JP2023136543A - Color prediction model creation method and color prediction model creation system - Google Patents

Abstract

To provide a color prediction model creation method capable of shortening the time required for generation of teacher data and generation of a color prediction model, and a color prediction model creation system.SOLUTION: A color prediction model creation method, which is a color prediction model creation method of creating a color prediction model by means of machine learning, using teacher data in which an ink amount set that is the combination of ink amounts of ink used for printing is mapped with the spectral reflectivity of a printed matter printed on a printing medium by means of the ink amount set, includes a step S22 of acquiring a medium identifier that represents a kind of the printing medium, a step S26 of acquiring an ink amount set matching an acquired medium identifier, a step S30 of generating teaching data, using the acquired ink amount set, and a step S31 of creating a color prediction model by performing machine learning by means of generated teacher data.SELECTED DRAWING: Figure 4

Description

The present invention relates to a color prediction model creation method and a color prediction model creation system.

A printing device that performs printing using multiple color inks can reproduce various color values by appropriately adjusting an ink amount set that is a combination of ink amounts of each ink. Color prediction models are created in order to predict in advance the color values that will be reproduced with an arbitrary set of ink amounts. The color prediction model is created, for example, by machine learning using a large amount of training data that associates an ink amount set with the calorimetric values of color patches printed with this ink amount set. Patent Document 1 discloses generation of training data used to create a color prediction model.

JP 2021-158398 Publication

Even when printing on the same print medium, if the reproduced colors vary due to post-print processing, etc., it is necessary to create a new color prediction model, and training data is required for this purpose. becomes. However, as the number of ink colors used in a printing device increases, the number of combinations of usable ink amounts becomes enormous, and it is difficult to select an appropriate ink amount set from among these to generate new training data each time. , there was a problem that it took a lot of time.

The color prediction model creation method uses training data that associates an ink amount set, which is a combination of ink amounts used for printing, with the spectral reflectance of a printed matter printed on a print medium using the ink amount set. A color prediction model creation method for creating a color prediction model by machine learning, comprising an identification information acquisition step of acquiring identification information representing the type of print medium, and an ink amount set corresponding to the acquired identification information. a step of acquiring an ink amount set; a teacher data generation step of generating the teacher data using the acquired ink amount set; and creating the color prediction model by performing machine learning using the generated teacher data. and a color prediction model creation step.

The color prediction model creation system uses training data that associates an ink amount set, which is a combination of ink amounts used for printing, with the spectral reflectance of a printed matter printed on a print medium using the ink amount set. A color prediction model creation system that creates a color prediction model by machine learning, the system comprising: an identification information acquisition unit that acquires identification information representing the type of print medium; and an ink amount set corresponding to the acquired identification information. an ink amount set acquisition unit that acquires an ink amount set, a teacher data generation unit that generates the teacher data using the acquired ink amount set, and creates the color prediction model by performing machine learning using the generated teacher data. and a color prediction model creation section.

1 is a block diagram illustrating a schematic configuration of a color prediction model creation system, and is a diagram illustrating an internal configuration of a server. FIG. 1 is a block diagram illustrating a schematic configuration of a color prediction model creation system, and is a diagram illustrating internal configurations of a terminal device and a printing device. FIG. The figure which shows an example of an identification table. A flowchart showing the procedure for creating a color prediction model.

Hereinafter, a color prediction model creation method and a color prediction model creation system according to embodiments will be described with reference to the drawings.
1 and 2 are block diagrams showing a schematic configuration of a color prediction model creation system 1 according to the present embodiment. FIG. 1 is a diagram showing the internal configuration of a server 100, and FIG. 2 is a block diagram showing a terminal device 200 and a printing device. 3 is a diagram showing the internal configuration of 300. FIG. Further, FIG. 3 is a diagram showing an example of the identification table 23.

As shown in FIG. 1, the color prediction model creation system 1 includes a server 100, a terminal device 200, and a printing device 300. The server 100 and the terminal device 200 are connected via a network NW such as a LAN (Local Area Network), and the printing device 300 is connected to the terminal device 200. The color prediction model creation system 1 creates a color prediction model that predicts the spectral reflectance of printed matter. This color prediction model predicts the spectral reflectance of a printed matter printed on a print medium P with the ink amount set based on the ink amount set, which is a combination of ink amounts of multiple colors of ink used in the printing device 300. do. In addition, this color prediction model measures the spectral reflectance of color patches actually printed with multiple ink amount sets, and uses multiple training data that associates the ink amount sets with the measured spectral reflectance. It is created by performing machine learning using

The server 100 is configured by a computer and includes a control section 10, a storage section 20, and a communication section 30. The storage unit 20 and the communication unit 30 are connected to the control unit 10 via a bus or the like. The control unit 10 includes a CPU (Central Processing Unit) and memories such as RAM (Random Access Memory) and ROM (Read Only Memory), and operates according to a control program (not shown) stored in the storage unit 20. By doing so, the operation of the server 100 is controlled. The control unit 10 includes an information acquisition unit 11, a medium discrimination unit 12, a medium discrimination learning unit 13, an ink amount set selection unit 14, a teacher data generation unit 15, and a color information acquisition unit 11, a medium discrimination unit 12, a medium discrimination learning unit 13, an ink amount set selection unit 14, a teacher data generation unit 15, and a color control unit 10, as functional units realized by a control program. It has a predictive model creation section 16. Details of each of these functional units will be described later.

The storage unit 20 is constituted by a storage device such as a hard disk drive or an SSD (Solid State Drive), and stores a data set storage unit 21, a medium discriminator 22, and an identification table 23 in addition to the above-mentioned control program. ing.

The dataset storage unit 21 stores color prediction models created in the past, a plurality of training data used to create this color prediction model, and a plurality of verification data used to verify the created color prediction models. It stores a dataset containing. The colors reproduced by the printing device 300 differ depending on the print medium P, the print mode represented by the resolution in the horizontal and vertical directions, the ink set used for printing, etc., so any of these may change. A new color prediction model is then created. The dataset storage unit 21 stores a plurality of datasets created so far, and each dataset is given a dataset ID that is a unique ID. Note that the teacher data and verification data are data in which the ink amount set and the spectral reflectance are associated, and the data includes the ink amount set, so storing the teacher data and verification data is the same as the teacher data This is synonymous with storing an ink amount set for the test data and an ink amount set for the verification data. The data set storage section 21 corresponds to a storage section.

The medium discriminator 22 is a tool used to identify the print medium P. The medium discriminator 22 of this embodiment is a learning model created by machine learning, and uses a plurality of training data in which the spectral reflectance of the print medium P is associated with a medium identifier representing the type of the print medium P. This is what was learned. Here, the spectral reflectance of the print medium P is the spectral reflectance of an area of the print medium P that is not printed. The medium discriminator 22 discriminates the type of print medium P based on the spectral reflectance of the input print medium P, and outputs a corresponding medium identifier. The medium identifier corresponds to identification information. Furthermore, if the medium discriminator 22 cannot discriminate the print medium P based on the input spectral reflectance, it outputs information indicating that the print medium P is unknown.

The identification table 23 is a table that associates data sets stored in the data set storage unit 21 with various conditions on which these data sets are based.
As shown in FIG. 3, the identification table 23 of this embodiment associates the medium identifier of the print medium P, the print mode, and the ink set with the data set ID. That is, the data set is stored in the data set storage unit 21 in association with the medium identifier, print mode, and ink set. By referring to this identification table 23, the control unit 10 can specify the data set corresponding to the combination of print medium P, print mode, and ink set.

Returning to FIG. 1, the communication unit 30 includes various circuits for communicating with external devices via the network NW. The communication unit 30 of this embodiment communicates with the terminal device 200 connected to the network based on the control of the control unit 10. The mode of communication may be wired communication or wireless communication.

An information acquisition unit 11 provided in the control unit 10 acquires various information necessary for creating a color prediction model from the terminal device 200. Specifically, the information acquisition unit 11 acquires medium information for identifying the print medium P and print information representing printing conditions from the terminal device 200. In this embodiment, the above-described spectral reflectance of the print medium P is acquired as the medium information, and the above-described print mode and ink set are acquired as the print information.

The medium discriminator 12 discriminates the type of the print medium P using the medium discriminator 22 based on the medium information acquired by the information acquisition section 11, that is, the spectral reflectance of the print medium P. Specifically, the medium discriminator 12 inputs the acquired spectral reflectance to the medium discriminator 22 and acquires the medium identifier output by the medium discriminator 22 or information indicating that the print medium P is unknown. . Since the medium determining unit 12 acquires a medium identifier as identification information, it corresponds to an identification information acquiring unit.

The medium discrimination learning unit 13 assigns a new medium identifier to the print medium P when the medium discriminator 22 outputs information indicating that the print medium P is unknown. Then, the medium discriminator 22 is caused to perform machine learning using training data in which the spectral reflectance of the print medium P is associated with the assigned medium identifier, and the medium discriminator 22 is updated. This allows the medium discriminator 22 to discriminate this unknown print medium P.

The ink amount set selection unit 14 selects, according to predetermined selection rules, from among a huge number of ink amount sets obtained by adjusting the ink amounts of multiple colors of ink that can be used in the printing device 300. A plurality of ink amount sets used for generating teacher data and a plurality of ink amount sets used for generating verification data are selected. Note that the method for selecting the ink amount set is not particularly limited, and for example, the method described in Patent Document 1 mentioned above can be adopted. Further, if the ink amount set to be selected can be obtained from the training data and verification data of the data set stored in the data set storage unit 21, the ink amount set selection unit 14 selects the ink amount set. Instead, the ink amount set is obtained from the training data and verification data included in the data set. Therefore, the ink amount set selection section 14 also corresponds to an ink amount set acquisition section.

The teacher data generator 15 uses the ink amount set for teacher data selected or acquired by the ink amount set selector 14 to generate teacher data for creating a color prediction model. Specifically, the printing unit 330 of the printing device 300 prints a color patch on the printing medium P using the ink amount set selected or acquired by the ink amount set selection unit 14, and the color measurement device 340 of the printing device 300 prints a color patch using the ink amount set selected or acquired by the ink amount set selection unit 14. After measuring the spectral reflectance of the color patch, the teacher data generation unit 15 acquires the spectral reflectance and generates teacher data in which the ink amount set and the spectral reflectance are associated with each other. Similarly, the teacher data generation unit 15 uses the ink amount set for verification data selected or acquired by the ink amount set selection unit 14 to generate verification data for verifying the color prediction model.

The color prediction model creation unit 16 creates a color prediction model by machine learning using the teacher data generated by the teacher data generation unit 15. The created color prediction model is verified using verification data generated by the teacher data generation unit 15.

As shown in FIG. 2, the terminal device 200 is configured by a computer and includes a control section 50, a storage section 60, a communication section 70, and an input/output I/F (interface) 80. The storage unit 60, the communication unit 70, and the input/output I/F 80 are connected to the control unit 50 via a bus or the like. The control unit 50 includes a CPU and a memory such as a RAM or ROM, and controls the operation of the terminal device 200 by operating according to a control program (not shown) stored in the storage unit 60.

The storage unit 60 is constituted by a storage device such as a hard disk drive or SSD, and stores the above-mentioned control program and the like. The communication unit 70 includes various circuits for communicating with external devices via the network NW. The communication unit 70 of this embodiment communicates with the server 100 connected to the network based on the control of the control unit 50.

The input/output I/F 80 is an interface for inputting/outputting various information with external devices, and the input/output I/F 80 of this embodiment includes an input device 210, a display device 220, and a printing device. The device 300 is connected. The input device 210 is configured with a keyboard, a pointing device such as a mouse, and receives user input operations and transmits them to the control unit 50. The display device 220 includes a display device such as a liquid crystal display or an organic EL (Electro Luminescence) display, and displays various information based on the control of the control unit 50.

The printing device 300 is, for example, an inkjet printer that forms dots on the print medium P and prints an image by ejecting ink onto the print medium P. The printing apparatus 300 of this embodiment can eject six colors of ink: cyan C, magenta M, yellow Y, black K, orange Or, and red R. The printing device 300 includes a control section 310, an input/output I/F 320, a printing section 330, and a color measurement device 340.

The control unit 310 includes a CPU and a memory such as a RAM or ROM, and controls the operation of the printing apparatus 300. The input/output I/F 320 is an interface for inputting/outputting various information with external devices, and the terminal device 200 is connected to the input/output I/F 320 of this embodiment.

The printing unit 330 includes a print head, and prints an image by ejecting ink droplets onto the print medium P under the control of the control unit 310.

Under the control of the control unit 310, the color measurement device 340 measures the color of the printed matter printed by the printing unit 330 and the printed matter printed by another printing device. The color measurement device 340 of this embodiment measures the spectral reflectance of a group of dots formed on the print medium P. Furthermore, the colorimetric device 340 can measure the spectral reflectance of the print medium P by measuring an area on the print medium P that is not printed. The spectral reflectance measured by the colorimeter 340 is output to the terminal device 200 via the input/output I/F 320. Note that the color measurement device 340 is not limited to a configuration provided integrally with the printing device 300, but can also be configured by a spectrophotometer, a scanner, or the like disposed outside the printing device 300. In this case, the colorimetric device 340 may be connected to the input/output I/F 80 of the terminal device 200.

Next, a color prediction model creation method for creating a color prediction model using the color prediction model creation system 1 configured as described above will be described.
FIG. 4 is a flowchart showing the procedure for creating a color prediction model, and shows steps S11 to S17, which are the operations of the terminal device 200, and steps S21 to S32, which are the operations of the server 100.

First, in step S11, the control unit 50 of the terminal device 200 causes the color measurement device 340 of the printing device 300 to measure the spectral reflectance of the print medium P based on the user's input operation, and creates the measurement result. Acquired as medium information related to the color prediction model. Then, in step S12, the control unit 50 acquires a print mode and an ink set as print information related to the color prediction model to be created based on the user's input operation. In step S13, the control unit 50 transmits the acquired medium information and print information, that is, the spectral reflectance, the print mode, and the ink set to the server 100.

Next, in step S21, the information acquisition unit 11 of the server 100 acquires the medium information and print information transmitted from the terminal device 200, that is, the spectral reflectance, the print mode, and the ink set. Thereafter, in step S22, the medium discriminator 12 inputs the acquired spectral reflectance to the medium discriminator 22, and causes the medium discriminator 22 to discriminate the type of print medium P. Then, the medium identifier output by the medium discriminator 22 or information indicating that the print medium P is unknown is acquired. Note that steps S21 and S22 correspond to an identification information acquisition step in order to acquire a medium identifier as identification information. Among these steps, step S21 corresponds to a medium information acquisition step because it acquires spectral reflectance as medium information, and step S22 corresponds to a medium discrimination step because it acquires a medium identifier using the medium discriminator 22. do.

In step S23, the medium discriminator 12 determines whether the type of print medium P is known based on the output from the medium discriminator 22. Then, when it is determined that the print medium P is known (step S23: YES), that is, when the medium identifier is output from the medium discriminator 22, the medium discriminator 12 moves the process to step S25. On the other hand, if it is determined that the print medium P is unknown (step S23: NO), that is, if the medium discriminator 22 outputs information indicating that the print medium P is unknown, the medium discriminator 12 Then, the process moves to step S24.

If it is determined that the print medium P is unknown and the process moves to step S24, the medium discrimination learning unit 13 uses the spectral reflectance acquired in step S21 as training data used for learning the medium discriminator 22. generates training data in which a new medium identifier is associated with a new medium identifier. Then, using this teacher data, the medium discriminator 22 is caused to perform machine learning to update the medium discriminator 22, and the process moves to step S27. Step S24 corresponds to a classifier learning step.

On the other hand, if the print medium P is known and the process moves to step S25, the medium discriminator 12 refers to the identification table 23 stored in the storage section 20, and in step S22, the medium discriminator 22 Based on the output medium identifier and the print mode and ink set acquired in step S21, it is determined whether or not there is a corresponding data set in the data set storage unit 21. Then, if the combination of the medium identifier output in step S22 and the print mode and ink set acquired in step S21 exists in the identification table 23, the medium discrimination unit 12 sets the data in the data set storage unit 21. It is determined that there is (step S25: YES), and the process moves to step S26. On the other hand, if the combination of the medium identifier, print mode, and ink set does not exist in the identification table 23, the medium determination unit 12 determines that there is no data set in the data set storage unit 21 (step S25: NO). The process moves to step S27.

If it is determined that there is a data set in the data set storage unit 21 and the process moves to step S26, the ink amount set selection unit 14 selects a combination of the medium identifier, print mode, and ink set in the identification table 23. The corresponding data set is identified based on the associated data set ID. Then, the ink amount set selection unit 14 acquires a plurality of ink amount sets constituting a plurality of teacher data from the specified data set as an ink amount set for the teacher data, and also configures a plurality of verification data. A plurality of ink amount sets are acquired as ink amount sets for verification data, and the process moves to step S28. Step S26 corresponds to an ink amount set acquisition step.

On the other hand, if it is determined that there is no data set in the data set storage unit 21 and the process moves to step S27, or if it is determined that the print medium P is unknown and the medium discriminator 22 is updated, the process moves to step S27. In this case, the ink amount set selection unit 14 selects a plurality of ink amount sets used for generating training data and a plurality of ink amount sets used for generating verification data according to a predetermined selection rule. Execute the process to newly select . The newly selected ink amount set here corresponds to the medium identifier output from the medium discriminator 22 in step S22 and the print mode and ink set acquired in step S21. Step S27 corresponds to an ink amount set selection step. When the ink amount set selection unit 14 completes the process of selecting an ink amount set, it moves the process to step S28.

In step S28, the ink amount set selection unit 14 transmits the plurality of ink amount sets acquired in step S26 or the plurality of ink amount sets selected in step S27 to the terminal device 200.

In step S14, the control unit 50 of the terminal device 200 acquires a plurality of ink amount sets for teacher data and a plurality of ink amount sets for verification data transmitted from the server 100. Then, in step S15, the control unit 50 causes the printing unit 330 of the printing device 300 to print a plurality of color patches for teacher data and a plurality of color patches for verification data based on the plurality of acquired ink amount sets. is printed on the print medium P. Further, the control unit 50 causes the color measurement device 340 of the printing device 300 to measure the spectral reflectance of these color patches.

In step S16, the control unit 50 sends the spectral reflectances of the plurality of color patches for teacher data and the spectral reflectances of the plurality of color patches for verification data to the server as measurement data measured by the colorimeter 340. Send to 100.

In step S29, the teacher data generation unit 15 of the server 100 acquires the measurement data transmitted from the terminal device 200, that is, the spectral reflectance. Then, in step S30, the teacher data generation unit 15 associates the ink amount set sent to the terminal device 200 in step S28 with the spectral reflectance acquired in step S29, and uses the teacher data generation unit 15 to create a color prediction model. data and verification data used to verify the color prediction model. Step S30 corresponds to a teacher data generation step.

In step S31, the color prediction model creation unit 16 creates a color prediction model by machine learning using the generated teacher data. Furthermore, the color prediction model creation unit 16 verifies the created color prediction model using the generated verification data, and if there is no problem with the verification result, the creation of the color prediction model is completed. Step S31 corresponds to a color prediction model creation step.

In step S32, the color prediction model creation unit 16 stores the created color prediction model, teacher data, and verification data as a data set in the data set storage unit 21. That is, the ink amount set newly selected in step S27 is also stored in the data set storage unit 21 as part of the teacher data and verification data. Step S32 corresponds to a data storage step. Further, the color prediction model creation unit 16 assigns a new data set ID to the stored data set and updates the identification table 23. As a result, the processing at the server 100 ends. Note that if the created color prediction model is created in step S26 using the ink amount set acquired from the data set stored in the data set storage unit 21, the same ink amount set is used. Since the dataset already exists in the dataset storage unit 21, there is no need to assign a new dataset ID or update the identification table 23.

When the data set is stored in the data set storage unit 21 of the server 100, the control unit 50 of the terminal device 200 acquires the generated color prediction model from the server 100 in step S17. From this point on, the user can utilize the acquired color prediction model to predict spectral reflectance.

As explained above, according to this embodiment, the following effects can be obtained.
(1) According to the present embodiment, when a data set generated using an ink amount set selected in the past for the same printing medium P exists in the data set storage unit 21, this ink amount set is Since it is reused to generate training data and create a color prediction model, it is possible to omit the time required to select an ink amount set, reducing the time required to generate training data and color prediction models. becomes possible.

(2) According to the present embodiment, if the data set corresponding to the medium identifier output by the medium discriminator 22 is not present in the data set storage unit 21, the data set storage unit 21 selects a new ink amount set. It becomes possible to create a color prediction model regardless of the presence or absence of the data set in 21.

(3) According to the present embodiment, when a new ink amount set is selected, the data set generated using this ink amount set is stored in the data set storage unit 21, so that the color prediction model can be used later. This ink amount set can be reused when creating an ink amount set.

(4) According to this embodiment, since the medium discriminator 22 that outputs a medium identifier based on the spectral reflectance of the print medium P is used, it becomes easy to discriminate the print medium P, that is, specify the medium identifier. .

(5) According to the present embodiment, when the print medium P is determined to be unknown by the medium discriminator 22, the medium discriminator 22 is caused to perform learning, so that the discrimination ability of the medium discriminator 22 can be improved. can.
Each of the above embodiments may be modified as follows.

In the above embodiment, the datasets stored in the dataset storage unit 21 include a color prediction model created in the past, training data used to create this color prediction model, and data used for verification of the color prediction model. However, the structure of the data set is not limited to this. The data set only needs to include at least an ink amount set for teacher data. As described above, the teacher data is data in which the ink amount set and the spectral reflectance are associated with each other, and the ink amount set can be obtained from the teacher data. Therefore, the ink amount set may be stored in the form of teacher data as in this embodiment, or may be stored as an ink amount set alone.

In the above embodiment, the color prediction model creation system 1 includes two computers, the server 100 and the terminal device 200, but it can also be configured with one computer, or three or more computers. It is also possible to distribute the processing to

In the above embodiment, a configuration is shown in which the server 100 acquires ink amount sets for teacher data and verification data from the data set stored in the data set storage unit 21 and sends them to the terminal device 200. For example, a configuration may be adopted in which the server 100 transmits teacher data and verification data to the terminal device 200, and the terminal device 200 acquires the ink amount set from the teacher data and verification data.

In the above embodiment, when the data set corresponding to the determined print medium P etc. is present in the data set storage unit 21, the server 100 transfers the ink amount set of the teacher data and verification data included in this data set to the terminal device 200. However, for example, instead of the ink amount set, a color prediction model included in the data set may be transmitted. In this aspect, the terminal device 200 acquires the transmitted color prediction model, allowing the user to try out the color prediction model.

In the embodiment described above, the server 100 generates the teacher data using the spectral reflectance acquired as measurement data from the terminal device 200, but the terminal device 200 uses the ink amount set acquired from the server 100 and the ink amount set measured by itself. The teacher data may be generated by associating the spectral reflectance with the calculated spectral reflectance, and the generated teacher data may be transmitted to the server 100.

In the above embodiment, the spectral reflectance of the print medium P is acquired from the terminal device 200 in step S21, and the medium identifier representing the type of the print medium P is output from the medium discriminator 22 in step S22. Although shown, the user may input a medium identifier to the terminal device 200 via the input device 210, and the server 100 may obtain this medium identifier from the terminal device 200. With this configuration, it is not necessary to provide the medium discriminator 22 in the server 100.

In the embodiment described above, the medium discriminator 22 discriminates the type of print medium P based on the spectral reflectance of the print medium P, but the medium information for discriminating the print medium P is limited to the spectral reflectance. Instead, the determination may be made based on physical property information other than spectral reflectance or other information. For example, spectral transmittance, reflectance distribution, an image captured by a visible light camera, thickness, moisture content, weight, friction coefficient, or ultrasonic inspection image can be used. Further, the medium discriminator 22 may be configured to discriminate the print medium P based on a plurality of pieces of information.

DESCRIPTION OF SYMBOLS 1... Color prediction model creation system, 10... Control unit, 11... Information acquisition unit, 12... Medium discrimination unit, 13... Medium discrimination learning unit, 14... Ink amount set selection unit, 15... Teacher data generation unit, 16... Color Prediction model creation unit, 20... Storage unit, 21... Data set storage unit, 22... Medium discriminator, 23... Identification table, 30... Communication unit, 50... Control unit, 60... Storage unit, 70... Communication unit, 80... Input/output I/F, 100... Server, 200... Terminal device, 210... Input device, 220... Display device, 300... Printing device, 310... Control section, 320... Input/output I/F, 330... Printing section, 340... Color measurement device, NW...network, P...printing medium.

Claims (7)

Color prediction is performed by machine learning using training data that associates an ink amount set, which is a combination of ink amounts used for printing, with the spectral reflectance of printed matter printed on a print medium with the ink amount set. A color prediction model creation method for creating a model, the method comprising:
an identification information acquisition step of acquiring identification information representing the type of the print medium;
an ink amount set acquisition step of acquiring an ink amount set corresponding to the acquired identification information;
a teacher data generation step of generating the teacher data using the acquired ink amount set;
a color prediction model creation step of creating the color prediction model by performing machine learning using the generated training data;
A color prediction model creation method comprising: The color prediction model creation method according to claim 1,
further comprising an ink amount set selection step of newly selecting an ink amount set corresponding to the identification information when there is no ink amount set corresponding to the acquired identification information;
The color prediction model creation method includes, in the teacher data generation step, generating the teacher data using the ink amount set selected in the ink amount set selection step. The color prediction model creation method according to claim 2,
A color prediction model creation method further comprising a data storage step of storing the ink amount set selected in the ink amount set selection step in a storage section. The color prediction model creation method according to any one of claims 1 to 3, comprising:
The identification information acquisition step includes a medium information acquisition step and a medium discrimination step,
The medium information acquisition step acquires medium information for identifying the print medium,
In the color prediction model creation method, the medium discriminating step acquires the identification information using a medium discriminator that receives the medium information as input and outputs the identification information. 5. The color prediction model creation method according to claim 4,
The medium classifier is a learning model created by machine learning,
further comprising a discriminator learning step of causing the medium discriminator to learn using training data in which new identification information is associated with the medium information when the medium discriminator determines that the print medium is unknown; Including, how to create a color prediction model. The color prediction model creation method according to any one of claims 1 to 3, comprising:
In the color prediction model creation method, the identification information acquisition step acquires the identification information input via an input device. Color prediction is performed by machine learning using training data that associates an ink amount set, which is a combination of ink amounts used for printing, with the spectral reflectance of printed matter printed on a print medium with the ink amount set. A color predictive model creation system that creates a model,
an identification information acquisition unit that acquires identification information representing the type of the print medium;
an ink amount set acquisition unit that acquires an ink amount set corresponding to the acquired identification information;
a teacher data generation unit that generates the teacher data using the acquired ink amount set;
a color prediction model creation unit that performs machine learning using the generated training data to create the color prediction model;
A color prediction model creation system.

Images