Estimation device, estimation method, and program

Abstract

To estimate the sign of germination in stored crops while preventing the commercial value of the crops from being impaired by the germination.SOLUTION: An estimation device comprises: an acquisition unit that acquires a target optical image obtained by photographing target crops; and an estimation unit that inputs the target optical image to a leaned model trained to, according to input of the optical image obtained by photographing the crops determined to be in a dormancy release state, output identification information indicating that the crops are in the dormancy release state, and thereby estimates weather the target crops are in the dormancy release state.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an estimation device, an estimation method, and a program.

Background Art

[0002] Conventionally, techniques for suppressing the germination of edible crops are known. For example, Patent Document 1 describes a technique for suppressing the germination of potatoes by supplying ethylene gas to stored potatoes. Further, Patent Document 2 describes a technique for suppressing the germination of potatoes by chemical treatment with pesticides such as chlorpropham.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] The above conventional technologies involve the use of raw material gases and pesticides, which may cause resistance among consumers or require special equipment and labor. Therefore, techniques for predicting the degree of future germination of potatoes without performing chemical treatment using methods such as gas chromatography-mass spectrometry (GC-MS) and near-infrared spectroscopy are also known. However, these techniques also involve destructive analysis of potato tubers or are not capable of detecting signs of tuber germination, and it is not possible to control the decrease in the commercial value of potatoes due to germination during storage using these as indicators.

[0005] The present invention has been made in consideration of such circumstances, and an object thereof is to provide an estimation device, an estimation method, and a program capable of estimating the sign of germination in a crop without impairing the commercial value of the stored crop due to germination.

Means for Solving the Problems

[0006] An estimation device according to an aspect of the present invention includes an acquisition unit that acquires a target optical image of a target crop (crop to be estimated), and a learned model that is learned to output identification information indicating that the crop is in a state of dormancy release in response to an input of a crop (reference crop used for acquiring learning data) determined to be in a state of dormancy release. An estimation unit that estimates whether the target crop is in a state of dormancy release by inputting the target optical image.

[0007] The crop and the target crop are potatoes, and the optical image and the target optical image may be optical images of the eyes of the potatoes.

[0008] The crop shown in the optical image may be determined to be in a state of dormancy release based on the relative frequency distribution of the pixel values of the T2 distribution image of the crop by MRI (Magnetic Resonance Imaging).

[0009] The relative frequency distribution may be the relative frequency distribution of the pixel values representing the T2 relaxation time in a predetermined range of the T2 distribution image.

[0010] The crop may be determined to be in a state of dormancy release based on the change in the T2 relaxation time.

[0011] The predetermined range may be the outer layer portion of the crop.

[0012] An estimation device according to an aspect of the present invention includes an acquisition unit that acquires a T2 distribution image of a crop by MRI (Magnetic Resonance Imaging), and an estimation unit that estimates whether or not the crop is in a state of dormancy release based on the relative frequency distribution of pixel values representing the T2 relaxation time in a predetermined range of the crop in the T2 distribution image.

[0013] An estimation method according to an aspect of the present invention is such that a computer acquires a target optical image of a target crop, and inputs the target optical image to a learned model that has been learned to output identification information indicating that the crop is in a state of dormancy release in response to the input of an optical image of a crop determined to be in a state of dormancy release, thereby estimating whether or not the target crop is in a state of dormancy release.

[0014] A program according to an aspect of the present invention causes a computer to acquire a target optical image of a target crop, and inputs the target optical image to a learned model that has been learned to output identification information indicating that the crop is in a state of dormancy release in response to the input of an optical image of a crop determined to be in a state of dormancy release, thereby causing the computer to estimate whether or not the target crop is in a state of dormancy release.

[0015] An estimation method according to an aspect of the present invention is such that a computer acquires a T2 distribution image of a crop by MRI (Magnetic Resonance Imaging), and estimates whether or not the crop is in a state of dormancy release based on the relative frequency distribution of pixel values representing the T2 relaxation time in a predetermined range of the crop in the T2 distribution image.

[0016] A program according to an aspect of the present invention causes a computer to acquire a T2 distribution image of a crop by MRI (Magnetic Resonance Imaging), and estimates whether or not the crop is in a state of dormancy release based on the relative frequency distribution of pixel values representing the T2 relaxation time in a predetermined range of the crop in the T2 distribution image. [Effect of the Invention]

[0017] According to the present invention, it is possible to estimate the sign of germination in a crop without impairing the commercial value of the stored crop due to germination.

Brief Description of Drawings

[0018]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Modes for Carrying Out the Invention

[0019] [First Embodiment] Hereinafter, with reference to the drawings, the estimation device 100 according to an embodiment of the present invention will be described. In the following embodiments, as an example, the case of estimating the sign of sprouting in a tuber of a potato during storage will be described, but the crop to be estimated for the sign of sprouting does not have to be a potato. The crop may be, for example, bulbous plants such as tulips, lilies, and garlic. More generally, in the present embodiment, the crop may be any crop as long as the T2 relaxation time described later decreases with growth, and it can also be applied to the estimation of the sign of sprouting of a crop without tubers.

[0020] [Overview] FIG. 1 is a diagram showing an example of the usage environment and configuration of the estimation device 100 according to the first embodiment. The estimation device 100 operates in cooperation with, for example, the camera 10 and the terminal device 20. The camera 10 is installed, for example, in a storage for storing potatoes P before shipment, and is a digital camera using a solid-state imaging device such as a CCD (Charge Coupled Device) or CMOS (Complementary Metal Oxide Semiconductor), or a smartphone or tablet terminal equipped with the function of a digital camera. The camera 10 is installed at a position where it can image the potato P stored in the storage, and transmits the image data obtained by imaging the potato P to the estimation device 100 via the network NW. Alternatively, the image data captured by the camera 10 may be manually transmitted to the estimation device 100 by the user of the estimation device 100 (for example, by connecting the camera 10 and the estimation device 100 via USB (universal serial bus). Further, the camera 10 is not limited to the image data of the entire potato P, and at least the image data of one or more eyes (concave portions) of the potato P may be transmitted to the estimation device 100.

[0021] The terminal device 20 is a computer device such as a personal computer, a smartphone, or a tablet terminal. The terminal device 20 communicates with the estimation device 100 via the network NW, receives the estimation result by the estimation device 100 as described later, and outputs it to a screen such as a display. Thereby, the user of the estimation device 100 can grasp the potatoes with signs of germination in the tubers and use them as a reference for determining the shipping time and the sales time.

[0022] The estimation device 100 is a server device such as a web server. The estimation device 100 includes, for example, an image acquisition unit 110, a germination estimation unit 120, an estimation result output unit 130, and a storage unit 150. Each of the image acquisition unit 110, the germination estimation unit 120, and the estimation result output unit 130 is realized, for example, by a hardware processor such as a CPU (Central Processing Unit) executing a program (software). Some or all of these components may be realized by hardware (including a circuit unit; circuitry) such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a GPU (Graphics Processing Unit), or may be realized by the cooperation of software and hardware. The program may be stored in advance in a storage device (a storage device having a non-transitory storage medium) such as an HDD (Hard Disk Drive) or a flash memory, or may be stored in a removable storage medium (a non-transitory storage medium) such as a DVD or a CD-ROM, and may be installed by mounting the storage medium on a drive device. The storage unit 150 is realized by a storage device such as an HDD, a flash memory, or a RAM (Random Access Memory). The storage unit 150 stores, for example, image data 152, a learned model 154, and estimation result data 156.

[0023] The image acquisition unit 110 acquires the optical image IM of the potato imaged by the camera 10 and stores it as image data 152. FIG. 2 is a diagram showing an example of the image IM acquired by the image acquisition unit 110. In the optical image IM shown in FIG. 2, the signs E1 to E4 represent the eyes of the imaged potato. Thus, the optical image IM to be processed in this embodiment includes at least one or more eyes of the imaged potato. The image acquisition unit 110 is an example of the "acquisition unit" in the claims.

[0024] The germination estimation unit 120 estimates whether the potato shown in the optical image IM is in a dormancy-release state (level 1, to be described later) by inputting the optical image IM acquired by the image acquisition unit 110 into the learned model 154. The germination estimation unit 120 stores, in the storage unit 150 as estimation result data 156, the information indicating whether the potato shown in the optical image IM output by the learned model 154 is in a dormancy-release state.

[0025] The estimation result output unit 130 transmits the estimation result by the germination estimation unit 120 to the terminal device 20 via the network NW, and the terminal device 20 displays the estimation result on, for example, a display. At this time, the estimation result output unit 130 may output information indicating whether the potato is in a dormancy-release state (that is, whether it is at level 1), or may output the state of the potato as one of a plurality of levels. Hereinafter, the details of the learned model 154 will be described. The germination estimation unit 120 is an example of the "estimation unit" in the claims.

[0026] [Details of the Learned Model 154] FIG. 3 is a diagram for explaining the relationship between the growth state of potatoes and the T2 distribution image that serves as the basis for generating the learned model 154. The upper part of FIG. 3 shows a time-series optical image of a sample in which the eye part of the potatoes stored in the warehouse after harvesting was grown, and the lower part of FIG. 3 shows the T2 distribution image (T2-weighted image) taken by an MRI (Magnetic Resonance Imaging) device corresponding to each time-series optical image. In the present embodiment, the growth state of the sample is classified into level 0 (dormant), level 1 (dormancy release), level 2 (germination start), level 3 (bud length 4 - 6 mm), and level 4 (bud length 1 cm or more).

[0027] After the state of level 2, that is, after the stage where the potato starts to germinate, the elongation of the bud progresses rapidly, leading to a decrease in the commercial value due to the shrinkage of the tuber and the reduction in yield, and at the germinating part, steroid glycoalkaloids that cause food poisoning are generated. Therefore, it is preferable to detect in advance that the potato has entered the state of level 1 and notify the user of the estimation device 100. Persons engaged in potato cultivation determine whether the potato is in a dormancy release state by visual inspection from the optical image of the potato, but it is difficult to visually distinguish between the dormant state of the potato (that is, the state of level 0) and the dormancy release state (that is, the state of level 1).

[0028] Against this background, the inventors of the present invention used an MRI device to acquire the T2 distribution image of the sample in time series and observed the change in the T2 relaxation time of the sample. As a result, as shown in the lower part of FIG. 3, for the outer layer (cortex) of the sample, a decreasing trend in the T2 relaxation time was confirmed over time. In other words, it was confirmed that the mobility of water in the outer layer decreases as germination progresses in the sample. Here, the outer layer of the sample can also be expressed as the region located outside the eyes and the bases of the buds of the potato and the vascular ring.

[0029] FIG. 4 is a diagram showing an example of a T2 distribution image at level 3 and a frequency distribution of the corresponding T2 relaxation time. The upper part of FIG. 4 represents the T2 distribution image at level 3, and the lower part of FIG. 4 represents the frequency distribution of the corresponding T2 relaxation time. The frequency distribution is obtained by calculating the T2 relaxation time for each pixel of the MRI image classified into either the outer layer or the inner layer of the sample, and aggregating the calculated T2 relaxation times (pixel values) for each of the outer layer and the inner layer. As shown in the frequency distribution of FIG. 4, it was confirmed that the frequency distribution of the pixel values constituting the outer layer has a shorter T2 relaxation time as the mode value compared to the pixels constituting the inner layer.

[0030] FIG. 5 is a diagram showing an example of a frequency distribution of the T2 relaxation time at each level of the growth state. FIG. 5 represents, for each level from level 0 to level 4, a result of creating a frequency distribution based on the T2 relaxation time of each pixel constituting the outer layer of the T2 distribution image and plotting the result, in the same manner as in the case of FIG. 4. As shown in FIG. 5, it was confirmed that as the level progresses, in other words, as germination progresses in the sample, the frequency distribution of the pixels constituting the outer layer has a shorter T2 relaxation time as the mode value.

[0031] FIG. 6 is a diagram showing an example of the result of performing principal component analysis using the frequency distribution of the T2 relaxation time. In FIG. 6, PC1 represents the first principal component axis obtained by performing principal component analysis on the frequency distribution of each sample, and PC2 represents the second principal component axis. When principal component analysis and significance test were performed on the frequency distribution of the T2 relaxation time for each of these samples, it was confirmed that for germination levels 0 to 2, the T2 relaxation time significantly decreases as the level progresses.

[0032] Based on the findings described with reference to FIGS. 3 to 6 above, in the present embodiment, an optical image of a sample determined to be in the state of level 1 based on the frequency distribution generated from the T2 distribution image is used as teacher data to generate a learned model 154. More specifically, the learned model 154 is learned to output identification information indicating whether the potato is in a state of dormancy release in response to an input of an optical image showing the eyes of the potato.

[0033] FIG. 7 is a diagram showing an example of teacher data of the learned model 154. As shown in FIG. 7, first, for each of a plurality of samples in which the eye portion of a potato is collected, an optical image and a T2 distribution image by MRI are acquired, and for the pixels constituting the outer layer of the T2 distribution image, a frequency distribution of the T2 relaxation time is created. Next, it is determined whether or not the created frequency distribution satisfies a predetermined condition (a condition characteristic of the T2 relaxation time at level 1). When it is determined that the created frequency distribution satisfies the predetermined condition, the optical image of the sample represents the eye of a potato in the state of level 1. Therefore, this optical image is associated with the label of level 1 and used as teacher data for generating the learned model 154. The optical image of a sample that satisfies any one of the examples of the predetermined conditions described below may be used as teacher data, or the optical images of samples that satisfy a plurality of the following examples of the predetermined conditions may be used as teacher data.

[0034] As a first example of the predetermined condition, the mode of the created frequency distribution is obtained, and it is determined whether or not the obtained mode falls within a predetermined range (for example, 52 to 56 ms). When it is determined that the obtained mode falls within the predetermined range, the corresponding optical image is associated with the label of level 1 and used as teacher data for generating the learned model 154. As shown in FIG. 5, since the T2 relaxation time at level 0 has 56 ms to 60 ms as the mode, this can also be expressed as a condition regarding the degree of change in the T2 relaxation time from level 0.

[0035] As a second example of the predetermined conditions, based on the mode of the frequency distribution, it is determined whether the ratio of the pixels showing pixel values smaller than the mode among all the pixels of the outer layer is equal to or greater than the first ratio, or the ratio of the pixels showing pixel values larger than the mode is equal to or less than the second ratio. When it is determined that the ratio of the pixels showing pixel values smaller than the mode among all the pixels of the outer layer is equal to or greater than the first ratio, or the ratio of the pixels showing pixel values larger than the mode is equal to or less than the second ratio, the corresponding optical image is associated with the label "level 1" and used as teacher data for generating the learned model 154. This is based on the inventors' finding that, compared to level 0, level 1 has a larger ratio of pixels showing pixel values smaller than the mode and a smaller ratio of pixels showing pixel values larger than the mode. The mode in the first example described above can vary depending on the season and the cultivation year, while the ratio of pixel values referred to in the second example shows a similar trend regardless of the season and the cultivation year, which is suitable for implementation.

[0036] As a third example of the predetermined conditions based on the above findings, the first quartile (25th percentile), the second quartile (50th percentile, i.e., the median), and the third quartile (75th percentile) are calculated from the created frequency distribution, and it is determined whether at least one value selected from the calculated first quartile, second quartile, and third quartile is equal to or less than a predetermined value. When it is determined that at least one value selected from the first quartile, second quartile, and third quartile is equal to or less than the predetermined value, the corresponding optical image is associated with the label "level 1" and used as teacher data for generating the learned model 154.

[0037] As a fourth example of the predetermined condition, a plurality of predefined statistics are calculated according to the range of T2 relaxation time of the created frequency distribution, a scatter diagram is created from any combination of the calculated statistics, and in the created scatter diagram, it is determined whether the combination of the calculated statistics is included in a predetermined range of the coordinate space. Here, the plurality of statistics represents, for example, any of median, mean, mode, standard deviation, skewness, kurtosis, second moment, third moment, etc., and the combination of the plurality of statistics according to the range of T2 relaxation time and the predetermined range are experimentally determined from potato samples that are distinguished into level 0 and level 1 in advance. Examples of combinations of a plurality of statistics include a combination of the median and skewness or kurtosis, a combination of the median and a statistic other than the mean and mode, and a combination of the standard deviation and the second moment.

[0038] In this way, a plurality of sample optical images determined to represent potato eyes in a level 1 state are prepared, and parameters of a machine learning model (e.g., a convolutional neural network) that performs image classification are adjusted (trained) so that these optical images are input and a label of level 1 is output. This makes it possible to generate a trained model 154 that inputs an optical image of a potato and outputs identification information indicating whether the potato is in a dormancy-released state.

[0039] In this embodiment, the trained model 154 is configured to have the above-mentioned input / output relationship, but the present invention is not limited to such a configuration, and the trained model 154 may be configured to output a level of a potato (i.e., any one of levels 0 to 4) in response to an input of an optical image of the potato. In this case, conditions corresponding to each level from level 0 to level 4 (for example, the conditions of the first to fourth examples above) are defined in advance, and each optical image of the sample is classified in advance into each level in accordance with the frequency distribution of the T2 relaxation time, and is used as training data.

[0040] Furthermore, in the present embodiment, for the sake of simplicity of explanation, the level determined for one eye of a certain potato is regarded as the level of the potato. However, more precisely, as shown in FIG. 2, a single potato may include a plurality of eyes. Therefore, as an example of the determination method, when at least one of the plurality of eyes is determined to be level 1, the level of the potato may also be determined to be level 1 in the same manner. Also, as another example of the determination method, among the levels determined for the plurality of eyes, the maximum level may be determined as the level of the potato.

[0041] Furthermore, in the present embodiment, as an example of the crop, the explanation is made with the potato as the subject. Therefore, based on the knowledge regarding the potato variety in which the T2 relaxation time decreases as the germination level progresses, the collection of teacher data and the generation of the learned model 154 are performed. However, the present invention is not limited to such a configuration, and even in the case of a potato variety (for example, Konafubuki) or a crop (for example, flower buds of fruit trees) in which the T2 relaxation time increases with growth, the above-described method can be applied by reversing the magnitude relationship. That is, the present invention can be more generally applied to crops in which the T2 relaxation time changes with growth.

[0042] [Flow of processing] Next, the flow of processing executed by the estimation device 100 according to the first embodiment will be described. FIG. 8 is a flowchart showing an example of the flow of processing executed by the estimation device 100 according to the first embodiment. The processing of the flowchart shown in FIG. 8 is executed, for example, at the timing when the camera 10 transmits an optical image of a potato in a dormancy-release state, which is stored in the storage, to the estimation device 100 via the network NW.

[0043] First, the image acquisition unit 110 acquires a target optical image of a potato that is the target for estimating the wake-up state (step S100). Next, the germination estimation unit 120 inputs the acquired target optical image into the learned model 154 to obtain an estimation result indicating whether the potato is in the wake-up state (step S102). Next, the estimation result output unit 130 outputs the acquired estimation result to the terminal device 20 for display (step S104). Thereby, the processing of this flowchart ends.

[0044] In the above description, the terminal device 20 outputs the estimation result output by the estimation result output unit 130 to a screen such as a display. However, some or all of the functions of the above-described camera 10 and estimation device 100 may be installed in the terminal device 20 as an application program (estimation app).

[0045] FIG. 9 is a diagram showing an example of an estimation app installed in the terminal device 20. As an example, FIG. 9 shows a case where the terminal device 20 is a smartphone and has the function of the camera 10 for imaging a potato that is the target for estimating the wake-up state. When the terminal device 20 acquires an optical image of a potato that is the target for estimating the wake-up state, an estimation result indicating whether the potato is in the wake-up state is obtained by inputting the acquired optical image into the learned model 154 incorporated in the estimation app.

[0046] When it is estimated that the potato is in a state of dormancy release, as shown in FIG. 9, the estimation application causes the terminal device 20 to display information indicating that the potato is estimated to be in a state of dormancy release. Thereby, the user can grasp whether the potato he / she owns is in a state of dormancy release or not, and can use it as a reference for determining the shipping time, sales time, and consumption time of the potato. Note that the estimation application has only an input function as the camera 10 for imaging the potato and an output function for outputting the estimation result of the dormancy release state, and the learned model 154 may be installed in an external server. In that case, the estimation application obtains the estimation result by inquiring an external server on which the learned model 154 is installed.

[0047] According to the first embodiment described above, the user of the estimation device 100 can grasp whether the potato to be estimated for the dormancy release state is in the dormancy release state or not by simply inputting an optical image into the learned model 154 without the need to obtain a T2 distribution image by MRI for the potato to be estimated for the dormancy release state. In this case, the user can utilize the estimation result by the learned model 154, for example, for visually determining the dormancy release state of the potato. Thus, according to the first embodiment, it is possible to estimate the sign of germination in the crop without impairing the commercial value of the stored crop by germination.

[0048] [Second Embodiment] The first embodiment estimates the dormancy release state of the potato to be estimated using the learned model 154 with the optical image of the potato estimated to be in the dormancy release state based on the T2 distribution image as the teacher data. However, the present invention is not limited to such a configuration, and as a second embodiment, the estimation device 200 may obtain a T2 distribution image by MRI for the potato to be estimated for the dormancy release state, and based on the obtained T2 distribution image, estimate whether the potato is in the dormancy release state or not. Hereinafter, the estimation device 200 according to the second embodiment will be described with an emphasis on the configuration different from that of the first embodiment.

[0049] FIG. 10 is a diagram showing an example of the usage environment and configuration of the estimation device 200 according to the second embodiment. The estimation device 200 is a server device such as a web server. The estimation device 200 operates in cooperation with, for example, the MRI device 12 and the terminal device 20. The estimation device 200 includes, for example, an image acquisition unit 210, a germination estimation unit 220, an estimation result output unit 230, and a storage unit 250. Each of the image acquisition unit 210, the germination estimation unit 220, and the estimation result output unit 230 is realized, for example, by a hardware processor such as a CPU (Central Processing Unit) executing a program (software). Some or all of these components may be realized by hardware (including a circuit unit; circuitry) such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a GPU (Graphics Processing Unit), or may be realized by the cooperation of software and hardware. The program may be stored in advance in a storage device (a storage device having a non-transitory storage medium) such as an HDD (Hard Disk Drive) or a flash memory, or may be stored in a removable storage medium (a non-transitory storage medium) such as a DVD or a CD-ROM, and may be installed by mounting the storage medium on a drive device. The storage unit 250 is realized by a storage device such as an HDD, a flash memory, or a RAM (Random Access Memory). The storage unit 250 stores, for example, image data 252 and estimation result data 254.

[0050] The MRI device 12 captures a plurality of tomographic images of potatoes that are targets for wake-up state estimation, and transmits the captured plurality of tomographic images to the estimation device 200 via the network NW. The image acquisition unit 210 calculates the T2 value from the plurality of tomographic images captured by the MRI device 12, acquires a T2 distribution image, and stores it as image data 252. Alternatively, the MRI device 12 may transmit the captured plurality of tomographic images to the terminal device 20 via the network NW, and the terminal device 20 may calculate the T2 value and acquire the T2 distribution image, and then transmit the acquired T2 distribution image to the estimation device 200.

[0051] Based on the T2 distribution image acquired by the image acquisition unit 210, the germination estimation unit 220 estimates whether the potato from which the T2 distribution image was acquired is in a wake-up state. More specifically, the germination estimation unit 220 creates a frequency distribution of the T2 relaxation time for the pixels constituting the outer layer of the T2 distribution image, and determines whether the created frequency distribution satisfies a predetermined condition. Here, the predetermined condition is any one or a combination of the first to fourth examples of the predetermined conditions described above. The germination estimation unit 120 stores the determination result in the storage unit 250 as estimation result data 254.

[0052] The estimation result output unit 230 transmits the estimation result by the germination estimation unit 220 to the terminal device 20 via the network NW, and the terminal device 20 displays the estimation result on, for example, a display. At this time, similar to the first embodiment, the estimation result output unit 130 may output information indicating whether the potato is in a wake-up state (that is, whether it is at level 1), or may output the state of the potato as any one of a plurality of levels.

[0053] [Flow of processing] Next, the process flow executed by the estimation device 200 according to the second embodiment will be described. FIG. 11 is a flowchart showing an example of the process flow executed by the estimation device 200 according to the second embodiment. The process of the flowchart shown in FIG. 11 is executed, for example, at the timing when the estimation device 200 acquires a plurality of tomographic images of potatoes to be the estimation target in the wake-up state from the MRI device 12, or at the timing when the terminal device 20 acquires the T2 distribution image of the potatoes.

[0054] First, the image acquisition unit 210 acquires a T2 distribution image of the potatoes to be the estimation target in the wake-up state (step S200). Next, the germination estimation unit 220 creates a frequency distribution of the T2 relaxation time for the pixels constituting the outer layer of the acquired T2 distribution image (step S202). Next, the germination estimation unit 220 determines whether the created frequency distribution satisfies a predetermined condition (a condition characteristic of the T2 relaxation time at level 1) (step S204).

[0055] When it is determined that the created frequency distribution satisfies the predetermined condition, the estimation result output unit 230 outputs information indicating that the potatoes are estimated to be in the wake-up state to the terminal device 20 for display (step S206). On the other hand, when it is determined that the created frequency distribution does not satisfy the predetermined condition, the estimation result output unit 230 outputs information indicating that the potatoes are not estimated to be in the wake-up state to the terminal device 20 for display (step S208). Thereby, the process of this flowchart ends.

[0056] As described above, the embodiments for carrying out the present invention have been described using the embodiments. However, the present invention is not limited to such embodiments, and various modifications and substitutions can be made without departing from the gist of the present invention.

Explanation of reference numerals

[0057] 10 Camera 12 MRI device 20 Terminal device 100 Estimation device 110 and 210 Image Acquisition Units 120 and 220 Bud Deduction Units 130 and 230 Deduction Result Output Units 150 and 250 Memory Units 152 and 252 Image Data 154 Trained Model 156 and 254 Deduction Result Data

Claims

1. An acquisition unit that acquires a target optical image of a target crop, An estimation unit that estimates whether the target crop is in a dormancy - released state by inputting the target optical image into a learned model that has been learned to output identification information indicating that the crop is in a dormancy - released state in response to the input of an optical image of a crop determined to be in a dormancy - released state, An estimation device.

2. The crop and the target crop are potatoes, and the optical image and the target optical image are optical images of the eyes of the potatoes, The estimation device according to claim 1.

3. The crop imaged in the optical image is determined to be in the dormancy release state based on the relative frequency distribution of the pixel values of the T ２ distribution image obtained by MRI (Magnetic Resonance Imaging) of the crop. The estimation device according to claim 1.

4. The relative frequency distribution is the T ２ relative frequency distribution of pixel values representing relaxation times in a predetermined range of the T ２ distribution image. The estimation device according to claim 3.

5. The crop is the T ２ determined to be in a state of dormancy release based on the change in relaxation time The estimation device according to claim 4.

6. The predetermined range is the outer layer of the crop, The estimation device according to claim 4.

7. Obtaining unit for acquiring T distribution image of crop by MRI (Magnetic Resonance Imaging) ２ and a distribution image acquisition unit The above-mentioned T ２ Among the distribution images, based on the relative frequency distribution of the pixel values representing the relaxation time T in a predetermined range of the crop, ２ an estimation unit that estimates whether or not the crop is in a state of dormancy release, and is provided with An estimation device.

8. A computer, Acquires a target optical image of a target crop, Estimates whether the target crop is in a dormancy - released state by inputting the target optical image into a learned model that has been learned to output identification information indicating that the crop is in a dormancy - released state in response to the input of an optical image of a crop determined to be in a dormancy - released state, An estimation method.

9. Causes a computer to, Acquire a target optical image of a target crop, Estimate whether the target crop is in a dormancy - released state by inputting the target optical image into a learned model that has been learned to output identification information indicating that the crop is in a dormancy - released state in response to the input of an optical image of a crop determined to be in a dormancy - released state, A program.

10. A computer, Obtain the T distribution image of the crop by MRI (Magnetic Resonance Imaging). ２ and The above-mentioned T ２ Among the distribution images, based on the relative frequency distribution of the pixel values representing the relaxation time T in a predetermined range of the crop, ２ estimate whether the crop is in a state of dormancy release. An estimation method.

11. Causes a computer to, Obtain the T distribution image of the crop by MRI (Magnetic Resonance Imaging). ２ And The above-mentioned T ２ Among the distribution images, based on the relative frequency distribution of the pixel values representing the relaxation time of T in a predetermined range of the crop, ２ estimate whether the crop is in a state of dormancy release. A program.

Images