Information processing device, information processing method, and program

Abstract

To provide an information processing device, an information processing method, and a program for inferring the color of a dental prosthesis using a learned inference model.SOLUTION: An information processing device includes: an inference unit having a learned inference model that has learned using teacher data pairing learning image data imaging a tooth row including a deficient area and correct answer data, which is the information on the color of a dental prosthesis disposed in the deficient area; and an inference object data acquisition unit for acquiring image data of an inference object that images the tooth row including the deficient area. The inference unit infers the color of a dental prosthesis corresponding to the image data of the inference object by inputting the image data of the inference object to the learned inference model.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and a program for inferring the color of a dental prosthesis using a trained model. [Background technology]

[0002] Conventionally, dental prostheses such as inlays, crowns, and bridges have been produced manually by dental technicians, such as by casting. In contrast, CAD / CAM systems, which use computers to design dental prostheses such as inlays, crowns, and bridges and then produce them by cutting, have been attracting attention (Patent Document 1).

[0003] The CAD / CAM system described in Patent Document 1 reads information on the three-dimensional shape of a tooth for which an abutment tooth or cavity has been formed, and in some cases information on the three-dimensional shape of adjacent teeth and opposing teeth.The system then designs the desired dental prosthesis on a computer based on the information on the tooth shape that has been read, and then sets a block-shaped material such as a hardened resin, a sintered ceramic, or a metal body in an automatic cutting machine and cuts it to produce the dental prosthesis. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2002-224142 Summary of the Invention [Problem to be solved by the invention]

[0005] In Patent Document 1, a dental prosthesis is formed by cutting a block material based on information on the three-dimensional shape of teeth, etc. using a CAD / CAM system, but coloring the dental prosthesis is costly because a dental technician must check the consistency with the remaining teeth. The present invention aims to provide an information processing device, information processing method, and program that infers the color of a dental prosthesis using a trained inference model. [Means for solving the problem]

[0006] In order to achieve the object of the present invention, an information processing device according to the present invention includes an inference unit having a trained inference model trained using training data that pairs training image data obtained by photographing a row of teeth including a missing area with ground truth data that is information regarding the color of a dental prosthesis to be placed in the missing area, and an inference target data acquisition unit that acquires image data of an inference target obtained by photographing a row of teeth including a missing area, and the inference unit inputs the image data of the inference target into the trained inference model to infer the color of the dental prosthesis corresponding to the image data of the inference target. In this regard, From a plurality of different trained inference models, an inference model to be used is selected based on the size of the missing area of ​​the tooth row in the image data of the inference target, and inference is performed. It is characterized by: [Effects of the Invention]

[0007] According to the present invention, the color of a dental prosthesis can be inferred using a trained model. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a diagram showing the configuration of a dental prosthesis manufacturing system according to the present invention. [Figure 2] FIG. 2 is a diagram showing a production flow of a dental prosthesis according to the present invention. [Figure 3] 1 is an image diagram of a production flow of a dental prosthesis according to the present invention. [Figure 4] 1 is a schematic diagram of an example of an apparatus for realizing each part of the dental prosthesis manufacturing system of the present invention. [Figure 5] 1 is a schematic diagram of an intraoral scanner 1000 of the present invention. [Figure 6] 1 is a schematic diagram of an information processing device 2000 (dental CAD 2000) according to the present invention. [Figure 7] 1 is a flowchart showing the inference of the color of a dental prosthesis by the information processing device 2000 (dental CAD 2000) of the present invention. [Figure 8] 1 is a schematic diagram of an inference model of the information processing device 2000 (dental CAD 2000) of the present invention during learning. [Figure 9] A learning flow diagram for an inference model by the information processing device 2000 of the present invention. [Figure 10] FIG. 10 is a flow diagram showing a modified example of the inference flow of the present invention. [Figure 11] 1 is a schematic diagram of a dental CAM 3000 of the present invention. [Figure 12] 1A to 1C are diagrams illustrating a flow of forming a dental prosthesis according to the present invention. [Figure 13] 1 is a schematic diagram of a coloring device 4000 of the present invention. [Figure 14] 3A to 3C are diagrams illustrating a coloring flow of the dental prosthesis of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

[0010] [First embodiment] FIG. 1 shows the configuration of a dental prosthesis manufacturing system according to the present invention.

[0011] The dental prosthesis manufacturing system of the present invention includes an intraoral information acquisition unit 100 that acquires information about the oral cavity, and a data generation unit 200 that generates three-dimensional shape data and color data for the dental prosthesis based on the acquired intraoral information.The dental prosthesis manufacturing system further includes a formation unit 300 that forms the dental prosthesis based on the generated three-dimensional shape data, and a coloring unit 400 that colors the surface of the formed dental prosthesis based on the generated color data.The formation unit 300 and the coloring unit 400 may be configured as a single device that performs both processes.

[0012] The flow of dental prosthesis production using the dental prosthesis manufacturing system shown in Figure 2 will be explained below with reference to the image diagram in Figure 3.

[0013] In step S200, the intraoral information acquisition unit 100 acquires intraoral information of the subject from a storage device or the like. The information acquired by the intraoral information acquisition unit 100 is information on the dentition including missing areas, and is image data including information on the three-dimensional shape of the teeth and tooth color information. The intraoral information acquisition unit 100 transmits the acquired intraoral information to the data generation unit 200 as image data to be inferred. Note that the intraoral information acquired by the intraoral information acquisition unit 100 here may be image data captured using an imaging means such as an intraoral scanner (see 301 in Figure 3). 301 in Figure 3 shows an image of acquiring intraoral information of a user.

[0014] In step S201, the data generation unit 200 infers the color of a dental prosthesis using a trained inference model for the acquired image data to be inferred. The trained inference model is a trained inference model trained using training data that pairs training image data of a tooth row including a missing area with ground truth data, which is information about the color of a dental prosthesis to be placed in the missing area. As will be described in detail later, the inference model may be based on a convolutional neural network (CNN) that applies ground truth labels corresponding to pixel values ​​indicating the color of the dental prosthesis as ground truth data. Alternatively, the inference model may be based on a generative adversarial network (GAN) such as pix2pix that has trained on ground truth image data as ground truth data. The data generation unit 200 further generates three-dimensional shape data and tooth color data based on the inferred color and shape (see 302 in FIG. 3). The inference process using the trained inference model in the data generation unit 200 and the generation of data based on the inference process may be performed by different devices. The data generating unit 200 may infer the three-dimensional shape of the dental prosthesis based on, for example, the difference between the master information of the tooth shape and the transmitted three-dimensional tooth shape information, and generate three-dimensional tooth shape data. After generating the three-dimensional tooth shape data and tooth color data, the data generating unit 200 transmits the generated three-dimensional shape data to the forming unit 300 and transmits the inferred color data to the coloring unit 400. Here, in 302 of Figure 3, the dental prosthesis is shown in dark color, and the existing teeth are shown in light color.

[0015] In step S202, the forming unit 300 forms a dental prosthesis by cutting and grinding the base material of the dental prosthesis based on the acquired three-dimensional shape data of the tooth. After the dental prosthesis is formed by the forming unit 300, the formed dental prosthesis is sent to the coloring unit 400.

[0016] In step S203, the coloring unit 400 colors the surface of the dental prosthesis acquired from the forming unit 300 based on the color data of the dental prosthesis inferred by the acquired trained model. 303 in FIG. 3 shows an image in which the forming unit 300 forms a dental prosthesis from a base material for the dental prosthesis, and the coloring unit 400 colors the formed dental prosthesis. Coloring the dental prosthesis according to this flow can generate a highly compatible dental prosthesis, as shown as an image in 304 in FIG. 3.

[0017] An example of a device that realizes each unit of the dental prosthesis manufacturing system will now be described with reference to FIG. 4. The function of the intraoral information acquisition unit 100 is realized, for example, by an intraoral scanner 1000 (described later with reference to FIG. 5). The intraoral information captured by the intraoral scanner 1000 is transmitted to an information processing device 2000 (described later with reference to FIG. 6, etc.) that has the function of a data generation unit 200. The three-dimensional shape data of the dental prosthesis generated by the information processing device is transmitted to a dental CAM 3000 (described later with reference to FIG. 11) that has the function of a formation unit 300. The color data of the dental prosthesis inferred by the trained inference model held by the information processing device 2000 is transmitted to a coloring device 4000 (described later with reference to FIG. 13) that has the function of a coloring unit 400. The color inference of the dental prosthesis in the information processing device 2000 and the data generation of the dental prosthesis may be realized by different devices. For example, the output of the trained inference model held by the information processing device 2000 is transmitted to another device such as a dental CAD, and the dental CAD generates data such as the color and shape of the dental prosthesis. Note that the devices that realize each part of the dental prosthesis manufacturing system are not limited to those described above. Each device that constitutes the above-mentioned dental prosthesis manufacturing system will be described below.

[0018] <Intraoral Scanner 1000> First, a schematic diagram of an intraoral scanner 1000 that realizes the functions of the intraoral information acquisition unit 100 in the dental prosthesis manufacturing system will be described with reference to Figure 5. The intraoral scanner 1000 includes a memory unit 501, an operation unit 502, a communication IF (Interface) 503, and an imaging control unit 504. The memory unit 501 stores acquired tooth information, etc., and the operation unit 502 accepts operations by the user. The communication IF 503 is realized by a LAN card or the like, and controls communication with an external device (e.g., a dental CAD 2000). The imaging control unit 504 acquires tooth image data.

[0019] The image data acquired by the imaging control unit 504 is stored in the storage unit 501 and transmitted to the information processing device 2000 as image data to be inferred using the communication IF 503. The intraoral information acquisition unit 100 may be replaced by acquiring pre-photographed intraoral information from a storage device or the like. The intraoral scanner 1000 is a scanner for acquiring intraoral information and does not necessarily need to be inserted into the oral cavity to capture images. For example, the intraoral information acquisition unit 100 may take an intraoral seal and scan it to acquire three-dimensional shape information of the teeth.

[0020] <Information Processing Device 2000> Next, an example of a schematic diagram of an information processing device that realizes the functions of the data generation unit 200 in the dental prosthesis manufacturing system is shown. As shown in Fig. 6, the information processing device 2000 includes a communication IF 601, a ROM 602, a RAM 603, a storage unit 604, an operation unit 605, a display unit 606, and a control unit 607.

[0021] The communication IF 601 is implemented by a LAN card or the like and controls communication with the external intraoral scanner 1000, dental CAM 3000, and coloring device 4000. However, if communication is not required, the communication IF 601 is not essential. The ROM 602 is implemented by a non-volatile memory or the like and stores various programs. The RAM 603 is implemented by a volatile memory or the like and temporarily stores various information. The storage unit 604 is implemented by a hard disk drive (HDD) or solid state drive (SSD) or the like and stores various information. The operation unit 605 is implemented by a keyboard, a mouse, or the like, and the display unit 606 is implemented by a display or the like and displays various information to the user. The operation unit 605 and the display unit 606 may function as components of the dental CAD 2000 as shown in this drawing, or may be configured as devices separate from the dental CAD 2000.

[0022] The control unit 607 includes an inference target data acquisition unit 608 that acquires image data of an inference target, and an inference unit 609 having a trained inference model trained using training data that pairs training image data of an image of a row of teeth including a missing area with ground truth data, which is information about the color of a dental prosthesis to be placed in the missing area. The inference unit 609 infers the color of a dental prosthesis corresponding to the image data of the inference target by inputting the image data of the inference target into the trained inference model. The information processing device 2000 also includes a data generation unit 610 that generates data of the dental prosthesis based on the inferred color and three-dimensional shape of the dental prosthesis, and a display control unit 611 that displays the generated data and information about the color of the dental prosthesis inferred by the inference unit 609 on the display unit 606.

[0023] The function of the data generating unit 610 may be realized by a different device.

[0024] 7, a flow of inferring the color of a dental prosthesis and generating data by the information processing device 2000 will be described. Note that each part of the information processing device 2000 may be configured as a plurality of devices, or may be configured as a device that realizes only some of the components.

[0025] In step S7001, the inference target data acquisition unit 608 acquires image data of the inference target obtained using an imaging means. Upon acquiring the image data of the inference target, the inference target data acquisition unit 608 transmits the image data of the inference target to the inference unit 609, and this step ends.

[0026] In step S7002, the inference unit 609 inputs the image data of the inference target into a trained inference model to infer the color of a dental prosthesis corresponding to the image data of the inference target. Here, the inference unit 609 has a trained inference model trained using training data that pairs training image data of an image of a tooth row including a missing area with ground truth data, which is information about the color of a dental prosthesis to be placed in the missing area. Here, the inference model is a model based on machine learning, such as a CNN-based model or a GAN-based model. The inference unit 609 may have multiple inference models. For example, consider a case where the inference unit 609 has inference models for cases where the missing area of ​​the tooth row is smaller than a predetermined value and larger than a predetermined value. The missing area is, for example, a value calculated by comparing a tooth row based on standard tooth row data with a tooth row including the missing area. When the missing area is large relative to the dentition, it may be preferable to take into account the colors of the teeth adjacent to the tooth to be placed and the corresponding tooth to be mated when inferring the color of the dental prosthesis to be placed in the missing area. On the other hand, when the missing area is small, it may be preferable to take into account the color of the tooth to be placed with the dental prosthesis. Therefore, by having multiple inference models with different training data depending on the size of the missing area, the accuracy of inferring the color of the dental prosthesis can be improved.

[0027] After the inference unit 609 infers the color of the dental prosthesis, it transmits the data to the data generation unit 610. Note that the data generation unit 610 may be configured as a different device, and the information processing device 2000 may function as an independent device that realizes the functions described above.

[0028] In step S7003, the data generation unit 610 generates color data of the dental prosthesis based on the transmitted color information of the dental prosthesis, and displays the generated color data on the display unit 606 via the display control unit 611. The generated color data is also transmitted to the coloring device 4000, and as will be described later, the coloring device 4000 colors the formed dental prosthesis based on the inferred color information.

[0029] Here, the inference unit 609 may infer a plurality of candidate colors for the surface of the dental prosthesis. In this case, the display control unit 611 displays a plurality of candidate colors for the surface of the dental prosthesis on the display unit 606. Furthermore, by further acquiring likelihood information for each color from the inference unit 609, the display control unit 611 can display a plurality of candidate colors together with the likelihood information. The user may perform selection acceptance via the operation unit 605 or the like, allowing the user to select color data from a plurality of candidate colors.

[0030] Here, the configuration for learning a trained inference model in the inference unit 609 will be described with reference to Figures 8 and 9. Figure 8 shows the configuration of the information processing device 2000 during learning, in which the control unit 807 is composed of a training data acquisition unit 808 that acquires training data, and a learning unit 809 that learns an inference model based on machine learning based on the acquired training data. Note that the information processing device having the inference unit 609 and the information processing device 2000 may be the same information processing device or different information processing devices. For example, the inference function is realized by copying the trained inference model generated by the learning unit 809 to the cloud or another information processing device.

[0031] The learning unit 809 learns the inference model and generates a trained inference model. The learning unit 809 stores the model information and parameter information acquired through the learning in the memory unit 604 in association with each other. The training data refers to data that pairs training image data of a row of teeth including a missing area with correct answer data, which is information about the color of a dental prosthesis to be placed in the missing area. The machine learning-based inference model may be a CNN-based model or a GAN-based model. When the inference model is a CNN-based model, the correct answer data is data composed of correct answer labels corresponding to pixel values ​​indicating the color of the dental prosthesis. When the inference model is a GAN-based model, the correct answer data is image data of the row of teeth. The learning unit 809 generates a trained inference model by learning the training image paired with the correct answer data. The learning flow will be described below with reference to FIG. 9.

[0032] In step S9000, the teacher data acquisition unit 808 acquires teacher data from the storage unit 604, etc. Here, teacher data is data that pairs learning image data and correct answer data. The teacher data acquisition unit 808 transmits the teacher data to the learning unit 809.

[0033] Here, the training data acquisition unit 808 may mask a portion of the dentition of a healthy individual and generate training image data with the masked region as a missing region. In this case, the color information masked by the masked region becomes the correct answer data. Furthermore, the training image data may be expanded by a combination of modifying the size, shape, and angle of the mask, the size and angle of the dentition image, brightness, etc.

[0034] In step S9001, the learning unit 809 performs a learning process on the inference model using the training data to generate a trained model. After generating the trained model, the learning unit 809 associates information about the inference model with information about the parameters and stores them in the storage unit 604. The learning unit 809 may train multiple inference models depending on the size of the missing area relative to the dentition in the training image data. For example, the learning unit 809 generates a first inference model trained with first training data including, as training data, training image data in which the missing area relative to the dentition is equal to or smaller than a predetermined standard. The learning unit 809 also generates a second inference model trained with second training data including, as training data, training image data in which the missing area relative to the dentition is larger than a predetermined standard. Here, "including" refers to the case where the training data includes other training data or where additional training is performed using different training data.

[0035] Next, the inference flow of the inference unit 609 when there are multiple models generated using different training data depending on the size of the missing area will be explained using Fig. 10. Here, the same processes as in Fig. 7 are assigned the same reference numerals, and explanations will be omitted as appropriate.

[0036] In step S10001, the inference unit 609 calculates the size of the missing area relative to the dentition in the acquired data to be inferred. Here, the size of the missing area relative to the dentition is, for example, a value calculated by comparing the dentition based on standard dentition data with the dentition including the missing area. Alternatively, this process may be replaced by an inference model that detects the missing area. The inference unit 609 determines whether the missing area relative to the dentition is equal to or smaller than a predetermined value. If the missing area is equal to or smaller than the predetermined value, the process proceeds to step S10002, where the color of the dental prosthesis is inferred using the first inference model trained using the first training data described above. On the other hand, if the missing area is greater than the predetermined value, the process proceeds to step S10003, where the color of the dental prosthesis is inferred using the second inference model trained using the second training data described above.

[0037] (Variation 1) In the above-described embodiment, it was explained that the inference unit 609 selects an inference model to perform inference based on the missing area in the tooth row and whether or not the training data of multiple inference models includes a training image of a specified missing area size, and performs inference.

[0038] Here, multiple trained models generated using different training data may be used to infer the image data for verification, and the accuracy of the multiple trained inference models may be associated with the size of the missing area and stored, and an inference model with high accuracy may be used to perform inference for the size of the missing area in the image data to be inferred. In this way, more accurate inference of the color of the dental prosthesis can be achieved for the image data to be inferred.

[0039] <Dental CAM3000> An example of the configuration of the dental CAM 3000 will be described with reference to FIG. 11, and a process for forming a dental prosthesis will be described with reference to FIG.

[0040] The dental CAM 3000 includes a communication IF 701, a processing unit 702, a memory unit 703, an operation unit 704, a display unit 705, and a control unit 706. The control unit 706 includes a processing control unit 707 and a display control unit 708.

[0041] In step S801, the dental CAM 3000 receives the three-dimensional shape data generated by the dental CAD 2000 via the communication IF 801. The received three-dimensional shape data is stored in the memory unit 803. The dental CAM 3000 may also acquire tooth color data. The tooth color data will be used in steps described below.

[0042] In step S802, the dental CAM 3000 selects a desired base material for the prosthesis based on the received three-dimensional shape data and color data. The selection of the base material for the prosthesis may be replaced by a base material selection made by the user. This step may be omitted as appropriate, for example, if the base material for the prosthesis has been set in advance. Here, the prosthesis components selected by the dental CAM 3000 are preferably base materials for the prosthesis that are close to the tooth color data received by the dental CAM 3000 from the dental CAD 2000. For example, the base material for the dental prosthesis selected is a base material for the prosthesis that is close to the color data and has a high degree of whiteness. The base material for the dental prosthesis may also contain ceramics and / or resin.

[0043] In step S803, the dental CAM 3000 controls the processing unit 802 via a processing control unit 807 constituting the control unit 806 based on the received three-dimensional shape data, and forms a dental prosthesis by processing a base material of the dental prosthesis. The results of processing and forming by the processing unit 802 can be displayed on the display unit 805 via a display control unit 808. Here, the processing unit 802 may form a dental prosthesis by processing the base material of the prosthesis by cutting and / or grinding. Alternatively, the processing unit 802 may process and form a dental prosthesis by repeatedly irradiating the base material of the prosthesis with light to sequentially sinter or melt and solidify the raw material. The processing unit 802 may form a dental prosthesis by an additive manufacturing method. Alternatively, the control unit 806 may be controlled via the operation unit 804. The dental prosthesis formed by the dental CAM 3000 is used in a coloring process by the coloring device 4000.

[0044] <Coloring device 4000> An example of a coloring device 4000 that realizes the function of the coloring unit 400 in the dental prosthesis manufacturing system will be described with reference to FIG. 13, and a dental prosthesis coloring process including this device will be described with reference to FIG.

[0045] The coloring device 4000 includes a communication IF 901, an operation unit 902, a memory unit 903, a rotation unit 904, a discharge unit 905, and a control unit 906. The control unit 906 includes a position detection unit 907, a shape detection unit 908, a rotation control unit 909, and a discharge control unit 910.

[0046] Hereinafter, a specific example of the process of coloring the formed dental prosthesis based on the color data generated by the data generating unit 200 will be described with reference to FIG.

[0047] In step S1001, the coloring device 4000 receives color data of the dental prosthesis generated by the information processing device 2000 via the communication IF 901.

[0048] The coloring device 4000 records the received color data in the memory unit 903. The coloring device 4000 holds the dental prosthesis processed and formed by the dental CAM 3000 in the rotation unit 904. The position detection unit 907 detects position data of the dental prosthesis processed and formed by the dental CAM 3000. The shape detection unit 908 detects the shape of the dental prosthesis by receiving the position data of the dental prosthesis detected by the position detection unit 907 or three-dimensional shape data from the dental CAD 2000. Combining the position data detected by the position detection unit 907 with the received three-dimensional shape data improves the accuracy of detection of the shape of the dental prosthesis by the shape detection unit 908. Detection of the shape of the dental prosthesis by the shape detection unit 908 may be replaced by receiving three-dimensional shape data.

[0049] In step S1002, the coloring device 4000 colors the surface of the dental prosthesis formed by the forming device 3000 based on the color data generated by the information processing device 2000. Specifically, the coloring device 4000 controls the rotation unit 904 with the rotation control unit 909 in accordance with the shape of the dental prosthesis detected by the shape detection unit 908, and colors the dental prosthesis processed and formed by the dental CAM 3000 with coloring material based on the color data while controlling the discharge unit 905 with the discharge control unit 910. After coloring the dental prosthesis, the process proceeds to step S1003. The discharge method of the discharge unit 905 may be inkjet or spray. Here, inkjet may be used for the discharge unit 905 when the amount of coloring material is small. Alternatively, spray may be used as the discharge unit 905 when the amount of coloring material is large. The discharge unit 905 may also have a mechanism that can switch between inkjet and spray. The coloring device 4000 may use two or more coloring materials to color the dental prosthesis, improving the reproducibility of the color data of the dental prosthesis. The coloring device 4000 may also have an applicator (not shown) that applies coloring materials to the dental prosthesis. When the coloring device 4000 applies coloring materials, the applicator may apply multiple layers of coloring materials using different coloring materials. Multi-layer application by the applicator improves the reproducibility of the coloring of the dental prosthesis according to the color data. The multiple coloring materials may have different refractive indices. Using coloring materials with different refractive indices to color the dental prosthesis increases reflectance, allowing for coloring reproducibility with a small amount of application. The coloring material used to color the dental prosthesis by the coloring device 4000 may be selected by referencing a lookup table that links tooth color data with coloring materials. If the coloring device 4000 has a lookup table stored in advance, the computational cost involved in determining the coloring material can be reduced. The coloring material may include B, Si, or resin. A wide color space can be achieved by including B, Si, or resin in the coloring material for coloring dental prostheses.

[0050] Furthermore, the coloring material used to color the dental prosthesis by the coloring device 4000 may contain a scattering material. By using a coloring material containing a scattering material, color reproducibility can be achieved with a smaller amount of application. The scattering material may have a high refractive index. The scattering material with a high refractive index may contain at least one of TiO2, SiO2, and ZrO2.

[0051] In step S1003, the coloring device 4000 determines whether the coloring process is complete. If it is determined that the coloring process is complete, the process ends. If it is determined that the coloring process is not complete, the process proceeds to step S1002, where the coloring process is performed. [Example]

[0052] The present invention will be described in more detail below with reference to examples, but the present invention is not limited to the following examples.

[0053] Example 1 Using the 3D tooth shape data, a Yb fiber laser was irradiated onto the raw material to create a 3D model. Powder 1 in Table 1 was used as the raw material. A 3D shape measuring machine was used to confirm that a prosthesis identical to the 3D tooth shape data had been created. The created prosthesis was then colored with a colorant using an inkjet printer.

[0054] [Table 1]

[0055] Example 2 As in Example 1, a prosthesis was fabricated using Powder 2 as the raw material.

[0056] It was confirmed that the prostheses of Examples 1 and 2 could be used as dental prostheses.

[0057] (Variation) The present invention can also be realized by executing the following process: software (programs) that realize the functions of the above-described embodiments are supplied to a system or device via a network or various storage media, and the computer (or CPU, MPU, etc.) of the system or device reads and executes the programs. [Explanation of symbols]

[0058] 608 Inference target data acquisition part 609 Reasoning section

Claims

1. an inference unit having a trained inference model trained using training data that pairs training image data obtained by photographing a row of teeth including a missing area with ground truth data that is information regarding the color of a dental prosthesis to be placed in the missing area; an inference target data acquisition unit that acquires image data of an inference target obtained by photographing a row of teeth including a missing region, The inference unit infers the color of a dental prosthesis corresponding to the image data of the inference target by inputting the image data of the inference target into the trained inference model, An information processing device characterized by selecting an inference model to be used from a plurality of different trained inference models based on the size of the missing area of ​​the tooth row in the image data of the inference target and performing inference.

2. A teacher data acquisition unit that acquires the teacher data; The information processing apparatus according to claim 1 , further comprising a learning unit that uses the acquired training data to learn the inference model based on machine learning.

3. 3. The information processing device according to claim 1, wherein the inference model is an inference model based on CNN.

4. 4. The information processing apparatus according to claim 1, wherein the correct answer data is data configured from correct answer labels corresponding to pixel values ​​indicating the color of the dental prosthesis.

5. 5. The information processing apparatus according to claim 1, wherein the learning image data includes image data in which a partial area of ​​a tooth row is masked, and the masked area is defined as the missing area.

6. 6. The information processing apparatus according to claim 5, wherein the teacher data is configured as learning image data in which the mask region is the missing region, and the image data of the row of teeth is configured as correct answer data.

7. 7. The information processing device according to claim 5, wherein the inference model is an inference model based on a global area network (GAN).

8. The information processing device according to claim 1 , wherein the inference unit selects an inference model depending on whether the size of the missing area for the tooth row is larger than a predetermined value.

9. The information processing device described in claim 8, characterized in that the multiple inference models include a first inference model learned based on first training data that includes training image data as training data in which the size of the missing area for the tooth row is less than the predetermined value, and a second inference model learned based on second training data that includes training image data as training data in which the size of the missing area for the tooth row is greater than the predetermined value.

10. The information processing device described in claim 8 or 9, characterized in that the inference unit causes the multiple trained models to infer verification image data, and selects a model to infer for the image data to be inferred based on the accuracy of inference of the multiple trained models for the verification image data and the size of the missing area of ​​the tooth row in the verification image data.

11. 10. The information processing device according to claim 8, wherein the size of the missing area is a value calculated by comparing a tooth row based on standard tooth row data with a tooth row including the missing area.

12. an acquisition step of inference target data for acquiring image data of an inference target obtained by photographing a row of teeth including a missing region; an inference step of inferring the color of the dental prosthesis by inputting the image data of the inference target to a trained inference model that has been trained using training data that pairs training image data of a row of teeth including a missing area with ground truth data that is information regarding the color of a dental prosthesis to be placed in the missing area; and An information processing method characterized in that in the inference step, an inference model to be used is selected from a plurality of different trained inference models based on the size of the missing area of ​​the tooth row in the image data to be inferred, and inference is performed.

13. A program for executing the information processing method according to claim 12 on a computer.

Images