Information processor, method of controlling information processor and program

Abstract

To accurately presume a type of a recording medium.SOLUTION: An information processor comprises acquiring means that acquires a characteristic value of a recording medium fed, registering means that registers types of recording media as objects of presumption by the number within a predetermined range, presuming means that presumes a type of a recording medium fed from among a plurality of types of recording media registered in the registering means using a presuming section, based on the characteristic value acquired by the acquiring means, change means that changes the type of the recording medium registered in the registering means, and output means that outputs an instruction to update the presuming section using a characteristic value corresponding to the changed type of the recording medium.SELECTED DRAWING: Figure 10

Description

[Technical Field]

[0001] The present disclosure relates to a technique for estimating the type of a recording medium. [Background technology]

[0002] In the commercial and industrial printing market, the uses of output are diverse, including CAD line drawings, posters, artwork, and signage. Therefore, a wide variety of recording media with diverse characteristics is used to suit these uses. As the number of recording media types increases, the task of users selecting the type of recording media fed into a recording device becomes more cumbersome. Recent recording devices are being equipped with functions that improve usability by automatically estimating the type of recording media fed. However, because this automatic estimation function makes an estimation based on information about pre-defined recording media, it is unable to estimate unknown print media. Therefore, it is desirable to be able to expand the types of recording media available to suit the user's intended use.

[0003] Patent Document 1 discloses a technology for estimating the type of recording medium using a trained model that has been trained in advance on spectral information of unprinted areas of the printing medium and an identifier indicating the type of printing medium. In Patent Document 1, when an unknown recording medium is estimated, the trained model is retrained using the spectral information of the unknown printing medium and the identifier indicating the type of printing medium, thereby updating the trained model and expanding the types of printing media that can be estimated. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2022-078426 Summary of the Invention [Problem to be solved by the invention]

[0005] In updating the learning model by re-learning as described in Patent Document 1, the more types of recording media are used, the more complex the learning model becomes and the lower the estimation accuracy becomes.In addition, as the amount of learning data increases, the capacity of the learning model increases, which puts pressure on the memory area in the device.

[0006] An object of the present disclosure is to estimate the type of recording medium with high accuracy. [Means for solving the problem]

[0007] An information processing device according to one aspect of the present disclosure is characterized by having an acquisition means for acquiring characteristic values ​​of a recording medium to be fed, a registration means for registering a number of types of recording media to be estimated within a predetermined range, an estimation means for estimating the type of the recording medium to be fed from among multiple types of recording media registered in the registration means based on the characteristic values ​​acquired by the acquisition means using an estimation unit, a change means for changing the type of recording media registered in the registration means, and an output means for outputting an instruction to update the estimation unit using characteristic values ​​corresponding to the changed type of recording media. [Effects of the Invention]

[0008] According to the present disclosure, the type of recording medium can be estimated with high accuracy. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a perspective view illustrating an example of a recording apparatus. [Figure 2] FIG. 2 is a cross-sectional view showing an example of a main part of a recording apparatus. [Figure 3] FIG. 2 is a block diagram showing an example of the configuration of a control system of the printing apparatus. [Figure 4] FIG. 2 is a cross-sectional view showing an example of the periphery of a media sensor. [Figure 5] 10 is a flowchart illustrating an example of a sheet type estimation process. [Figure 6] FIG. 10 is a diagram illustrating an example of characteristics of a sheet. [Figure 7]FIG. 10 is a diagram illustrating an example of a sheet type estimation table. [Figure 8] Schematic diagram of a DNN. [Figure 9] This is a list of recording media uses, categories, and types. [Figure 10] 10 is a flowchart illustrating an example of a manual selection process. [Figure 11] 10 is an example of a screen when selecting an application in manual selection. [Figure 12] 10 is an example of a screen when a category is selected in manual selection. [Figure 13] FIG. 10 is a diagram showing an example in which a list is displayed in order of usage history when individual selection is performed in manual selection. [Figure 14] FIG. 10 is a diagram showing an example in which a list is displayed in order of usage history when individual selection is performed in manual selection. [Figure 15] 10 is a flowchart illustrating an example of an operation of a first process of automatic selection. [Figure 16] 10 is a flowchart illustrating an example of an operation of a second process of automatic selection. [Figure 17] FIG. 10 is a diagram illustrating an example of a trained model update process. DETAILED DESCRIPTION OF THE INVENTION

[0010] Preferred embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. The following embodiments do not limit the scope of the present disclosure, and not all combinations of features described in the following embodiments are necessarily essential to the solutions of the present disclosure. The same components are denoted by the same reference numerals. Furthermore, the relative positions, shapes, and the like of the components described in the embodiments are merely examples, and are not intended to limit the scope of the present disclosure to those alone.

[0011] In the following description of the embodiments, "recording" not only refers to the formation of meaningful information such as characters and figures, but also broadly includes the formation of images, designs, patterns, etc. on a sheet. In this embodiment, a roll sheet is assumed as the sheet, but cut paper, cloth, plastic film, etc. may also be used. Furthermore, "ink" should be broadly interpreted and refers to a liquid that can be applied to a sheet to form an image, design, pattern, etc., or to process the sheet, or to be used for ink processing.

[0012] <<First Embodiment>> <Inkjet recording device> FIG. 1 is a perspective view showing an inkjet recording apparatus as an example of a recording apparatus 101 that performs industrial and commercial printing in this embodiment. FIG. 2 is a cross-sectional view showing an example of the main parts of the recording apparatus 101. FIG. 3 is a block diagram showing an example of the control configuration of the recording apparatus 101. The configuration of the recording apparatus 101 will be described below with reference to FIGS. 1 to 3.

[0013] In this embodiment, as will be described later, processing using a trained model is performed. In this embodiment, processing using a trained model is performed in the recording device 101. That is, the recording device 101 is an information processing device that uses a trained model. Note that the information processing device that uses a trained model is not limited to the recording device 101. A server (for example, a cloud server) that can exchange various types of data with the recording device 101 may also be used as the information processing device that uses a trained model.

[0014] Hereinafter, the transport direction in which the sheet S is transported in the recording apparatus 101 is referred to as the +Y direction. The direction in which the recording head 204 ejects ink onto the sheet S is referred to as the -Z direction. The direction in which the recording head 204 moves from the standby position is referred to as the +X direction.

[0015] The recording device 101 rotatably holds a roll sheet R, which is a sheet S wound into a roll. By rotating the roll sheet R with a roll drive motor 308, the sheet S is supplied from the roll sheet R to a conveyance roller 203. The conveyance roller 203 can rotate while holding the sheet S. By rotating the conveyance roller 203 with a conveyance roller drive motor 309, the sheet S is conveyed to a position where a recording head 204 can record on the sheet S. The recording head 204 is mounted on a carriage (not shown) and configured to move back and forth in the X direction. While the recording head 204 is moving in the X direction, ink is ejected from the recording head 204 onto the conveyed sheet S, thereby recording an image on the sheet S. The sheet S with the recorded image is discharged from a discharge section located downstream of the recording head 204 in the conveyance direction and stacked in a basket 103.

[0016] The operation panel 102 is an interface module that accepts various operations from the user. The user can use various switches or a touch panel provided on the operation panel 102 to make various settings for the recording device 101. The various settings for the recording device 101 include, for example, settings for the size and type of the sheet S. The operation panel 102 also displays estimation results, which will be described later.

[0017] A sheet detection sensor 202 is disposed upstream of the conveyance roller 203 in the conveyance direction. When the sheet detection sensor 202 detects that a user has supplied a sheet S from a rolled sheet R, the conveyance operation of the sheet S is initiated. The conveyance operation of the sheet S is performed by synchronously driving a roll drive motor 308 and a conveyance roller drive motor 309. At this time, the recording apparatus 101 can estimate the type of the sheet S by estimating the sheet type, which will be described later. Details will be described later.

[0018] In the conveying direction, a media sensor 206 and an ultrasonic transmitter 207 are arranged upstream of the sheet detection sensor 202. The media sensor 206 is arranged above the sheet S in the direction of gravity (Z direction), and the ultrasonic transmitter 207 is arranged below the sheet S in the direction of gravity. The media sensor 206 and the ultrasonic transmitter 207 are used to estimate the sheet type, which will be described later.

[0019] An image is recorded on the sheet S as follows. First, the recording device 101 performs a transport operation to transport the sheet S to a position facing the recording head 204. Next, a recording operation is performed in which the recording head 204 is scanned in a cross direction intersecting (orthogonal to) the transport direction of the sheet S while ejecting ink, thereby recording an image on an area of ​​the sheet S corresponding to the recording head 204. Next, after the sheet S is transported a predetermined distance, ink is ejected while the recording head 204 is scanned in the cross direction. In this way, a desired image is recorded on the sheet S by alternately performing the transport operation of the sheet S and the image recording operation. The sheet S on which the image has been recorded is sequentially transported downstream of the recording head 204 in the transport direction. The transported sheet S is cut by a cutter 205 provided in the discharge section. The cut sheet S is stacked in the basket 103.

[0020] FIG. 3 is a block diagram showing an example of the configuration of a control system in the recording apparatus 101. The recording apparatus 101 has an operation panel 102, a recording head 204, a CPU 301, a sensor control unit 302, and an input / output interface (IF) 303. The recording apparatus 101 also has a USB port 304, memory 305, a motor control unit 306, and RAM 320. The recording apparatus 101 also has a sheet detection sensor 202, a media sensor 206, an ultrasonic transmitter 207, and a carriage encoder 307. The recording apparatus 101 also has a roll drive motor 308, a conveyance roller drive motor 309, a carriage drive motor 310, a lift drive motor 311, a cutter drive motor 312, and a media sensor lift motor 313. The memory 305 stores a program 351 and a trained model 352. The trained model 352 includes a trained model for coarse classification, a trained model for detailed classification 1, and a trained model for detailed classification 2, which will be described later. In this embodiment, when predetermined data is input to the trained model 352, a predetermined estimation result is output from the trained model 352. In other words, the trained model 352 is an estimation unit that performs estimation.

[0021] The motor control unit 306 controls each drive motor according to a program 351 stored in the memory 305. The roll drive motor 308 rotates the spool 201 to transport the sheet S from the roll sheet R in the transport direction. The transport roller drive motor 309 rotates the transport roller 203 to transport the sheet S to a position facing the print head 204. The transport roller drive motor 309 is provided with an encoder that detects the amount of rotation to detect the transport distance of the sheet S. The amount of rotation of the encoder can be measured to detect the transport distance of the sheet S. The carriage drive motor 310 rotates a carriage belt (not shown) to move a carriage (not shown) and the print head 204 mounted on the carriage. The lift drive motor 311 moves the carriage and print head 204 up and down. The cutter drive motor 312 drives the cutter. The media sensor lift motor 313 lifts and lowers the media sensor 206.

[0022] Various setting information and the like is input to CPU 301 via input / output IF 303 by user operation from operation panel 102 or a PC connected to USB port 304 or a LAN port (not shown). The input information is stored in memory 305. CPU 301 can read out the information stored in memory 305 as needed and perform various processes on the read out information. In other words, CPU 301 includes a processing unit that executes various processes.

[0023] The CPU 301 controls the carriage encoder 307, the sheet detection sensor 202, the media sensor 206, and the ultrasonic transmitter 207 via the sensor control unit 302, and obtains output data from each unit. The CPU 301 executes various controls based on input from the carriage encoder 307, the sheet detection sensor 202, and the media sensor 206. The CPU 301 controls the carriage encoder 307, the sheet detection sensor 202, the media sensor 206, and the ultrasonic transmitter 207 via the sensor control unit 302, and obtains information. The CPU 301 also executes various controls based on input from the carriage encoder 307, the sheet detection sensor 202, and the media sensor 206. The RAM 320 is used as a temporary work area.

[0024] In this embodiment, a process of estimating the type of sheet using a machine learning trained model will be described.

[0025] <Assuming the seat type> The operation of estimating the type of sheet S in this embodiment will be described with reference to FIGS. 4 to 7. FIG. 4 is a cross-sectional view showing an example of the media sensor 206 and its surroundings. FIG. 5 is a flowchart showing an example of a process for estimating the type of sheet S from paper feed. FIG. 6 is a diagram showing an example of data obtained by detecting the characteristics of sheet S. FIG. 6 is an example of data obtained by detecting the characteristics of sheet S using the media sensor 206 and the ultrasonic transmitter 207. FIG. 7 is a diagram showing an example of an estimation table for the type of sheet S in this embodiment. The estimation table in FIG. 7 defines the type of sheet S corresponding to the output value y of a trained model described below. In this embodiment, the types of sheet S to be estimated are nine types, recording medium 1 to recording medium 9. In other words, the trained model 352 that estimates the type of sheet S is configured to estimate, as the type of sheet S, the recording medium that is most suitable for the input feature values ​​from among recording medium 1 to recording medium 9. The trained model 352 is described as being stored in memory 305, but it may also be provided outside the recording device 101, and the CPU 301 of the recording device 101 may use the trained model provided outside it.

[0026] The process of the flowchart shown in Figure 5 is realized by the CPU 301 of the recording device 101 reading out the program 351 stored in the memory 305 or the like into the RAM 320 and executing it. Note that some or all of the functions of the steps in Figure 5 may be realized by hardware such as an ASIC or electronic circuit. The symbol "S" in the explanation of each process indicates a step in the flowchart (the same applies to the flowcharts hereinafter in this specification). The process shown in Figure 5 is started, for example, when a user sets the rolled sheet S in the recording device 101. Alternatively, the process may be started when it is detected that a predetermined operation by the user has been input on the operation panel 102 after the user has set the rolled sheet S in the recording device 101.

[0027] In S501, sheet S is fed. Specifically, CPU 301 detects that a user has set rolled sheet R in the recording apparatus 101. Then, CPU 301 rotates rolled sheet R using roll drive motor 308. As a result, sheet S is fed from rolled sheet R to conveyance rollers 203. Then, sheet detection sensor 202, which is disposed upstream of conveyance rollers 203, detects that sheet S has reached conveyance rollers 203. When sheet detection sensor 202 detects that sheet S has reached conveyance rollers 203, CPU 301 stops driving roll drive motor 308. At the position where sheet detection sensor 202 detects sheet S, sheet S has been conveyed to a position opposite media sensor 206 and ultrasonic transmitter 207. Then, CPU 301 proceeds to the process of S502.

[0028] In S502, the CPU 301 performs sensing. That is, the CPU 301 measures the characteristics of the sheet S by controlling the media sensor 206 and the ultrasonic transmitter 207 via the sensor control unit 302. As shown in FIG. 4, the media sensor 206 has a CIS (Contact Image Sensor) 401 and a microphone 402. A roller 403 is disposed opposite the CIS 401. The ultrasonic transmitter 207 is disposed opposite the microphone 402. The CPU 301 lowers the media sensor 206, which is separated from the sheet S, using the media sensor lift motor 313, thereby enabling the CIS 401 and the roller 403 to sandwich the sheet S. By sandwiching the sheet S, the characteristics of the sheet S can be stably measured. The CPU 301 reads the surface image of the sheet S using the CIS 401 while conveying the sheet S again with the sheet S sandwiched between the CIS 401 and the roller 403. When the sensing is completed, the CPU 301 moves the media sensor 206 away from the sheet S by raising the media sensor 206 using the media sensor lifting motor 313.

[0029] 6(a) and (b) show examples of surface images of sheet S acquired using CIS 401. FIGS. 6(c) and (d) show examples of electrical signals of ultrasonic waves transmitted through sheet S acquired using ultrasonic transmitter 207 and microphone 402.

[0030] The CIS 401 is a line sensor extending in the width direction of the sheet S and acquires one-dimensional (one line) image data. With the sheet S sandwiched between the CIS 401 and rollers 403, the CPU 301 synchronizes the roll drive motor 308 and the conveyance roller drive motor 309 to acquire image data of the sheet S using the CIS 401. By reading the image of the sheet S using the CIS 401 while conveying the sheet S in this manner, two-dimensional image data such as those shown in FIGS. 6(a) and 6(b) can be acquired. FIG. 6(a) is an example of a surface image of Japanese paper, and FIG. 6(b) is an example of a surface image of synthetic paper. In FIGS. 6(a) and 6(b), the CIS direction corresponds to the width of the CIS 401 (the width in the X direction intersecting the sheet S), and the conveyance direction corresponds to the conveyance amount of the sheet S measured by the CIS 401. While an example of measurement using a one-dimensional sensor as the CIS 401 is shown here, the surface image of the sheet S may also be measured using a two-dimensional sensor. Furthermore, while the CIS 401 is measuring the surface image, the ultrasonic transmitter 207 and the microphone 402 (sound collection sensor) obtain ultrasonic electrical signals as shown in FIGS. 6(c) and 6(d). FIG. 6(c) is an example of an ultrasonic electrical signal transmitted through Japanese paper, and FIG. 6(d) is an example of an ultrasonic electrical signal transmitted through synthetic paper. In this embodiment, an example is described in which the ultrasonic electrical signal is obtained in conjunction with the measurement of the surface image by the CIS 401, but this is not limiting. The measurement of the surface image and the acquisition of the ultrasonic electrical signal may be performed separately. Furthermore, the ultrasonic electrical signal may be obtained without conveying the sheet S. Thereafter, the CPU 301 proceeds to the process of S503.

[0031] In S503, the CPU 301 derives characteristic values. That is, the CPU 301 derives a feature amount related to the surface information of the sheet S and a feature amount related to the cross-sectional information of the sheet S from the characteristics of the sheet S measured in S502 described above, using a feature amount derivation method previously stored in the memory 305. That is, the CPU 301 derives a feature amount related to the surface information of the sheet S and a feature amount related to the cross-sectional information of the sheet S from the surface image of the sheet S and the electrical signal of the ultrasonic waves.

[0032] The CPU 301 derives three feature amounts related to the surface information of the sheet S from the surface image of the sheet S as shown in FIGS. 6(a) and 6(b). The first feature amount is brightness. Brightness is derived as the average value of all pixel values. The second feature amount is unevenness in the CIS direction. The unevenness in the CIS direction is derived as the average value of the absolute values ​​of the differences between adjacent pixel values ​​in the CIS direction. The third feature amount is unevenness in the transport direction. The unevenness in the transport direction is derived as the average value of the absolute values ​​of the differences between adjacent pixel values ​​in the transport direction.

[0033] The CPU 301 derives three feature quantities related to the cross-sectional information of the sheet S from the electrical signal of the ultrasonic waves that have passed through the sheet S, as shown in FIGS. 6(c) and 6(d). The first feature quantity is Peak 1. Peak 1 is derived as the maximum voltage value during the period from time t1 to time t2. The second feature quantity is Peak 2. Peak 2 is derived as the maximum voltage value during the period from time t2 to time t3. The third feature quantity is Peak 3. Peak 3 is derived as the maximum voltage value during the period from time t3 to time t4. Note that although the maximum voltage value is described here, it may also be the minimum voltage value. The cross-sectional information of the sheet S corresponds to information such as the thickness and basis weight of the sheet.

[0034] In this way, in S503, the CPU 301 can derive six feature amounts relating to the surface information and cross-sectional information of the sheet S from the characteristics of the sheet S measured in S502.

[0035] In this embodiment, Japanese paper and synthetic paper are used as examples. An example of the relationship between the feature quantities of the two is shown below. Synthetic paper has higher brightness than Japanese paper. Japanese paper has higher unevenness in the CIS direction than synthetic paper. Japanese paper has higher unevenness in the transport direction than synthetic paper. For example, a whiter, flatter sheet S has higher brightness. Furthermore, the greater the unevenness of the surface, the greater the unevenness in both the CIS direction and the transport direction. Note that some sheets have either a longitudinal or transverse grain, and only one type of unevenness is greater. The thicker the sheet, the smaller the peak value of the ultrasonic electrical signal. Japanese paper is thicker than synthetic paper. Therefore, the peak value is larger for synthetic paper than for Japanese paper. Even with the same thickness, the peak value varies depending on the cross section (the material or density of the sheet). Specifically, the peak value tends to be smaller when a material (medium) with a higher acoustic impedance is used. The CPU 301 then proceeds to step S504.

[0036] In S504, the CPU 301 classifies the type of the sheet S. That is, the CPU 301 estimates the type of the sheet S from the six feature amounts derived in S503 using the trained model of rough classification stored in advance in the memory 305 and the estimation table shown in FIG. 7. To estimate the type of the sheet S, the CPU 301 acquires an output value y output from the trained model of rough classification. The trained model of rough classification receives the six feature amounts related to the sheet S derived in S503 and outputs the probability of the type of the sheet S being estimated as an array for each type of sheet S. Each element of the output array is associated with a type of sheet S. That is, the index of each element of the output array is associated with a respective sheet type. In this embodiment, the index of the element in the output array with the highest probability is set as the output value y of the trained model of rough classification. The type of sheet S associated with the output value y becomes the estimation result.

[0037] As shown in the estimation table in FIG. 7 , in S504, when the output value y obtained by inputting the feature quantities of sheet S into a learned model for rough classification (described later) is 0, CPU 301 estimates the type of sheet S as recording medium 1. Similarly, when the output value y is 1, CPU 301 estimates the type of sheet S as recording medium 2, when the output value y is 2, when the output value y is 3, when the output value y is 4, when the output value y is 5, the estimated type of sheet S is recording medium 6 and recording medium 7. In this case, recording medium 6 and recording medium 7 are collectively classified as the first recording medium group. In other words, when the output value y is 5, CPU 111 estimates the type of sheet S as the first recording medium group. Similarly, when the output value y is 6, the estimated type of sheet S is recording medium 8 and recording medium 9. In this case, recording medium 8 and recording medium 9 are collectively classified as the second recording medium group. That is, when the output value y is 6, the CPU 111 estimates that the type of the sheet S is the second recording medium group. After the process of S504, the CPU 301 proceeds to the process of S505.

[0038] Thus, in S504, when the output value y output from the trained model for rough classification is between 0 and 4, the CPU 301 can uniquely estimate the type of sheet S as recording medium 1 to 5, respectively. On the other hand, when the output value y is 5, the CPU 301 estimates the type of sheet S as recording medium 6 and recording medium 7. That is, in this case, the CPU 301 estimates that the type of sheet S is classified into the first recording medium group in the rough classification. Similarly, when the output value y is 6, the CPU 301 estimates that the type of sheet S is classified into the second recording medium group in the rough classification. That is, when the output value y is 5 and 6, the CPU 111 cannot uniquely estimate the type of sheet S, and uniquely estimates the repair of sheet S as the type of a recording medium group consisting of multiple recording medium types. In this embodiment, the characteristic values ​​corresponding to recording media 6 and 7 are close to each other within a predetermined range. Similarly, the characteristic values ​​corresponding to recording media 8 and 9 are close to each other within a predetermined range. Therefore, estimation using the trained model for rough classification is configured to obtain estimation results that collectively represent multiple recording media.

[0039] In S505, the CPU 301 determines whether to perform first detailed classification in S507 or second detailed classification in S508, which will be described later, based on the estimation result of the type of sheet S in S504. Specifically, when the estimation result in S504 corresponds to the first recording medium group, the CPU 301 determines to perform first detailed classification and proceeds to processing of S507. When the estimation result in S504 corresponds to the second recording medium group, the CPU 301 determines to perform second detailed classification and proceeds to processing of S508. When the estimation result in S504 does not correspond to either the first recording medium group or the second recording medium group, the CPU 301 proceeds to processing of S506. In S506, the CPU 111 displays the estimation result on the operation panel 102.

[0040] In S507, the CPU 301 estimates the type of sheet S from the six feature quantities derived in S503 described above, using a trained model for first detailed classification stored in advance in the memory 305 and the estimation table shown in FIG. 7. That is, in S507, the CPU 301 estimates the type of sheet S using the same six feature quantities as those used in estimating the type of sheet S in S504. However, the output value y acquired in S507 is the output value of a trained model different from the trained model used in S504. In other words, if the trained model used in S504 is the first trained model, the trained model used in S507 is a second trained model different from the first trained model. Here, a different trained model means, at least, a model trained using different data as the data used for training. To estimate the type of sheet S, the CPU 301 acquires the output value y of the trained model, as in S504 described above.

[0041] 7, in S507, when the output value y obtained by inputting the feature amount of sheet S into the trained model for first detailed classification is 0, CPU 301 estimates the type of sheet S as recording medium 6. Similarly, when the output value y is 1, CPU 301 estimates the type of sheet S as recording medium 7. Thereafter, CPU 301 proceeds to the processing of S506. In this way, in S507, CPU 301 can uniquely estimate the type of sheet S, which was estimated as being in the first recording medium group in S504, as recording medium 6 or recording medium 7.

[0042] In S508, CPU 301 estimates the type of sheet S from the six feature amounts derived in S503 described above, using a trained model for second detailed classification stored in advance in memory 305 and the estimation table shown in FIG. 7. That is, in S508, CPU 301 estimates the type of sheet S using the same six feature amounts as those used in estimating the type of sheet S in S504. However, the output value y obtained in S508 is the output value of a trained model different from those in S504 and S507. In S508, in order to estimate the type of sheet S, the output value y of the trained model for second detailed classification is obtained, as in S504 and S507 described above.

[0043] 7, in S508, when the output value y obtained by inputting the feature amount of sheet S into the trained model of the second detailed classification is 0, CPU 301 estimates the type of sheet S as recording medium 8. Similarly, when the output value y is 1, CPU 301 estimates the type of sheet S as recording medium 9. Thereafter, CPU 301 proceeds to the processing of S506. In this way, in S508, CPU 301 can uniquely estimate the type of sheet S, which was estimated as being of the second recording medium group in S504, as recording medium 8 or recording medium 9.

[0044] As described above, in S506 following S507 and S508, the CPU 301 also displays the estimation result on the operation panel 102. When the process of S506 ends, the CPU 301 proceeds to S509.

[0045] In S509, the CPU 301 creates training data to be used to update the trained model. The training data is a type of learning parameter used to train the trained model. This training data includes the feature values ​​of the sheet S acquired in S503 and information about the type of the sheet S (estimated result) acquired in S506. In S510, the CPU 301 saves the created training data in a specified save destination. Even for the same type of recording medium, characteristic values ​​may not be completely identical due to individual differences. Therefore, even if the type of the sheet S is appropriately estimated, the characteristic values ​​used at that time and the estimated result can be saved as training data and used to update the trained model, thereby optimizing the trained model as appropriate. Note that the training data created and saved here is data used in each trained model. For example, assume that the estimated result in S506 is recording medium 6. In this case, the estimated result indicating the first recording medium group is used as the training data for the trained model used for rough classification. On the other hand, the training data for the trained model used in the first detailed classification uses an estimation result indicating that the recording medium is 6. In this way, when detailed classification is performed, training data for detailed classification is created and saved separately from training data for coarse classification.

[0046] The storage destination of the training data is preferably the same as the storage destination of the device or system that updates the trained model. This is to reduce various costs required for updating the trained model. Therefore, in a configuration in which the trained model is updated inside the recording device 101, the storage destination of the training data may be memory 305 shown in FIG. 3. In a configuration in which the trained model is updated by an external device other than the recording device 101, the storage destination of the training data is a machine learning device (not shown). The machine learning device may be a user's PC (Personal Computer) or a server PC. It may also be a server system consisting of multiple servers. When transferring training parameters from the recording device 101 to an external storage device, the transfer may be via USB port 304 or LAN (Local Area Network) port 314.

[0047] <Pre-trained model> 8 is a schematic diagram of a DNN (Deep Neural Network). The three trained models 352 of this embodiment will be described with reference to FIG.

[0048] The trained model 352 of this embodiment is a DNN as shown in the schematic diagram of FIG. 8. The DNN receives data at an input layer 801, propagates the data through an intermediate layer 802, and outputs the data at an output layer 803. Each layer has multiple nodes indicated by circles. The input data is propagated toward the output layer while being weighted, biased, and the like between the nodes in each layer. Adjusting parameters such as weighting and bias so that a specified output is obtained for a specified input is referred to as training a model. The trained model is also called a trained model. As mentioned above, the dataset of input data used to train a model and the output data associated with it is called training data.

[0049] In this embodiment, the input data for the training data are six feature amounts prepared for each type of sheet S. The six feature amounts are the same as the feature amounts derived in S503 described above, namely, brightness, unevenness in the CIS direction, unevenness in the conveying direction, peak 1, peak 2, and peak 3. In this embodiment, the same feature amounts are used in the training of all models. In addition, the input layer 801 of the trained model has six nodes. Each node receives an input of a feature amount.

[0050] In this embodiment, the output data of the training data is an integer value indicating the type of sheet S to be estimated. When there are two types of sheets S to be estimated, the output data of the training data is 0 or 1. When there are seven types of sheets S to be estimated, the output data of the training data is 0 to 6. When actually training a model, integer values ​​converted into one-hot vectors are used. In this embodiment, a different combination of sheet S types is used for each model to be trained. In addition, the output layer 803 of the trained model has the same number of nodes as the types of sheets S to be estimated. Each node outputs the probability that the sheet S is of each sheet type. If the output of the trained model is viewed as an array, each element of the output array can be considered to be the probability that it is of each sheet type. If each element of the output array is associated with each sheet type, the index of each element is also associated with each sheet type. In this embodiment, the estimation result of the trained model is the index of the element with the highest probability.

[0051] The first trained model is a trained model for rough classification used in S504 described above. The types of sheets S to be estimated are all recording media from recording medium 1 to recording medium 9. However, given the difficulty of estimating all recording media with high accuracy using a single trained model, types of sheets S with similar characteristics are grouped together in advance. In this embodiment, recording media 6 and recording media 7 are grouped together as a first recording media group. Similarly, recording media 8 and recording media 9 are grouped together as a second recording media group. The output data of the training data for rough classification is an integer value indicating the type of sheet S, with recording media 1 being 0, recording media 2 being 1, recording media 3 being 2, recording media 4 being 3, recording media 5 being 4, the first recording media group being 5, and the second recording media group being 6.

[0052] The trained model trained using the above training data is referred to as the trained model for rough classification. In the trained model for rough classification, as shown in Figure 7, when feature quantities corresponding to recording medium 1 to recording medium 9 are input, 0 to 6 are output as the respective estimation results. When feature quantities corresponding to recording medium 6 or recording medium 7 are input, 5 is output as the estimation result. When feature quantities corresponding to recording medium 8 or recording medium 9 are input, 6 is output as the estimation result.

[0053] The second trained model is the trained model of the first detailed classification used in S507 described above. The types of sheet S to be estimated are recording medium 6 and recording medium 7. The output data of the training data of the first detailed classification is an integer value indicating the type of sheet S, with recording medium 6 being 0 and recording medium 7 being 1.

[0054] The trained model trained using the training data is referred to as the trained model for the first detailed classification. As shown in Figure 7, when a feature corresponding to recording medium 6 or recording medium 7 is input, the trained model for the first detailed classification outputs 0 or 1 as the estimation result, respectively.

[0055] The third trained model is a trained model for the second detailed classification used in S508 described above. The types of sheet S to be estimated are recording medium 8 and recording medium 9. The output data of the training data for the second detailed classification is an integer value indicating the type of sheet S, with recording medium 8 being 0 and recording medium 9 being 1.

[0056] The trained model trained using the training data is defined as the trained model for the second detailed classification. As shown in Figure 7, when a feature corresponding to recording medium 8 or recording medium 9 is input, the trained model for the second detailed classification outputs 0 or 1 as the respective estimation result.

[0057] The above three trained models are generated by a machine learning device (not shown). The machine learning device is, for example, a PC. The machine learning device can record the trained models and the calculation methods required for estimation using the trained models in memory 305 via USB port 304, input / output IF 303, and CPU 301. CPU 301 measures the characteristics of sheet S using media sensor 206 and ultrasonic transmitter 207, derives feature amounts related to surface information and cross-sectional information of sheet S from the measurement data, and inputs the derived feature amounts into the trained model to estimate the type of sheet S.

[0058] In the examples described so far, estimation is performed using a trained model created by a machine learning device (not shown) based on training data from a predetermined recording medium. That is, in the example of FIG. 7, estimation is performed using a trained model trained to estimate recording media 1 to 9. On the other hand, recording device 101 may be equipped with a recording medium of a different type from the predetermined recording medium. In such a case, it is necessary to update the trained model for the newly equipped recording medium.

[0059] Here, taking FIG. 7 as an example, one possible method is to update the trained model by adding new recording media, such as recording media 10 and 11, to the recording media to be estimated in addition to recording media 1 through 9. However, as the number of recording media types increases, i.e., as the number of possible values ​​for the output value y of the trained model increases, the parameters of the trained model become more complex. This may result in a larger capacity for the trained model or a decrease in operating speed. On the other hand, by limiting the types of recording media within a certain range, i.e., by limiting the possible values ​​for the output value y of the trained model to a predetermined value, such phenomena can be avoided. For example, in the example estimation table of FIG. 7, seven types of recording media are registered as the types of recording media to be estimated, including recording media groups as one type. Furthermore, as described in the rough estimation classification, the fewer the number of estimation targets, the more accurate the estimation. Although the recording device 101 of this embodiment can be equipped with more than 100 types of recording media, the number of recording media actually used by users tends to be within a predetermined range (e.g., 10 types).

[0060] In this embodiment, taking the above points into consideration, a process for updating a trained model will be described based on the user's usage trends of the recording medium. A method for changing the recording medium to be estimated and a method for updating the trained model will be described below. In this embodiment, the description will be given assuming that the recording medium to be estimated is changed by the recording device 101, and that the work of updating the trained model is performed by a machine learning device (not shown).

[0061] <How to change the recording medium of the estimated target> A method for changing the recording medium to be estimated will be described with reference to Figs. 9 to 16. When changing the recording medium to be estimated, a process for selecting the recording medium to be estimated is performed. The user can select the method for selecting the recording medium from either manual selection or automatic selection using the operation panel 102. The process for changing the recording medium by manual selection is called the first change process, and the process for changing the recording medium by automatic selection is called the second change process.

[0062] FIG. 9 is a list of print uses, categories, and types of recording media. FIG. 10 is a flowchart showing an example of manual selection processing. FIG. 11 is an example of a screen when use selection is performed in manual selection. FIG. 12 is an example of a screen when category selection is performed in manual selection. FIG. 13 is a diagram showing an example of a list displayed in order of usage history when individual selection is performed in manual selection. FIG. 14 is a diagram showing an example of a list displayed in order of usage history when individual selection is performed in manual selection. FIG. 15 is a flowchart showing an example of the operation of a first process of automatic selection. FIG. 16 is a flowchart showing an example of the operation of a second process of automatic selection.

[0063] The list shown in FIG. 9 lists the types of recording media that can be estimated in the recording device 101, and is stored in, for example, the memory 305. This list can be updated as needed. The example shown in FIG. 9 is merely an example and is not limited to this example. An example in which the recording media to be estimated is changed from the list shown in FIG. 9 will be described below. In FIG. 9, the types of recording media corresponding to the categories are linked. The categories also correspond to print uses. For example, one print use may correspond to multiple categories. Specifically, if the print use is "poster," the category corresponds to all of "plain paper," "coated paper," and "film paper." As shown in FIG. 9, "print use" and "category" are each classifications in groups that include multiple types of recording media.

[0064] First, the manual selection process will be described. When the user selects "manual" as a method for changing the recording medium on a change screen (not shown) displayed on the operation panel 102 for changing the recording medium to be estimated, the process shown in FIG. 10 starts.

[0065] The processing of the flowchart shown in Fig. 10 is realized by the CPU 301 of the recording device 101 reading out a program 351 stored in memory 305 or the like into RAM 320 and executing it. Note that some or all of the functions of the steps in Fig. 5 may be realized by hardware such as an ASIC or electronic circuit. The same applies to the automatic selection flowcharts shown in Figs. 15 and 16.

[0066] In the manual selection process shown in FIG. 10, the user selects according to the printing purpose, a batch selection by recording medium category, or an individual selection by recording medium type in response to a specification made on the operation panel 102.

[0067] In S1001, the CPU 301 determines whether the user has specified that the selection be made using the print purpose. For example, the CPU 301 displays a screen on the operation panel 102 that prompts the user to specify whether the selection be made using the print purpose, and makes the determination based on an operation instruction from the user. If the user has specified that the selection be made using the print purpose, the CPU 301 proceeds to S1004. If the user has not specified that the selection be made using the print purpose, the CPU 301 proceeds to S1002.

[0068] In S1002, CPU 301 determines whether the user has specified that selection be made using the category of the recording medium. For example, CPU 301 displays a screen on operation panel 102 that prompts the user to specify whether selection be made using the category, and makes the determination based on the user's operation instruction. If the user has specified that selection be made using the category of the recording medium, CPU 301 proceeds to S1004. If the user has not specified that selection be made using the category of the recording medium, CPU 301 proceeds to S1003.

[0069] In S1003, CPU 301 determines that the user has designated individual selection using the type of recording medium, and proceeds to S1004. While the present example describes an example in which S1001, S1002, and S1003 are processed sequentially, this is not limiting. For example, the designation processing corresponding to S1001, S1002, and S1003 may be performed based on a single designation operation by the user via operation panel 102.

[0070] In S1004, the CPU 301 displays a selection screen according to the user's specifications in S1001, S1002, and S1003.

[0071] If selection by purpose is specified in S1001, the CPU 301 displays a selection screen, for example, as shown in FIG. 11. In FIG. 11, "poster," "CAD," "photo," and "sign" are displayed on the operation panel 102 as print purposes. When the user selects a print purpose displayed on the operation panel 102 on this selection screen, the recording media assigned to each print purpose are selected all at once, as shown in FIG. 9. For example, if "photo" is selected on the selection screen of FIG. 11, the CPU 301 selects coated paper 1-2 and glossy paper 1-4, as shown in FIG. 9. Note that the selection screen of FIG. 11 may be configured to allow the user to select multiple print purposes. The selection screen of FIG. 11 includes check boxes, allowing the user to specify multiple purposes. For example, in a configuration in which up to two print purposes can be selected, if the print purposes of "poster" and "CAD" are selected, the CPU 301 selects plain paper 1-3, coated paper 1-2, and film paper 1-3, as shown in FIG. 9. In this manner, the user can manually select a recording medium suitable for the print purpose. That is, the user can freely select a recording medium suitable for the printing purpose. This selection method is suitable, for example, when the recording device 101 is used for a purpose other than the printing purpose.

[0072] If selection by category is specified in S1002, the CPU 301 displays a selection screen, for example, as shown in FIG. 12. In FIG. 12, "plain paper," "coated paper," "glossy paper," "film paper," and "vinyl chloride paper" are displayed on the operation panel 102 as categories of recording media. When the user selects a category displayed on the operation panel 102 on this selection screen, a recording medium assigned to each category is selected, as shown in FIG. 9. Note that a configuration may be adopted in which multiple categories can be selected. For example, in a configuration in which up to two categories can be selected, if "plain paper" and "coated paper" are selected, plain paper 1-3 and coated paper 1-2 are selected, as shown in FIG. 9. In this way, the user can select a recording medium appropriate for the recording medium category. This selection method is suitable, for example, when a recording medium of a category different from the category that has been used so far is to be used in the recording device 101.

[0073] If individual selection of recording medium types is specified in S1003, the CPU 301 displays a selection screen such as that shown in FIG. 13 or 14. FIG. 13 is a selection screen in which recording medium types are listed in category order on the operation panel 102. FIG. 14 is a selection screen in which recording medium types are displayed with priority given to recording medium types with a usage history. Both FIGS. 13 and 14 show how plain paper 1, coated paper 1-2, and glossy paper 1 are individually selected. In this way, the user can select multiple recording media to be estimated from the displayed recording medium types.

[0074] In this embodiment, regardless of whether the selection method of "use," "category," or "type" shown in FIG. 9 is used, the number of types of recording media to be estimated is not limited. One type of recording media may be selected, or multiple types of recording media may be selected. Here, an example has been described in which the user selects the type of recording media to be estimated in the recording device 101. That is, an example has been described in which the type of recording media selected in FIGS. 11 to 14 becomes the type of recording media to be estimated in the recording device 101. That is, an example has been described in which changing the recording media involves changing all of the types of recording media currently used as estimation targets in the recording device 101. However, this example is not limiting. For example, the user may previously select unnecessary types of recording media from the types of recording media currently used as estimation targets in the recording device 101 via a selection screen (not shown). Then, instead, the types of recording media selected in FIGS. 11 to 14 may be added as types of recording media to be estimated. Furthermore, a configuration may be adopted in which the types of recording media selected in FIGS. 11 to 14 are added as types of recording media to be estimated without selecting unnecessary types of recording media. In either case, as explained above, if the number of types of recording media exceeds a predetermined range, various costs of the trained model may increase or the estimation accuracy may decrease. Therefore, it is preferable to change the recording media to be estimated so that the number of types of recording media does not exceed the predetermined range. Therefore, for example, if the number of types of recording media resulting from manual selection exceeds the predetermined range, a warning may be issued to the user. Note that an example of the predetermined range is 2 to 10, but of course, this is just an example and is not limited to this range.

[0075] The above is an example of the manual selection process. Next, the automatic selection process will be described. Unlike manual selection, automatic selection is a process in which the CPU 301 determines the type of recording medium to be estimated in accordance with predetermined conditions without a selection instruction from the user. In the automatic selection process of this embodiment, the CPU 301 automatically selects the recording medium to be estimated based on the usage history of the recording medium in the recording device 101. Here, the usage history may be information based on the count value of the number of times a recording medium has been fed by the recording device 101. The timing for executing the automatic selection process may be, for example, when the number of times paper has been fed exceeds a specific number in the usage history. Furthermore, the CPU 301 may periodically execute the automatic selection process using a clock function inside the recording device 101. In this embodiment, two types of automatic selection process, a first process and a second process, will be described.

[0076] First, the first process of automatic selection will be described with reference to Figures 9 and 15. The first process is a process of inferring the user's print purpose from the usage history and determining the recording medium to be inferred based on the print purpose.

[0077] The first process will be described using the flowchart in Figure 15. In S1501, CPU 301 obtains the number of times each recording medium has been attached (count value) from the usage history. In S1502, CPU 301 classifies each candidate recording medium, whose attachment count was obtained in S1501, by "use" shown in Figure 9, and determines the total number of times the recording medium has been attached for each use. For example, if the attachment counts for the recording media obtained in S1501 were 5 times for glossy paper 1, 3 times for glossy paper 2, 2 times for coated paper 1, and 2 times for film paper 1, the total for each use will be 10 times for "photo," which is the sum of glossy paper 1, glossy paper 2, and coated paper 1. "Poster" will be 4 times, which is the sum of coated paper 1 and film paper 1. "CAD" will be 2 times for film paper 1 only, and "sign" will be 2 times for film paper only.

[0078] Next, in S1503, the CPU 301 compares the sums of the count values ​​determined in S1502 and estimates the use with the highest count value as the print use. In the example described above, "photo" is counted 10 times, "poster" is counted 4 times, "CAD" is counted 2 times, and "signature" is counted 2 times. If the recording device 101 is configured to allow up to two print uses to be selected, "photo" and "poster" are selected. Next, in S1504, the CPU 301 determines the type of recording medium corresponding to the estimated use as the recording medium to be estimated. For example, if "photo" and "poster" are selected as the print uses, glossy paper 1-4, coated paper 1-2, and film paper 1-3 are determined as the recording media to be estimated.

[0079] In the above example, in S1503, the total sum of the count values ​​is compared to estimate the print use. However, it is also possible to calculate a ratio from each count value and estimate the one with the highest ratio as the print use.

[0080] In the above example, the print purpose is automatically estimated from the usage history. However, the category may be automatically estimated from the usage history. That is, in S1502, the printer may be classified by the "category" shown in FIG. 9, in S1503, the frequently used category may be estimated, and in S1504, the recording media belonging to that category may be automatically determined. For example, if the count values ​​acquired in S1501 are Glossy Paper 1 five times, Glossy Paper 2 three times, Coated Paper 1 two times, and Film Paper 1 two times, the totals for each category are Glossy Paper eight times, Coated Paper two times, and Film Paper one time. In S1503, the CPU 301 compares the count values ​​acquired in S1503. For example, if up to two categories can be selected, the top-ranked "Glossy Paper" and "Coated Paper" are estimated as the categories. In S1504, the CPU 301 determines the types of recording media corresponding to the estimated category as candidate recording media to be estimated. For example, if "glossy paper" and "coated paper" are selected as categories, then the printing media to be estimated are determined to be glossy paper 1 to 4 and coated paper 1 and 2. The above is an example of the first process.

[0081] In the first process, a recording medium corresponding to a purpose or category based on the usage history is automatically determined as a recording medium to be estimated. Therefore, the type of an unused recording medium may be determined as a recording medium to be estimated.

[0082] Next, an example of the second automatic selection process will be described with reference to the flowchart shown in Fig. 16. The second automatic selection process is a process for automatically and individually selecting the type of each recording medium from the usage history and determining it as an estimation target.

[0083] In S1601, CPU 301 obtains the number of times each recording medium has been installed from the usage history. Next, in S1602, CPU 301 selects as tentative candidates those recording media that have been installed a predetermined number of times or more. For example, if the same type of recording media has been installed three or more times, that recording media is selected as the tentative candidate. Next, in S1603, CPU 301 determines whether the number of tentative candidates selected in S1602 is less than a predetermined number. If it is less than the predetermined number, the process proceeds to S1604, where CPU 301 determines the types of the tentative candidate recording media as the types of recording media to be estimated. For example, if the predetermined number is 10 and the number of tentative candidates actually selected is nine, then in S1604 the types of the nine candidate recording media are determined as the types of recording media to be estimated.

[0084] On the other hand, if the number of tentative candidates is equal to or greater than the predetermined number in S1603, CPU 301 proceeds to S1605. In S1605, CPU 301 changes the candidate conditions for selecting a tentative candidate. For example, CPU 301 adds "1" to the number of times a recording medium has been attached when selected as a tentative candidate, returns to S1602, and performs tentative candidate selection again. In this case, in S1602 again, recording media of the same type that have been attached four or more times will be selected as tentative candidates. In this way, the process is repeated until the condition in S1603 is met, and finally, in S1604, the estimated target recording medium is determined.

[0085] In the second process, the recording medium to be estimated is automatically determined based on the usage history of each individual recording medium. Therefore, unlike the first process, the second process is configured so that the type of unused recording medium is not determined as the type of recording medium to be estimated. Therefore, the second process is useful for optimization in situations where, for example, unused recording media will not be used.

[0086] As described above, by using the first and second automatic selection processes, it is possible to automatically determine the estimation target of the recording medium based on the usage history. As described above, the system may be configured so that the user is prompted to select both the first and second processes when a selection between manual selection and automatic selection is received from the user. Alternatively, the system may be configured so that the user is prompted to select either the first or second process after receiving the automatic selection.

[0087] <Updating a trained model> FIG. 17 is a diagram showing an example of a trained model update process. FIG. 17(a) is a transition diagram of a trained model update process. FIG. 17(b) is a flowchart showing an example of a trained model update process in a trained model update system. In this embodiment, a series of trained model update processes are performed using the recording device 101 and the machine learning device 170. The step numbers shown in FIG. 17(a) and the step numbers shown in FIG. 17(b) indicate the same steps. Hereinafter, a trained model update method used in S504 of FIG. 5 will be described with reference to FIG. 17.

[0088] Steps S1701 and S1704 are processes performed by the recording device 101. Steps S1702 and S1703 are processes performed by the machine learning device 170.

[0089] In S1701, the recording device 101 transmits to the machine learning device 170 information indicating the type of recording medium to be estimated, determined by the manual setting or automatic setting described above. That is, the recording device 101 transmits to the machine learning device 170 information indicating the type of recording medium to be registered as an estimation target. In S1702, the machine learning device 170 reads training data that matches the transmitted type of recording medium to be estimated. The machine learning device 170 stores training data for predetermined types of recording media. This training data stores information on the features and type of the recording media derived by the same method as in S503 and S506 described above. The machine learning device 170 also stores the training data saved in S510. Therefore, in S1702, the machine learning device 170 reads, from the information on recording media previously stored in the machine learning device 170, information that matches the estimation target sent in S1701 as training data.

[0090] Next, in S1703, the machine learning device 170 updates the trained model using the acquired training data. The training model may be updated by a machine learning method using the DNN (Deep Neural Network) described above in FIG. 8. The input data for the training data of the DNN is input with the feature values ​​of each recording medium. The output data for the training data of the DNN is set with information indicating the type of each recording medium. In this way, the machine learning device updates the trained model by relearning the training data related to the recording medium to be estimated. Finally, in S1704, after detecting the completion of the update of the trained model in S1703, the recording device 101 performs a process of replacing the trained model in the recording device 101.

[0091] As described above, the trained model used in S504 is a trained model used for rough classification. Therefore, the training data may be processed so that the results of recording media with feature differences within a predetermined range, i.e., recording media with similar feature values, are estimated as a group of recording media.

[0092] Also, while an example of updating the trained model used in S504 has been described here, it is possible to update the trained model used for detailed classification in S507 or S508 in a similar manner.

[0093] <Example of usage> Next, an example of a specific usage pattern to which the above-described embodiment is applied will be described. First, when a new recording device 101 is installed, the type of recording medium to be estimated is determined by manual selection by the user. Thereafter, as the recording device 101 is used, the types of recording media to be estimated are divided into used recording media and unused recording media. Therefore, at an appropriate time, re-learning is performed by automatic selection based on the usage history so that frequently used recording media are used as recording media to be estimated. Conversely, it can be said that the re-learning process is performed so as to exclude infrequently used recording media types from the types to be estimated.

[0094] Furthermore, during use, a completely new recording medium may be attached to the recording device 101. In the above-described process for displaying the estimation result in S506, an example in which the probability of the classification result is output has been described. However, if the probability is below a predetermined value, an error may be output as the estimation result in S506. When a completely new recording medium is attached, the user manually selects a different recording medium to be estimated in response to the output of the error, and re-learning is performed based on this change. As described above, the machine learning device 170 stores training data for each recording medium, and a trained model that has been re-trained using training data for the recording medium manually selected by the user is applied to the recording device 101. Then, by performing the estimation process again, it becomes possible to appropriately estimate the newly attached recording medium.

[0095] As described above, according to this embodiment, the type of recording medium can be estimated with high accuracy. That is, in this embodiment, by limiting the recording media to be estimated to a predetermined range of numbers of recording media, the complexity of the trained model can be suppressed, and the recording medium can be easily estimated. In addition, the amount of information to be trained can be reduced, and the capacity of the trained model can be reduced.

[0096] <<Other embodiments>> The above-described embodiments are merely examples of implementing the present disclosure, and are not intended to limit the present disclosure. For example, the present disclosure may be applied not only to a recording device that ejects ink onto a sheet to form an image, but also to a scanner that reads an image on a sheet, a post-processing machine that processes a sheet, or the like.

[0097] The estimation of the sheet type is not limited to being performed by the CPU 301 mounted on the recording apparatus 101, but may also be performed by a scanner, a post-processing machine, a PC, or the like.

[0098] The number of detailed classifications (second estimation) shown in S507 and S508 is not limited to two, but may be one or more. For example, if the types of sheets S that are likely to be erroneously estimated in the rough classification (first estimation) of S504 are divided into five groups, five detailed classifications corresponding to each group may be provided. Furthermore, the number of recording medium types in the detailed classification corresponding to each group is not limited to two, but may be two or more. For example, if there are three types of sheets S that are likely to be erroneously estimated in the rough classification, and these three types can be accurately estimated in the detailed classification, the three types may be classified into one group in the rough classification.

[0099] The machine learning device 170 may be mounted on the recording device 101. The trained model may also be generated by the recording device 101. Training data used to generate the trained model may be stored in the memory 305.

[0100] The above-mentioned feature amounts are not limited to six, and the color or thickness of the sheet S may also be used. Furthermore, in the above-mentioned embodiment, the type of sheet S is estimated from six feature amounts, but this is not limited to this. For example, the type of sheet S may be estimated from at least one feature amount related to the surface information of the sheet S and at least one feature amount related to the cross-sectional information of the sheet S. Furthermore, data acquired by the media sensor 206 or the like may be used as input data without deriving feature amounts. The output data of the training data is not limited to the sheet type, and the sheet model name or a sheet name defined by the user may also be used.

[0101] The trained model may be recorded outside the recording device 101. For example, it may be recorded on a PC or the like connected via the input / output IF 303 and USB port 304 of the recording device 101. Furthermore, the trained model is not limited to a DNN, and may be a decision tree or the like. Furthermore, in the above-described embodiment, an example has been described in which the trained model is applied to rough classification and detailed classification to estimate the type of the sheet S, but this is not limited to this. For example, the trained model may be applied to either rough classification or detailed classification. Specifically, the type of the sheet S may be estimated by applying the trained model to rough classification and deriving the output value y in detailed classification. In other words, the type of the sheet S may be estimated by combining the trained model and the output value y.

[0102] In the above-described embodiment, a configuration using a trained model for coarse classification and a trained model for fine classification has been described, but the present invention is not limited to such an example. When changing the type of recording medium to be estimated, the recording medium to be estimated may be determined so that the number of recording media falls within a predetermined range. The trained model may then be updated so that the thus-determined recording medium can be estimated. For example, when operating using a trained model for coarse classification and a trained model for fine classification, changing the recording medium to be estimated may result in the need to use the trained model for fine classification. That is, the type of recording medium may be uniquely estimated using only the trained model for coarse classification. Conversely, when the type of recording medium can be uniquely estimated using only the trained model for coarse classification, changing the recording medium to be estimated may result in the need for a trained model for fine classification. Alternatively, when the type of recording medium can be uniquely estimated using only the trained model for coarse classification, changing the recording medium to be estimated may result in the type of recording medium being uniquely estimated using only the trained model for coarse classification. In this way, the above-described embodiment is also applicable to a configuration in which a trained model for coarse classification, i.e., a single trained model, is used. Furthermore, even in a configuration in which one trained model is used, as described above, the recording medium can be estimated with high accuracy by training the trained model to estimate the type of recording medium within a predetermined number range.

[0103] In the above-described embodiment, the explanation of FIG. 5 described an example in which training data is created in S509 and stored in S510. The explanation of FIG. 17 also described an example in which training data corresponding to each recording medium is stored in advance. To improve learning accuracy, it is preferable to create training data in S509 and store the training data in S510, as explained in the embodiment. However, the trained model may be updated using pre-stored training data. That is, a configuration may be adopted in which S509 and S510 are not performed.

[0104] In the above-described embodiment, an example in which both manual selection and automatic selection are performed when changing the recording medium to be estimated is described, but this is not limiting. Only one of manual selection and automatic selection may be performed.

[0105] The present disclosure can also be realized by supplying a program that realizes one or more functions of the above-described embodiments to a system or device via a network or a storage medium, and having one or more processors in the computer of the system or device read and execute the program. It can also be realized by a circuit (e.g., ASIC) that realizes one or more functions.

[0106] The disclosure of the present embodiment includes configurations typified by the following information processing device example, information processing device control method example, and program example.

[0107] <Configuration 1> an acquisition means for acquiring characteristic values ​​of a recording medium to be fed; a registration means for registering a predetermined number of types of recording media to be estimated; an estimation unit that estimates the type of the recording medium to be fed from among a plurality of recording medium types registered in the registration unit based on the characteristic value acquired by the acquisition unit; a change means for changing the type of recording medium registered in the registration means; an output means for outputting an instruction to update the estimation unit using a characteristic value corresponding to the changed type of recording medium; An information processing device comprising:

[0108] <Configuration 2> 2. The information processing apparatus according to configuration 1, wherein the change means performs a first change process to change the type of the recording medium based on an operation by a user.

[0109] <Configuration 3> 3. The information processing device according to configuration 2, wherein the change means is configured to accept an instruction to change the type of recording medium in units of a group including a plurality of types of recording medium in the first change process.

[0110] <Configuration 4> 3. The information processing apparatus according to configuration 2, wherein the change means is configured to accept an instruction to change the type of each recording medium individually in the first change process.

[0111] <Configuration 5> 5. The information processing device according to any one of configurations 1 to 4, wherein the change means performs a second change process to change the type of the recording medium based on a usage history of the recording medium in the information processing device.

[0112] <Configuration 6> The information processing device according to configuration 5, wherein the change means, in the second change process, changes the type of recording medium on a group basis based on usage history on a group basis including multiple types of recording medium.

[0113] <Configuration 7> 6. The information processing apparatus according to configuration 5, wherein the change means changes the type of each recording medium individually based on the usage history of each type of recording medium in the second change process.

[0114] <Configuration 8> 8. The information processing device according to any one of configurations 1 to 7, wherein the change means changes the types of the recording media so that the number of types of recording media falls within the predetermined range.

[0115] <Configuration 9> The information processing device according to any one of configurations 1 to 8, wherein the estimation unit is a trained model trained based on training data including input data corresponding to the characteristic value and output data indicating the type of recording medium corresponding to the input data.

[0116] <Configuration 10> The information processing device according to configuration 9, wherein the output means outputs an instruction to update the trained model using training data corresponding to the type of recording medium changed by the change means.

[0117] <Configuration 11> 11. The information processing device according to configuration 9 or 10, wherein the teacher data includes characteristic values ​​acquired by the acquisition means and used in the estimation means, and estimation results by the estimation means.

[0118] <Configuration 12> The information processing device according to any one of configurations 1 to 11, characterized in that the acquisition means acquires features relating to surface information of the recording medium to be fed and features relating to cross-sectional information of the recording medium to be fed, and acquires the characteristic value based on the acquired features.

[0119] <Configuration 13> The information processing device according to configuration 12, characterized in that the acquisition means acquires features related to the surface information from a sensor that acquires a surface image of the recording medium, and acquires features related to cross-sectional information of the recording medium from a sensor that acquires an electrical signal of ultrasonic waves that have passed through the recording medium.

[0120] <Configuration 14> 14. The information processing apparatus according to configuration 12 or 13, wherein the surface information includes information regarding at least one of the brightness and the unevenness of the recording medium to be fed.

[0121] <Configuration 15> 15. The information processing apparatus according to any one of configurations 12 to 14, wherein the cross-section information includes information about at least one of the thickness and basis weight of the recording medium to be fed.

[0122] <Configuration 16> 16. The information processing device according to any one of configurations 1 to 15, further comprising a display means for displaying the result estimated by the estimation means.

[0123] <Configuration 17> the estimation unit is configured to estimate a group of recording media including a first recording medium and a second recording medium as one type of recording medium; The information processing device according to any one of configurations 1 to 16, characterized in that when the estimation result by the estimation unit indicates the group of recording media, the estimation means estimates the type of the recording medium to be fed from the group of recording media including the first recording medium and the second recording medium based on the characteristic value acquired by the acquisition means, using a second estimation unit different from the estimation unit.

[0124] <Configuration 18> The information processing device described in configuration 17, characterized in that the second estimation unit is a trained model trained using training data corresponding to the types of recording media included in the recording medium group including the first recording medium and the second recording medium.

[0125] <Configuration 19> The information processing device according to configuration 17 or 18, characterized in that, when the changed type of recording medium includes a type of recording medium included in the recording medium group, the output means outputs an instruction to update the second estimation unit using characteristic values ​​corresponding to the type of recording medium included in the recording medium group.

[0126] <Configuration 20> obtaining a characteristic value of the recording medium to be fed; a step of registering a predetermined number of types of recording media to be estimated; a step of estimating the type of the recording medium to be fed from among the plurality of registered recording medium types using an estimation unit based on the acquired characteristic value; changing the type of the registered recording medium; outputting an instruction to update the estimation unit using a characteristic value corresponding to the changed type of recording medium; 1. A method for controlling an information processing device, comprising:

[0127] <Configuration 21> 21. A program for causing a computer to execute the control method for an information processing device according to claim 20. [Explanation of symbols]

[0128] 101 Recording device 301 CPU 305 memory

Claims

1. an acquisition means for acquiring characteristic values ​​of a recording medium to be fed; a registration means for registering a predetermined number of types of recording media to be estimated; an estimation unit that estimates the type of the recording medium to be fed from among a plurality of recording medium types registered in the registration unit based on the characteristic value acquired by the acquisition unit; a change means for changing the type of recording medium registered in the registration means; an output means for outputting an instruction to update the estimation unit using a characteristic value corresponding to the changed type of recording medium; An information processing device comprising:

2. 2. The information processing apparatus according to claim 1, wherein the change unit performs a first change process to change the type of the recording medium based on an operation by a user.

3. 3. The information processing apparatus according to claim 2, wherein the change unit is configured to accept an instruction to change the type of recording medium in units of a group including a plurality of types of recording medium in the first change process.

4. 3. The information processing apparatus according to claim 2, wherein the change unit is configured to accept an instruction to change the type of each recording medium individually in the first change process.

5. 2. The information processing apparatus according to claim 1, wherein the change unit performs a second change process to change the type of the recording medium based on a usage history of the recording medium in the information processing apparatus.

6. 6. The information processing apparatus according to claim 5, wherein the change unit changes the type of recording media in units of groups based on a usage history of the groups including a plurality of types of recording media in the second change process.

7. 6. The information processing apparatus according to claim 5, wherein said change means changes the type of each recording medium individually in said second change process based on the usage history of each type of recording medium.

8. 2. The information processing apparatus according to claim 1, wherein said change means changes the type of said recording medium so that the number of types of recording medium falls within said predetermined range.

9. 2. The information processing device according to claim 1, wherein the estimation unit is a trained model trained based on training data including input data corresponding to the characteristic value and output data indicating the type of recording medium corresponding to the input data.

10. The information processing device according to claim 9 , wherein the output means outputs an instruction to update the trained model using training data corresponding to the type of recording medium changed by the change means.

11. 10. The information processing apparatus according to claim 9, wherein the teacher data includes characteristic values ​​used in the estimation means and acquired by the acquisition means, and an estimation result by the estimation means.

12. 2. The information processing apparatus according to claim 1, wherein the acquisition means acquires a feature amount relating to surface information of the recording medium to be fed and a feature amount relating to cross-sectional information of the recording medium to be fed, and acquires the characteristic value based on the acquired feature amount.

13. The information processing device according to claim 12, characterized in that the acquisition means acquires features related to the surface information from a sensor that acquires a surface image of the recording medium, and acquires features related to cross-sectional information of the recording medium from a sensor that acquires an electrical signal of ultrasound that has passed through the recording medium.

14. 13. The information processing apparatus according to claim 12, wherein the surface information includes information regarding at least one of brightness and unevenness of the recording medium to be fed.

15. 13. The information processing apparatus according to claim 12, wherein the cross-section information includes information about at least one of the thickness and the basis weight of the recording medium to be fed.

16. 2. The information processing apparatus according to claim 1, further comprising a display unit that displays the result of estimation by said estimation unit.

17. the estimation unit is configured to estimate a group of recording media including a first recording medium and a second recording medium as one type of recording medium; 17. The information processing device according to claim 1, wherein when the estimation result by the estimation unit indicates the group of recording media, the estimation means estimates the type of the recording medium to be fed from the group of recording media including the first recording medium and the second recording medium based on the characteristic value acquired by the acquisition means using a second estimation unit different from the estimation unit.

18. 18. The information processing device according to claim 17, wherein the second estimation unit is a trained model trained using training data corresponding to the types of recording media included in the group of recording media including the first recording medium and the second recording medium.

19. The information processing device according to claim 17, characterized in that, when the changed type of recording medium includes a type of recording medium included in the recording medium group, the output means outputs an instruction to update the second estimation unit using characteristic values ​​corresponding to the type of recording medium included in the recording medium group.

20. obtaining a characteristic value of the recording medium to be fed; a step of registering a predetermined number of types of recording media to be estimated; a step of estimating the type of the recording medium to be fed from among the plurality of registered recording medium types using an estimation unit based on the acquired characteristic value; changing the type of the registered recording medium; outputting an instruction to update the estimation unit using a characteristic value corresponding to the changed type of recording medium; 1. A method for controlling an information processing device, comprising:

21. A program for causing a computer to execute the method for controlling an information processing device according to claim 20.

Images