target drawing. ), and compare the vectors of the target drawing and the comparison drawing to calculate the degree of similarity between the target drawing and the comparison drawing. It is also possible to compare the respective histograms of the object drawing and the comparative drawing, and to calculate the degree of similarity between the object drawing and the comparative drawing based on the comparison result.

That is, the similar drawing search unit 45 obtains a histogram of one or more comparison areas of the target drawing requested by the quotation system 4 and one or more comparison areas of the existing drawings (comparative drawings) accumulated in the drawing DB 63. By comparing the histograms, the degree of similarity between the object drawing and the comparative drawing is calculated (calculated), and the numerical value of the similarity and the comparative drawing are output as the calculation (calculation) result in correspondence.
The similar drawing search unit 45 passes the comparative drawing associated with the degree of similarity and its thumbnail as a search result to the similar drawing presenting unit 46 .

The similar drawing presenting unit 46 displays a list screen 91 (see FIG. 10) in which the thumbnails of the drawing candidates searched by the similar drawing searching unit 45 are arranged in a predetermined order (for example, in descending order of numerical similarity). It is displayed on the sales person terminal 4a.

The learning unit 47 performs machine learning using a plurality of pieces of learning information, using a known drawing and information about one or more areas previously specified from the known drawing as learning information. A region-designated AI model 61 that designates and outputs one or more comparison regions is generated and stored in the storage unit 18 .
The learning unit 47 considers not only the mere similarity of the areas of the drawings, but also the supplementary information attached to the drawings (the company that requested the quotation, the products handled, the person in charge, etc.), so that the annotation for each company such as a customer Rules are also learned to update the similarity calculation AI model 62 . Annotation rules are rules for adding information (metadata) related to data as annotations.

In addition, the learning unit 47 performs machine learning using a plurality of pieces of learning information, using a known drawing and a histogram related to one or more regions specified in advance from the known drawing as learning information. A similarity calculation AI model that inputs a comparison area and a comparison area of an existing drawing, compares the histograms of each comparison area, calculates (calculates) the similarity of each area, and outputs the similarity calculation result. 62 is generated and stored in the storage unit 18 .

The learning unit 47 has a model generation unit 51 and a model update unit 52 .
When a predetermined drawing is input, the model generation unit 51 generates an area-designated AI model 61 that designates and outputs one or more comparison areas, and stores it in the storage unit 18 .
The model updating unit 52 performs machine learning using a plurality of pieces of learning information, using a known drawing and information about one or more regions previously specified from the known drawing as learning information, and stores the learning information in the storage unit 18. update the region-specifying AI model 61 that is
In addition, the model update unit 52 updates the similarity calculation AI model 62 by adjusting parameters for similarity calculation based on feedback information from the FB information acquisition unit 48 .

The FB information acquisition unit 48 obtains information including at least the search result of the similar drawing of the similar drawing search unit 45 for the target drawing and the evaluation of the salesperson (user) on the search result as feedback information such as a user evaluation value. It acquires and outputs to the learning unit 47 .

According to the server 1 configured as described above, the area designating section 42, the similar drawing searching section 45, and the similar drawing presenting section 46 are provided, so that the sales person can search by drawing number, drawing name, etc. By simply preparing a target drawing, similar drawings that describe articles more similar to the target drawing are read from the drawing DB 63 and displayed in a list from the existing drawings. Can create quotation drawings. As a result, the salesperson can efficiently create an estimate sheet attached with an estimate drawing.

Next, the operation of the server in the information processing system of the embodiment will be described with reference to FIG. FIG. 5 is a flow chart showing the operation of the server in FIG. 4 for similar drawing retrieval.

In the server 1, in step S101, the area designating unit 42 designates one or more comparison areas to be compared with the comparative drawing in the target drawing to be estimated.

In step S102, the similar drawing search unit 45 compares one or more comparison areas of the specified target drawing with one or more comparison areas of the comparison drawing, thereby determining the degree of similarity between the target drawing and the comparison drawing. Calculation is performed, and the degree of similarity of the calculation result is associated with the comparison drawing.

In step S103, the similar drawing presenting unit 46 displays the calculation result calculated by the similar drawing searching unit 45 and a list screen (see FIG. 10) in which one or more comparative drawings similar to the target drawing are arranged on the salesperson terminal 4a. indicate.

In this way, according to the operation of the server 1, the server 1 designates a comparison area (graphic area) from the target drawing to be quoted, and compares it with one or more comparison areas of the comparison drawing. The degree of similarity with the drawing is calculated, and a list of those with the highest degree of similarity is displayed on the salesperson's terminal 4a. Since similar existing drawings can be read out, it is possible to efficiently create a written estimate to which the read out existing drawings are attached.

Hereinafter, similar drawing search processing in the server will be specifically described with reference to FIGS. 6 to 11. FIG.
First, referring to FIG. 6, the designation operation of the graphic area (comparison area) of the target drawing in the server will be described.
FIG. 6 is a diagram showing a state in which a figure area (comparison area) is specified in the target drawing.

As shown in FIG. 6, when a search button 72 provided on the screen 70 is operated by the salesperson with the target drawing 71 displayed on the screen 70 of the salesperson terminal 4a, the region specifying unit 42 inputs the target drawing 71 to the region-designating AI model 61, and based on the output (region coordinate information) from the region-designating AI model 61, one or more comparison regions 73 and 74 are determined in the target drawing 71. Specify (area setting).

At this time, the area designation AI model 61 geometrically designates the area. For example, a region is specified by regarding a portion of a shape drawn with thicker lines than other lines as an object to be enclosed. Specifically, based on the outer diameter line (body line) of the graphic area article body, a point (coordinates) separated from the part by an interval of XX pixels (XX is a numerical value) is specified as a part of the area. .
In addition, for example, an area may be specified by machine learning how to enclose an area when the area is enclosed by an operator's operation.

As a result, the work efficiency can be improved by automating the designation of the target when retrieving parts. Among drawings handled in the information processing system, for example, a CAD drawing or the like is provided with a title column 75 as shown in FIG. The title column 75 contains confidential information such as the name of the customer. By applying the processing, the confidentiality of the original drawing can be maintained even when the existing drawing is used.

Next, the operation of searching for similar drawings from the comparison area will be described with reference to FIGS. 7 and 8. FIG.
FIG. 7 is a diagram showing how feature points are extracted from a designated comparison area.
FIG. 8 is a diagram showing an example of additional information to be added to search conditions.

When the comparison area 73 is designated in FIG. 6, the comparison area 73 is cut out from the drawing by the drawing processing unit 43 (FIG. 7 is an image diagram of the cut-out area).
The similar drawing search unit 45 extracts feature points 77 from the area in FIG.
In addition, the incidental information acquisition unit 44 acquires incidental information (customer, material, plate thickness, material type, type, dimension, etc.) related to the article to be quoted from the core system 2 and outputs it to the similar drawing retrieval unit 45. .

The incidental information can be set by selecting or keying in the incidental information setting screen 80 (see FIG. 8) for the item to be estimated.
As shown in FIG. 8, the incidental information setting screen 80 is provided with a pull-down menu for customer, material, etc., and the customer, material, etc. can be selected and set as incidental information, as well as revision, plate thickness, etc. can also be entered as one of the incidental information.

The similar drawing search unit 45 combines the feature points 77 extracted from the area of FIG. Calculate the similarity of the drawings to each other.

Here, an example of comparing the histograms of the target drawing and the comparative drawing will be described with reference to FIG.
FIG. 9 is a diagram showing a comparative example of histograms generated from feature points and incidental information.
As shown in FIG. 9, a histogram 81 of the comparison area of the target drawing and histograms 82 to 84 of the comparison area of the comparison drawing (existing drawing) are compared to calculate the degree of similarity between the drawings.
Here, the similar drawing search algorithm to be used is based on, for example, either Euclidean distance or cosine similarity. List as

Next, referring to FIG. 10, a list screen of drawings retrieved based on the calculated similarity will be described.
FIG. 10 is a diagram showing a list screen of drawings searched according to the degree of similarity.

When the similar drawing search unit 45 lists drawings with a degree of similarity greater than or equal to a certain threshold as candidates, the list is output to the similar drawing presenting unit 46 .
The similar drawing presenting unit 46 displays similar drawing candidates on a list screen 91 shown in FIG. 10 based on the list input from the similar drawing searching unit 45 .
On the list screen 91, similar drawing candidates are listed according to a certain rule, for example, in order of search, order of high similarity, or order of similarity of incidental information.

As the information of each candidate of the similar drawing, the thumbnail image of the retrieved candidate drawing, the drawing name, the degree of similarity, the access button for referring to the past cost, and the like are displayed.

In addition, a check box 92 is provided at the head of each candidate for a similar drawing, and the user checks the check box 92 for a drawing to be used for an estimate or a drawing that the user considers to be similar, thereby obtaining a search result. You can make your own assessment of

Next, tuning of the AI model will be described with reference to FIG.
FIG. 11 is a diagram showing how the AI model is tuned based on the user's evaluation of the similar drawing search results.

The FB information acquisition unit 48 acquires the user's evaluation result of the similar drawing search results, and outputs the result to the learning unit 47 . As a result, the weights of the parameters of the similarity calculation AI model 62 are adjusted by the learning unit 47 .
Specifically, in the list screen 91 shown in FIG. 10, when the salesperson checks a desired check box 92 out of the check boxes 92, this is used as the user's evaluation of the search result, and this evaluation is used as the FB It is acquired by the information acquisition unit 48 and output to the learning unit 47 .

In the learning unit 47, the model update unit 52 (model update AI) calculates a cluster number to which weight should be placed based on the user's evaluation and the tendency of the user's unique thinking.
In the example of FIG. 11, the range of the ellipse 85 in the histogram 81, that is, the vicinity of the cluster numbers "150" to "400" is the portion to be weighted.

Then, the model update unit 52 (model update AI) adjusts the weight of the parameter so as to place weight around the calculated cluster number, and feeds it back to the similarity calculation AI model 62 (search engine).

As described above, according to the information processing system of this embodiment, the following effects can be obtained.
First, the region designation AI model 61 automatically designates and encloses the region of the main body of the article among the figures drawn on the target drawing, so that the work of designating the range by the user can be omitted. .
Second, the similarity calculation AI model 62 extracts feature points and lines of figures existing in the enclosed area, develops them into distributions (histograms (vectors)), and then performs searches (histograms (vectors)). )), it is possible to search for drawings in which an article having a shape more similar to the shape of the article in the target drawing is drawn.
The third is that the model update unit 52 (model update AI) determines how to surround the figure drawn in the target drawing and search method based on the user's evaluation (feedback information) acquired by the FB information acquisition unit 48. By adjusting the method, it becomes possible to retrieve a drawing that is closer to the drawing that the user has requested.

In the above-described embodiment, extraction of feature points and lines of figures of articles included in the target drawing, development of distribution (histogram (vector)), and retrieval of similar drawings containing similar figures have been described. When retrieving drawings, information other than the shape of the article, i.e., additional information about the article (character strings such as the dimensions and materials of the article included in the drawing), is recognized instead of the shape of the article (main information). It is possible to search for similar drawings using both information and feature quantities (feature distribution map). good too.

The second embodiment will be described below with reference to FIGS. 12 to 16. FIG.
The second embodiment is an embodiment of a detection function of incidental information by image recognition and a part configuration collective similarity search function. In describing the second embodiment, the same components as in the above-described embodiment (FIGS. 2 to 4) are described with the same reference numerals.

In the second embodiment, the functional configuration of FIG. 4 is further provided with an incidental information detection unit, an incidental information processing unit, a parts price acquisition unit, an estimation unit, a recommendation unit, and the like.
In addition to the drawing DB 63, the storage unit 18 of FIG. 4 also stores an estimate master. The estimate master stores information on drawings used for newly created estimates and information on drawings used for estimates in the past. As the drawing information, the identification information (for example, part number, etc.) of the drawing indicating the predetermined article or the parts constituting the predetermined article is stored in association with the processing performance (price) for each part.

The area designating unit 42 shown in FIG. 4 further designates an area including additional information as an additional information area in the target drawing.

The similar drawing search unit 45 calculates the degree of similarity between the target part drawing and each of the plurality of comparison part drawings each time one of the target part drawings is sequentially set as the target part drawing.
Specifically, the similar drawing search unit 45 selects one of a plurality of target component drawings constituting the search device of FIG. etc. are sequentially set in the part drawing of interest, and the part drawing of interest and the quotation master or a plurality of comparison part drawings in the drawing DB 63 (existing D part drawing, E part drawing, F part drawing, etc.) Calculate similarity.
Supplementary information is information other than the main information when the shape (figure) of the article is taken as the main information. For example, as shown in FIG. (in this example, 140, etc.) and alphanumeric characters (in this example, SS400, etc.) indicating the material described in the title column. For example, if the material of the part is SS400, that is, stainless steel, it is possible to estimate whether the part can be processed in-house, and the processing costs and the like in the case of in-house processing can be estimated based on past records. That is, the incidental information is information that can specify the estimated price of the desired product or part. Note that the numerical values indicating the dimensions and the alphanumeric characters indicating the materials of the parts are only examples, and additional information is sufficient.
The similar drawing search unit 45 further compares the target article drawing (for example, the drawing of the search device in FIG. 16) with each article of the plurality of types of comparative article drawings (drawings of the α device, β device, γ device, etc.). The similarity is calculated, and for the comparative article (α device) indicated by the comparative article diagram (α device diagram) that satisfies the conditions based on the similarity, the target part diagram (search device in FIG. 16, etc.) and the comparative part diagram (α device), and the similarity calculation is performed.
Conditions based on similarity include, for example, a condition of the highest similarity and a threshold condition of a similarity of 80% or more.
The similar drawing search unit 45 performs the calculation, for example, with the α device indicated by the figure of the α device having the highest degree of similarity as the object of the similarity calculation with the search device of the drawing of the search device. As a result, it is possible to determine the α device to be subjected to the similarity calculation of the part drawing.

The incidental information detection section detects incidental information from the incidental information area.
Specifically, the incidental information detection unit performs character recognition processing on the incidental information area specified by the area designating unit 42, and extracts one or more incidental information (for example, the external dimensions of the part in the CAD drawing of FIG. 12). alphanumeric characters of the material of the part in the numerical value and the title block, etc.).

The incidental information processing section executes predetermined processing using one or more incidental information detected by the incidental information detecting section.
Specifically, the incidental information processing section executes, as the predetermined processing, a process of calculating the estimated price of the predetermined article using one or more of the detected incidental information.

The part price acquisition unit obtains a comparative article shown in the comparative article diagram that has a degree of similarity that allows each of a plurality of target articles (for example, A part, B part, and C part in FIG. 16) constituting the target article to be determined to be the same part. Acquire the estimated amount (price) associated with (D part, E part, F part, etc.).
For example, if the target part is A part, the comparative article drawing having a degree of similarity (usually the highest degree of similarity) that can be determined as the same part as the A part from the material, outer shape, etc. is the drawing of D part. Therefore, the price of 300 yen for the actual machining associated with the D component shown in the diagram of the D component is the unit price corresponding to the A component. For parts for which the comparative product drawing is not retrieved (for example, C part in FIG. 16), a new quotation is obtained, and the unit price input by the person in charge of quotation is obtained as the price of the C part.

The estimating unit calculates the estimated price of the target article based on the prices obtained for each of the plurality of target parts (for example, parts A to C in FIG. 16).
Specifically, the estimating unit calculates the estimated price of a target article composed of a plurality of parts by accumulating the machining results, the unit prices, etc. acquired for each of the plurality of target parts. As a result, it is possible to estimate the target product by accumulating similar parts in past performance units.

Based on information on processing conditions such as material, hardness, workpiece size, tolerance, etc. obtained as supplementary information of parts or articles (products), the recommendation department sells whether it is possible to respond with the capacity of our own factory, necessary processes and machines, etc. A recommendation is made to the person in charge terminal 4a.
The storage unit 18 stores in advance a processing table that associates the specifications (capacity) of equipment and equipment in the company's own factory with the processing conditions for products that can be processed by the company. Based on processing conditions such as material, hardness, dimensions (workpiece size), tolerance, etc. obtained as incidental information of (product), from the processing table, information such as whether it can be handled by our own factory capacity, necessary processes and machines, etc. is read out, and recommendation information including the information (for example, it can be processed because there is such a machine or facility) is sent to the salesperson terminal 4a and notified to the salesperson as recommended information.
As a result, it is possible to estimate the retrieval device after automatically selecting the similar device corresponding to the retrieval device to be estimated and the component parts of the similar device using the estimation master or the past estimation record. Become.

Next, with reference to FIG. 12, the function of detecting incidental information by image recognition in the second embodiment will be described.
In the above-described first embodiment, the feature points 77 extracted from the area in FIG. 7 and the incidental information about the article acquired by the incidental information acquisition unit 44 are combined into a histogram for each of the target drawing and the comparative drawing, and vector Although calculation of similarity between drawings has been described above, in the second embodiment, a method of using incidental information particularly regarding an article will be specifically described.

FIG. 12 is a diagram showing an example of a CAD drawing as a target drawing.
In the case of a general CAD drawing (hereinafter referred to as “drawing”), as shown in FIG. 12, a plan view and a side view including the shape of a part, and a title block are arranged in the drawing. There are holes, dimension lines, and center lines in the plan view and side view, and each line has a numerical value (15, 70) or a symbol (φ140, φ8×4, R2, etc.) indicating the size as necessary. are placed. Here, the shape of the part itself is defined as main information, and other information is defined as incidental information (dimension lines, numerical values, symbols, title column (confidential area) and attribute information related to the part placed in that column, etc.) do. The attribute information about parts includes, for example, drawing name, drawing number, approval seal, material, quantity, date, company name, scale, creator name, and the like.

The server 1 designates a specific position of this drawing, for example, an image area where the numerical value of the outer diameter of the part is arranged or an image area where the material of the title box is described, and character image recognition is performed for each image area. Accompanying information of the part is detected by executing the processing (OCR processing).
For example, by performing character recognition on the image area in which the numerical values of the outer diameter are arranged, the outer dimensions (for example, 140, etc.) of the parts described in the drawings are detected.
Also, the material of the parts (for example, SS400, etc.) described in the drawing is detected by character recognition of the image area in which the material is described in the title column.

Supplementary information such as the outer shape and material of the parts detected in this way is included in the attribute information for searching for similar drawings, and by executing image search processing for similar drawings from the quotation master and past results, it is possible to select more carefully. You can search for similar drawings. In addition, when searching for similar drawings, by adding supplementary information to processing conditions such as part material, hardness, workpiece size, and tolerances, the retrieved information can be used to determine whether or not the company's factory is capable of handling it and the necessary processes. Information such as machinery can be extracted. The information can be recommended or used for parts estimation (multiplying of parts, etc.).

In this way, in addition to searching for similar drawings, by automatically estimating the processing costs of parts, companies that personalize the estimation work, capitalize the tacit knowledge of veteran know-how, and improve the estimation work of companies. Promote digital transformation (DX).

Next, the component structure collective similarity search function will be described with reference to FIGS. 13 to 16. FIG.
FIG. 13 is a diagram showing the information of the estimate master and the flow of automatic summation.

Information on products for which estimates have been made in the past is stored in the quotation master of the server 1. By inputting the quotation number or the like from the salesperson terminal 4a or the like, the information of the quotation master is read out and is shown in FIG. An estimate screen 133 such as can be displayed.
The estimate screen 133 of FIG. 13 displays the item names, dimensions, tolerances, material codes, weights, number of estimates, unit names, estimated unit prices, estimated amounts, cost unit prices, cost amounts, cost breakdowns, and unit prices of many parts that make up the product. Estimated information such as betting rate classification, unit price betting rate, gross profit amount, gross profit rate, etc. is displayed for each part.
Conventionally, based on the product drawing (working drawing) of a new product, the part drawing (working drawing) of the parts that make up the product is linked, and further, the parts of each part drawing are investigated and competitive quotations are performed. We accumulated man-hours and created estimates for new products, but for products with hundreds or even thousands of parts, it takes a lot of time and effort to estimate each part even for a veteran. was required.

Therefore, in the second embodiment, past performance drawings (similar drawings) that are similar in product shape and incidental information (product attributes, etc.) are searched from the product drawing (working drawing), and the product included in the similar drawing is searched. As for parts composition, we search for similar past performance drawings (similar drawings) in the same way as above, integrate the parts quotation information of the parts drawings (engineering drawings) of the searched similar parts, and estimate the product to be quoted. make an estimate of
Specifically, the part drawing to be estimated is sequentially compared with one or more part drawings to be compared, and the past cost of the product configuration that is comprehensively closest from the comparison results is set for each close part. do.
In addition, for example, similar parts for each part may be scanned across past production numbers, and the past cost of similar parts may be set as that of a new estimated part.

As a result, the salesperson can estimate the product to be estimated without estimating each of the many parts that make up the product, so that effort is not required regardless of past experience. You can create a quote for a product without

Next, with reference to FIGS. 14 and 15, a similarity search method will be described in the case where a device (product) is composed of one or more units, and one unit is composed of one or more parts.
FIG. 14 is a diagram showing an example of an assembled product (apparatus) composed of units and parts.
FIG. 15 is a diagram showing the search operation for similar drawings in the case where the device is composed of units and parts as shown in FIG.
As in the assembly drawing shown in FIG. 14, when a product (equipment) is composed of multiple units and multiple detailed parts, it is difficult to understand from the drawing, considering the differences in detailed specifications. Cannot search.

In such a case, by extracting feature values from the plan view of each part, accumulating the similarity by adding the arrays together for each unit, and then accumulating it for each device, even large products can be It is possible to search for similarity from the part shape of the drawing.

As shown in FIG. 15, for example, an inspection device is a product, and the inspection device is composed of a T unit and a Q unit. The T unit includes parts such as a base plate, a side plate, and a sensor. , base, lever and belt holder.
In this case, the server 1 expands the component configuration of the inspection apparatus and extracts feature amounts from each of the end component drawings, thereby generating a feature distribution map for each component.
For example, for the T unit, the feature amount of the shape of the base plate in the drawing of the base plate, the feature amount of the shape of the side plate in the drawing of the side plate, and the feature amount of the shape of the sensor in the drawing of the sensor are calculated. Do the same for the Q unit.

Subsequently, the server 1 calculates the feature amount of the unit by adding the feature amounts of the arrays of the individual parts and accumulating the feature amounts, and generates a feature distribution map for each unit.
Specifically, the feature amount of the T unit is obtained by integrating the feature amount of the shape of the base plate, the feature amount of the shape of the side plate, and the feature amount of the shape of the sensor in the drawing of the sensor. Do the same for the Q unit.

Furthermore, the server 1 calculates the feature amount of the inspection apparatus by adding the feature amounts of the arrays of the T units and the Q units and accumulating the feature amounts, and generates a feature distribution map for each product.

In this way, for an assembly product (equipment), by accumulating the feature values from the end parts that make up the assembly product (equipment) to the unit, and from the unit to the equipment, the original assembly product (equipment) ) can be calculated.

Next, referring to FIG. 16, the operation of generating an estimate by comparing the target drawing and the comparison drawing will be described.
FIG. 16 is a diagram showing how a similar product is retrieved for each part constituting the retrieval device and an estimated price is output.
As shown in FIG. 16, the target drawing is, for example, a drawing of a product such as a retrieval device. It is assumed that drawings have been retrieved.

Assume that the degrees of similarity of the drawings of the α device, β device, and γ device to the drawing of the search device are 82%, 75%, and 50%, respectively.
Among these, if the condition under which the drawing can be compared as a similar drawing is the highest degree of similarity, the drawing of the α device is selected as the drawing to be compared. Also, if the condition is that the preset threshold value is 80 or more, the drawing of the α device is selected as the drawing to be compared. Other conditions may be used.
In this example, it is assumed that the drawing of the α device is selected as the drawing for comparison under the condition of the highest degree of similarity.

In this case, the server 1 reads the drawings of the D, E, and F parts that make up the existing α device from the estimate master, extracts the respective feature amounts, and extracts the feature amounts and the A parts that configure the search device in advance. The feature values extracted from the part, the B part, and the C part are compared in a round-robin manner to search for a similar product (equivalent product) corresponding to the part of the search device.
It should be noted that the machining results of D, E, and F parts that make up the existing α equipment are stored in the quotation master, and the machining results of D, E, and F parts are 300 yen in order. , 500 yen, and 700 yen.

As a result of the search, in this example, component D is determined to be similar to component A, component F is determined to be similar to component B, and there is no similar component to component C. shall be assumed.
In such a case, the server 1 assumes that the A part is equivalent to the D part, and adds the estimated price (for example, 300 yen) of the D part stored in the estimation master to the configuration of the estimation master search device (estimate target product). Set to A part, which is one of the parts.
Further, the server 1 regards the B part as equivalent to the F part, and sets the estimated price (for example, 700 yen) of the F part stored in the estimation master for the B part. Note that the server 1 determines that there is no applicable part for the C part, and makes a new estimate.
As a result, A part, B part, and C part that constitute the retrieval device are 300 yen and 700 yen, respectively, and the new estimated price (for example, 400 yen), and the total amount of 1400 yen is the estimated price of the retrieval device. calculated as

In this way, it is possible to make an estimate for the search device only by accumulating the parts of the existing similar product and the new estimate of the part different from the existing parts. Compared to , it is possible to greatly reduce the burden of the sales person in estimating, and it becomes possible to estimate a new product regardless of experience.

Although the information processing system including the server of one embodiment of the information processing apparatus of the present invention has been described above, the present invention is not limited to the above-described embodiment. Moreover, the effects described in the present embodiment are merely enumerations of the most suitable effects resulting from the present invention, and the effects of the present invention are not limited to those described in the present embodiment.

In the above embodiment, the "machine drawing" was explained as the target drawing of the article to be estimated, but in addition to this, "architectural drawing", "equipment drawing", "planning drawing", "working drawing", "approval drawing", etc. and the type of drawing is not limited.

In the above-described embodiment, the region designating unit 42 directly relates to the shape of the article, such as dimension lines and title blocks, in the target drawing acquired by the target drawing acquiring unit 41, for comparison regions to be compared with existing comparative drawings. Although one or more body line areas are specified ignoring non-existent lines, the comparison area may be specified by other methods. It is enough to specify the area.

In the above embodiment, the similar drawing search unit 45 generates histograms of the target drawing and the comparative drawing, and calculates the similarity between the drawings on the vector. The similarity between drawings may be calculated by comparison, or the similarity between drawings may be calculated by comparing feature vectors for each unit area of each drawing. That is, it suffices to calculate the degree of similarity between the target drawing and the comparison drawing by comparing one or more comparison regions of the target drawing and one or more comparison regions of the comparison drawing.

Also, the system configuration shown in FIG. 2, the hardware configuration shown in FIG. 3, and the functional configuration shown in FIG. 4 are merely examples for achieving the object of the present invention, and are not particularly limited.

In other words, the functional configuration shown in FIG. 4 is merely an example and is not particularly limited.
That is, it is sufficient if the information processing system has a function capable of executing the above-described series of processes as a whole, and what kind of functional blocks, AI models, and databases are used to realize this function are particularly shown in FIG. Examples are not limiting. Also, the locations of the functional blocks, AI model, and database are not particularly limited to those shown in FIG. 4, and may be arbitrary. For example, the functional blocks, AI model and database of the server 1 may be transferred to the core system 2, the design terminal 3, the quotation system 4, the salesperson terminal 4a, or another information processing device (not shown).
Also, one functional block, AI model and database may be composed of hardware alone, software alone, or a combination thereof.

Further, for example, when a series of processes is executed by software, a program constituting the software is installed in a computer or the like from a network or a recording medium. The computer may be a computer built into dedicated hardware. Also, the computer may be a computer capable of executing various functions by installing various programs, such as a server, a general-purpose smart phone, or a personal computer.

A recording medium containing such a program is not only constituted by the removable medium 31 of FIG. It consists of a recording medium, etc. provided to The removable medium 31 is composed of, for example, a magnetic disk (including a floppy disk), an optical disk, or a magneto-optical disk. Optical discs are composed of, for example, CD-ROMs (Compact Disk-Read Only Memory), DVDs (Digital Versatile Disks), and the like. The magneto-optical disk is composed of an MD (Mini-Disk) or the like. Also, the recording medium provided to the user in a state of being incorporated in the main body of the apparatus in advance is composed of, for example, the ROM 12 in FIG.

In this specification, the steps of writing a program recorded on a recording medium are not only processes that are performed chronologically in that order, but also processes that are not necessarily chronologically processed, and that are performed in parallel or individually. It also includes the processing to be performed.
Further, in this specification, the term "system" means an overall device composed of a plurality of devices, a plurality of means, or the like.

In summary, the information processing apparatus to which the present invention is applied only needs to have the following configuration, and can take various embodiments.
That is, an information processing apparatus to which the present invention is applied (for example, the server 1 in FIG. 4, etc.)
A drawing of a target (such as an item to be quoted) used by a user (salesperson, etc.) is set as a target drawing, and the target drawing and a comparison drawing in which one or more areas to be compared are set in advance as a comparison area. In the information processing system to be compared,
In the target drawing, one or more comparison regions are defined as regions to be compared with the comparison drawing (regions of body lines ignoring lines that are not directly related to the shape of the article, such as dimension lines and title blocks) as comparison regions. area specifying means (for example, the area specifying unit 42 in FIG. 4, etc.) for specifying the
Similarity calculation means (for example, Similar drawing search unit 45 etc. in FIG. 4),
Prepare.
In this way, in the target drawing to be used by the user (sales person, etc.), one or more comparison regions (regions of main lines) are automatically selected by using technology such as object detection, with the region to be compared with the comparison drawing as the comparison region. , and by comparing one or more specified comparison areas with one or more comparison areas of the comparison drawing corresponding to these areas, the degree of similarity between the target drawing and the comparison drawing is calculated. By simply preparing a target drawing, the user can retrieve a drawing describing an article more similar to the article in the target drawing from existing drawings. can be searched.
As a result, regardless of the user's (salesperson's, etc.) experience, it is possible to read existing drawings that are more similar to the target drawing from the target drawing to be quoted, and to efficiently create an estimate.

A known drawing and information about one or more regions previously specified from the known drawing are used as learning information, and machine learning is performed using a plurality of pieces of the learning information, so that when a predetermined drawing is input, the comparison region is determined. learning means for generating a region specification model to be output by specifying one or more (learning unit 47 in FIG. 4, etc.);
further comprising
The area specifying means (for example, the area specifying unit 42 in FIG. 4, etc.), based on the output of the area specifying model (for example, the area specifying AI model 61 in FIG. 4, etc.) to which the target drawing is input, in the target drawing , to specify the one or more comparison regions.
In this way, by using the area designation model (for example, the area designation AI model 61 in FIG. 4, etc.) to designate one or more comparison areas of the target drawing, it is possible to increase the accuracy of the model. It becomes possible to search for similar existing drawings.

The area specifying means (for example, the area specifying unit 42 in FIG. 4) further designates an area containing confidential information (for example, the title column 75 of the CAD drawing in FIG. 6) in the target drawing as a confidential area,
Drawing processing means (for example, the drawing processing unit 43 in FIG. 4, etc.) for processing the confidential area of the target drawing (for example, masking the title column 75),
further provide.
As a result, areas other than the shape of the main body of the article can be excluded from the drawing of interest, and the shape of the main body can be compared with the comparison drawing, making it possible to find a drawing closer to the article of the drawing of interest.
In addition, it is possible to prevent confidential information from being leaked when using an existing drawing.

Supplementary information (for example, customer, revision, material, plate thickness, material type, model size, etc. in FIG. 2 information, etc.) is further provided with additional information acquisition means (for example, the additional information acquisition unit 44 in FIG. 4, etc.),
The comparative image includes additional information (for example, the customer, revision, material, plate thickness, material type, are associated in advance as information managed by the core system 2 such as model dimensions,
In addition to comparing the comparison areas of the target drawing and the comparison drawing, the similarity calculation means (for example, the similar drawing search unit 45 in FIG. 4) further compares the target drawing and the comparison drawing. The degree of similarity between the target drawing and the comparison drawing is calculated based on the comparison of the incidental information (for example, customer, revision, material, plate thickness, material type, model size, etc.).
In this way, the degree of similarity is calculated not only by comparing drawings (images) but also by taking into account the incidental information associated with each (for example, customer, revision, material, plate thickness, grade, model size, etc.) By doing so, drawings similar to the target drawing can be found from among the comparative drawings.

Histograms (histograms 82 to 84 in FIG. 9, etc.) and vectors each showing the frequency of a plurality of elements including at least feature points extracted from figures (main body shape of an article, etc.) included in the comparison area of the comparison drawing. at least one of
The similarity calculation means (for example, the similar drawing search unit 45 in FIG. 4) is a histogram showing the frequency of a plurality of elements including at least feature points extracted from the graphic included in the comparison area of the target drawing. (histograms 82 to 84 in FIG. 9, etc.) and vectors, and generate at least one of the histograms (histogram 81 and histograms 82 to 84, etc. in FIG. 9) of each of the target drawing and the comparison drawing, or the vector Based on the comparison, a degree of similarity between the target drawing and the comparison drawing is calculated.
In this way, for a plurality of elements including at least feature points extracted from figures included in the comparison area of the target drawing, at least one of a histogram and a vector indicating the frequency of each element is generated, and a comparison between the target drawing and the comparison drawing is generated. By comparing each histogram or vector and calculating the similarity between the target drawing and the comparison drawing, it is possible to calculate the similarity based on the features of the figures. It is possible to detect drawings of articles similar to

Information about feature points extracted from a figure included in a known drawing is used as learning information (teaching data), and machine learning is performed using a plurality of pieces of learning information (teaching data) to obtain a predetermined number using the plurality of elements. learning means (for example, the learning unit 47 in FIG. 4) that generates a similarity calculation model (such as the similarity calculation AI model 62 in FIG. 4) that calculates the similarity between a drawing and another drawing;
further comprising
The similarity calculation means (for example, the similar drawing search unit 45 in FIG. 4, etc.) calculates the similarity between the target drawing and the comparison drawing based on the similarity calculation model (similarity calculation AI model 62, etc. in FIG. 4). Compute degrees.
By providing learning means (for example, the learning unit 47 in FIG. 4, etc.) in this way, similarity of figures (shapes) between articles included in a given drawing and another drawing can be obtained using a plurality of elements by machine learning or deep learning. By learning the degree calculation result, it is possible to improve the matching accuracy of the figure (shape) of the article.

Information including at least the calculation result of the similarity calculation means (for example, the similar drawing search unit 45 in FIG. 4) for the target drawing and the user's evaluation of the calculation result are obtained as feedback information (for example, user evaluation value, etc.) further comprising a feedback information acquisition means (for example, the FB information acquisition unit 48 in FIG. 4),
The learning means (for example, the model update unit 52 (model update AI) of the learning unit 47 in FIG. 4) uses the feedback information (for example, user evaluation value, etc.) to calculate the similarity calculation model (similarity calculation in FIG. 4). AI model 62, etc.) is updated.
In this way, by updating the similarity calculation model (such as the similarity calculation AI model 62 in FIG. 4) using feedback information including at least the user's evaluation (e.g., user evaluation value, etc.), the drawing requested by the user is updated. can select an existing drawing that is closer to the

The area specifying means (for example, the area specifying unit 42 in FIG. 4, etc.) further designates an area containing incidental information (eg, the dashed line area in FIG. 12) as an incidental information area in the target drawing,
detection means (incidental information detection unit) for detecting the incidental information (for example, 140 or SS400 in FIG. 12) from the incidental information area;
incidental information processing means (incidental information processing unit) for executing a predetermined process using the detected one or more incidental information;
further provide.
By detecting incidental information through image recognition in this way, it is possible to narrow down the search for similar parts by size and material, rather than simply searching for similar parts based on part shape alone. can be improved.

The target drawing is a diagram showing a predetermined article or a part of the predetermined article,
The incidental information is information that can specify the estimated price of the predetermined product or the part,
The incidental information processing means executes, as the predetermined processing, processing for calculating an estimated price of the predetermined article.
By executing the estimation process as the predetermined process in this manner, it becomes possible to estimate the target product for estimation from similar products, and the burden on the person (for example, a salesperson) who makes the estimate can be greatly reduced.

The target drawing includes a plurality of target part drawings respectively showing a plurality of target parts constituting a predetermined target article,
The comparative drawing includes a plurality of target part drawings showing each of a plurality of comparative parts constituting a predetermined comparative article,
Each time the similarity calculation means (for example, the similar drawing search unit 45 in FIG. 4) sequentially sets one of the plurality of target component drawings as the target component drawing (the drawing of the A component→the drawing of the B component→the C Each time a drawing of a part is sequentially set in the target part drawing, the degree of similarity between the target component drawing and each of the plurality of comparison component drawings (D component drawing, E component drawing, F component drawing) is calculated.
As a result, it is possible to collectively perform a similarity search for parts constituting a device to be searched (for example, a search device).

Each of the comparative parts diagrams is associated with an estimated price of the corresponding comparative parts (for example, the price of the actual machining shown in FIG. 16) (300 yen for D parts, 500 yen for E parts, and 500 yen for F parts). 700 yen, etc.), each of the plurality of target articles (A part, B part, C part) constituting the target article has a degree of similarity that can be determined as the same part. D Therefore, parts price acquisition means (parts price acquisition unit) for acquiring a price of 300 yen associated with D parts shown in the diagram of D parts,
an estimate calculating means (estimating unit) for calculating an estimated price of the target article based on the price obtained for each of the plurality of target parts;
Prepare.
As a result, it becomes possible to estimate the target part or the target product by accumulating similar parts in past performance units.

The target drawing further includes a target product drawing (for example, a product drawing (working drawing) in FIG. 13) showing the target product,
The comparative drawing further includes a comparative product drawing (for example, a product drawing (working drawing) in FIG. 13, etc.) showing the comparative product,
There are a plurality of types of comparative product drawings (three types such as a diagram of α device, a diagram of β device, and a diagram of γ device in FIG. 16),
The similarity calculation means (for example, the similar drawing search unit 45 in FIG. 4, etc.)
Further, calculating the degree of similarity between the target article drawing and each of the plurality of types of comparative article drawings,
About the comparative article (α device) indicated by the comparative article diagram (α device diagram) that satisfies the conditions based on the similarity (for example, the highest similarity condition, the threshold condition of 80% or more similarity, etc.) executes the calculation as a target for calculating the degree of similarity between the target part drawing and the comparison part drawing. As a result, it is possible to determine a device (α device) to be subjected to the similarity calculation of the part drawing.

DESCRIPTION OF SYMBOLS 1... Server, 2... Core system, 3... Design terminal, 4... Quotation system, 41... Target drawing acquisition unit, 42... Area designation unit, 43... Drawing processing Part 44 Supplementary information acquisition unit 45 Similar drawing search unit 46 Similar drawing presentation unit 47 Learning unit 48 Feedback information acquisition unit (FB information acquisition unit) , 51 ... model generation unit, 52 ... model update unit (model update AI), 61 ... region designation AI model, 62 ... similarity calculation AI model, 63 ... drawing DB

Claims (14)

In an information processing system for comparing a drawing to be used by a user as a target drawing and comparing the target drawing with one or more comparative drawings in which one or more regions to be compared are set in advance as comparison regions,
area designating means for designating one or more comparison areas in the target drawing as areas to be compared with the comparison drawing as comparison areas;
a similarity calculation means for calculating a similarity between the target drawing and the comparison drawing by comparing the one or more comparison regions of the target drawing and the one or more comparison regions of the comparison drawing;
An information processing system comprising A known drawing and information about one or more regions previously specified from the known drawing are used as learning information, and machine learning is performed using a plurality of pieces of the learning information, so that when a predetermined drawing is input, the comparison region is determined. learning means for generating an area specification model to be output by specifying one or more;
further comprising
The region specifying means specifies the one or more comparison regions in the target drawing based on the output of the region specifying model to which the target drawing is input.
The information processing system according to claim 1. The area designating means further designates an area containing confidential information as a confidential area in the target drawing,
drawing processing means for processing the confidential area in the target drawing;
The information processing system according to claim 1 or 2, further comprising: Further comprising additional information acquisition means for acquiring information about the target or the user as additional information,
Information about a predetermined target or a predetermined person is associated in advance with the comparative image as incidental information,
The similarity calculating means compares the comparison areas of the target drawing and the comparative drawing, and further, based on the comparison of the incidental information of the target drawing and the comparative drawing, calculates the object calculating the degree of similarity between the drawing and the comparative drawing;
The information processing system according to any one of claims 1 to 3. At least one of a histogram and a vector indicating the frequency of a plurality of elements including at least feature points extracted from figures included in the comparison area of the comparison drawing is prepared,
The similarity calculating means generates at least one of a histogram and a vector indicating frequencies of a plurality of elements including at least feature points extracted from figures included in the comparison area of the target drawing, and calculating the degree of similarity between the target drawing and the comparative drawing based on the comparison of the respective histograms or vectors of the drawing and the comparative drawing;
The information processing system according to any one of claims 1 to 4. Information about feature points extracted from a figure included in a known drawing is used as learning information, and machine learning is performed using a plurality of pieces of the learning information. learning means for generating a similarity calculation model for calculating the degree;
further comprising
The similarity calculation means calculates the similarity between the target drawing and the comparison drawing based on the similarity calculation model.
The information processing system according to any one of claims 1 to 5. feedback information acquisition means for acquiring, as feedback information, information including at least the calculation result of the similarity calculation means for the target drawing and the user's evaluation of the calculation result;
updating the similarity calculation model using the feedback information;
The information processing system according to claim 6. The area designating means further designates an area containing additional information as an additional information area in the target drawing,
a detecting means for detecting the incidental information from the incidental information;
Supplementary information processing means for executing a predetermined process using the detected one or more supplementary information;
The information processing system according to any one of claims 1 to 7, further comprising: The target drawing is a diagram showing a predetermined article or a part of the predetermined article,
The incidental information is information that can specify the estimated price of the predetermined product or the part,
The incidental information processing means executes a process of calculating an estimated price of the predetermined item as the predetermined process.
The information processing system according to claim 8. The target drawing includes a plurality of target part drawings respectively showing a plurality of target parts constituting a predetermined target article,
The comparative drawing includes a plurality of target part drawings showing each of a plurality of comparative parts constituting a predetermined comparative article,
Each time one of the plurality of target part drawings is sequentially set as the target part drawing, the similarity calculation means calculates the similarity between the target part drawing and each of the plurality of comparison part drawings.
The information processing system according to any one of claims 1 to 7. Each of the comparison part drawings is associated with a price of the corresponding comparison part,
parts price obtaining means for obtaining the estimated price associated with the comparative article shown in the comparative article drawing having a degree of similarity enabling determination of the same part to each of the plurality of target articles constituting the target article; ,
estimating means for calculating an estimated price of the target article based on the price obtained for each of the plurality of target parts;
The information processing system according to claim 10, comprising: The target drawing further has a target article drawing showing the target article,
The comparative drawing further includes a comparative article drawing showing the comparative article,
There are multiple types of comparative product drawings,
The similarity calculation means is
Further, calculating the degree of similarity between the target article drawing and each of the plurality of types of comparative article drawings,
For the comparison article drawing that satisfies the condition based on the similarity, the calculation is performed as a target for calculating the similarity between the target part drawing and the comparison part drawing.
The information processing system according to claim 10 or 11. In an information processing method executed by an information processing system for comparing a drawing to be used by a user as a target drawing and comparing the target drawing with one or more comparative drawings in which one or more regions to be compared are set in advance as comparison regions,
a region designation step of designating one or more comparison regions in the target drawing as comparison regions to be compared with the comparison drawing;
a similarity calculation step of calculating a similarity between the target drawing and the comparison drawing by comparing the one or more comparison regions of the target drawing and the one or more comparison regions of the comparison drawing;
Information processing method including. A computer that compares a target drawing to be used by a user with a comparison drawing in which one or more comparison regions are set in advance as a region to be compared with the target drawing,
a region designation step of designating one or more comparison regions in the target drawing as comparison regions to be compared with the comparison drawing;
a similarity calculation step of calculating a similarity between the target drawing and the comparison drawing by comparing the one or more comparison regions of the target drawing and the one or more comparison regions of the comparison drawing;
A program that executes control processing including

Images