Information processing device, information processing method, and control program

Abstract

This makes it possible to estimate the progression of plant diseases that show visible symptoms at a lower cost and in a simpler manner. [Solution] The information processing device (10) includes an acquisition unit (12) that acquires a target image of a target plant, and an estimation unit (13) that uses a learning model that takes the target image as input and outputs target classification information together with at least one of the following: (1) target probability information including a first probability that the target plant is classified into a first class and a second probability that it is classified into a second class, and (2) target accuracy information including a first correct answer rate when the target plant is classified into a first class and a second correct answer rate when it is classified into a second class, and estimates the progression of a predetermined plant disease in the target plant based on at least one of the target probability information and the target accuracy information.

Description

Technical Field

[0001] One aspect of the present invention relates to an information processing apparatus, an information processing method, and a control program.

Background Art

[0002] Techniques for estimating whether a plant included in an image is suffering from a predetermined plant disease and for estimating the progress of the plant disease in the plant are known. In Patent Document 1, a plant disease diagnosis system is disclosed that can easily diagnose plant diseases by saving the labor of manually extracting image feature data of plant diseases from images. In Patent Document 2, an optimal control recipe providing device is disclosed that analyzes crop video data and environmental information of a cultivation area to determine the type and progress of pests and diseases.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the conventional technologies as described above, in order to estimate the progress of plant diseases, hyperspectral images and environmental information measured using a plurality of devices are required, and there is a problem that expensive equipment and complicated operations are necessary for data acquisition. In addition, there is a problem that it is impossible to estimate the progress of plant diseases that appear in the appearance, such as the degree of wilting of plants.

[0005] One aspect of the present invention has been made in view of the above problems, and an object thereof is to enable the progress of a plant disease whose symptoms appear in the appearance to be estimated at a lower cost and more simply.

Means for Solving the Problems

[0006] To solve the above problems, an information processing device according to one aspect of the present invention includes: an acquisition unit that acquires a target image of a target plant; and an estimation unit that takes the target image as input and outputs target classification information indicating whether the target plant is classified into a first class indicating that it is suffering from a predetermined plant disease, or into a second class indicating that it is not suffering from a predetermined plant disease, together with at least one of the following: (1) target probability information including a first probability that the target plant is classified into the first class and a second probability that it is classified into the second class, and (2) target precision information including a first correct answer rate when the target plant is classified into the first class and a second correct answer rate when it is classified into the second class, wherein the estimation unit estimates the progression of the predetermined plant disease in the target plant based on at least one of the target probability information and the target precision information.

[0007] To solve the above problems, an information processing method according to one aspect of the present invention is an information processing method performed by a device, comprising: an acquisition step of acquiring a target image of a target plant; and an estimation step using a learning model that takes the target image as input and outputs target classification information indicating whether the target plant is classified into a first class indicating that it is suffering from a predetermined plant disease, or into a second class indicating that it is not suffering from the predetermined plant disease, together with at least one of: (1) target probability information including a first probability that the target plant is classified into the first class and a second probability that it is classified into the second class, and (2) target precision information including a first correct answer rate when the target plant is classified into the first class and a second correct answer rate when it is classified into the second class, wherein the estimation step estimates the progression of the predetermined plant disease in the target plant based on at least one of the target probability information and the target precision information.

[0008] To solve the above problems, an information processing device according to one aspect of the present invention includes: an acquisition unit that acquires an image of an object; and an estimation unit that takes the image of the object as input and outputs object classification information indicating whether the object is classified into a first class or a second class in which it is qualitatively classified, together with at least one of the following: (1) object probability information including at least one of a first probability that the object is classified into the first class and a second probability that it is classified into the second class, and (2) object precision information including at least one of a first correct answer rate when the object is classified into the first class and a second correct answer rate when it is classified into the second class, and estimates a predetermined quantitative value in the object based on at least one of the object probability information and the object precision information.

[0009] Each aspect of the present invention may be implemented by a computer, in which case a control program for the information processing device that enables the computer to implement the information processing device by operating the computer as each part (software element) of the information processing device, and a computer-readable recording medium on which the program is recorded, also fall within the scope of the present invention. [Effects of the Invention]

[0010] According to one aspect of the present invention, it is possible to estimate the progression of a plant disease that exhibits visible symptoms at a lower cost and in a simpler manner. [Brief explanation of the drawing]

[0011] [Figure 1] This is an example of a functional block diagram for an information processing system. [Figure 2] This is an example of a diagram showing a feature space where features extracted from images containing plants are mapped. [Figure 3] This is an example of a flowchart illustrating the flow of processing performed by an information processing system. [Figure 4] This figure shows an example of plant images of diseased individuals exhibiting mosaic symptoms at each stage. [Figure 5] This is an example of a table that organizes the values ​​of the first probability and the first correct answer rate according to the stages corresponding to the progression of mosaic disease symptoms in the target plant. [Figure 6] This figure shows examples of plant images of diseased individuals exhibiting wilting symptoms at each stage. [Figure 7] This is an example of a table that organizes the values ​​of the first probability and the first correct answer rate according to the progression of wilting symptoms in the target plant. [Modes for carrying out the invention]

[0012] One embodiment of the present invention will be described in detail below.

[0013] [1. Example of an information processing system configuration] The information processing system 1 according to this embodiment will now be described. Figure 1 is an example of a functional block diagram of the information processing system 1 according to this embodiment. The information processing system 1 is a system for estimating the presence and progression of plant diseases in plants included in an image, and comprises an information processing device 10, an imaging device 21, and a display device 22. Note that each component of the information processing system 1 is not limited to one, but may be multiple. Furthermore, the function of a single component of the information processing system 1 may be realized by multiple other components, and the functions of multiple components of the information processing system 1 may be realized by a single other component.

[0014] The information processing device 10 is a device that can be implemented as a personal computer, tablet, or smartphone, and comprises a control unit 11 and a storage unit 19.

[0015] The control unit 11 is a control device such as a CPU that oversees the entire information processing device 10, and also operates as an acquisition unit 12, an estimation unit 13, and a learning unit 14.

[0016] The acquisition unit 12 acquires a target image obtained by imaging a target plant from the imaging device 21 or the storage unit 19. Here, the target plant is a plant for which the presence and progress of a plant disease are to be estimated. In many cases, the target plant is an arbitrary plant having leaves, but this is not necessarily the case.

[0017] The estimation unit 13 estimates the progress of a predetermined plant disease in the target plant based on the target probability information and the target accuracy information output by a learning model that takes the target image as an input and outputs the target classification information together with at least one of the target probability information and the target accuracy information. The predetermined plant diseases include plant diseases derived from plant pathogens such as viruses, bacteria, fungi, viroids, mycoplasmas, or bacterium-like microorganisms, and plant diseases in which disease symptoms appear in the appearance as mosaic symptoms or wilting. Here, the disease symptoms mean the characteristics of a plant individual suffering from a plant disease.

[0018] In addition, the target classification information is information indicating whether the target plant is classified into a disease class (class classification) indicating that the target plant is suffering from a plant disease or a healthy class indicating that the target plant is not suffering from a plant disease. Here, the disease class is an example of the first class in the present disclosure, and the healthy class is an example of the second class in the present disclosure. In addition, a plant suffering from a plant disease means a plant in which the progress of the plant disease is equal to or greater than a predetermined threshold. Assuming that the progress is represented as a percentage, the threshold may define any value greater than at least 0%, and may be, for example, 50% or 100%.

[0019] In addition, the aforementioned target probability information is information including at least one of a first probability that the target plant is classified into the disease class and a second probability that the target plant is classified into the healthy class. The target accuracy information is information including at least one of a first correct answer rate when the target plant is classified into the disease class and a second correct answer rate when the target plant is classified into the healthy class. In the case of two-class classification, since one value of the first probability and the second probability determines the other value, the estimation based on the first probability and the estimation based on the second probability are substantially synonymous. The same explanation also applies to the first correct answer rate and the second correct answer rate.

[0020] FIG. 2 is a diagram for supplementing with an example regarding the classification process performed by the estimation unit 13, and is an example of a diagram showing a feature amount space obtained by mapping the feature amounts extracted from the target image.

[0021] In the example of FIG. 2, the symptom of the plant disease is that the surface of the plant exhibits yellow, and the surface of a healthy individual plant that is not suffering from the plant disease is green. Further, the feature point 31a is a feature point extracted from an image of a diseased individual plant suffering from the plant disease, and the feature point 31b is a feature point extracted from an image of a plant individual not suffering from the plant disease. Further, the boundary curve 32 is a boundary curve used for two-class classification that classifies the plant individual corresponding to each feature point into either a diseased class or a healthy class. Specifically, the upper left side of the boundary curve 32 corresponds to the diseased class, and the lower right side corresponds to the healthy class. When the feature point is located on the boundary curve 32, the plant individual corresponding to the feature point may be classified into any of the pre-specified classes. Also, in the example of FIG. 2, the feature point located on the boundary curve 32 means that the progress degree of the plant disease in the plant individual corresponding to the feature point is 50%.

[0022] Also, the farther the feature point is mapped from the boundary curve, the more the symptom or the features of the healthy individual appear in the appearance of the plant individual corresponding to the feature point, and the reliability in the classification process is high. On the contrary, the closer the feature point is mapped to the boundary curve, the more the symptoms and the features of the healthy individual are mixed in the appearance of the plant individual corresponding to the feature point, and the reliability in the classification process is low.

[0023] Furthermore, the more a feature point is mapped to the upper left position in Figure 2, the higher the first probability and first correct answer rate for the plant individual corresponding to that feature point to be classified into the disease class, and the lower the second probability and second correct answer rate for being classified into the healthy class. The first probability can also be rephrased as the probability that the feature point is mapped to the upper left of the boundary curve 32. Conversely, the more a feature point is mapped to the lower right position, the lower the first probability and first correct answer rate for the plant individual corresponding to that feature point to be classified into the disease class, and the higher the second probability and second correct answer rate for being classified into the healthy class. The second probability can also be rephrased as the probability that the feature point is mapped to the lower right of the boundary curve 32.

[0024] Furthermore, the accuracy information of the plant individuals corresponding to the feature points may be calculated by the estimation unit 13 based on the accuracy rate when classifying plant individuals corresponding to feature points that were previously mapped to the vicinity of the feature point in the feature space.

[0025] The learning unit 14 trains the aforementioned learning model using a pair of specimen images and classification information indicating whether or not the specimen plants contained in the images are suffering from a predetermined plant disease as training data. For example, the training data includes multiple specimen images as explanatory variables and classification information corresponding to each of the multiple specimen images as the target variable. In other words, the specimen images are images used for training or validating the learning model. Hereafter, when there is no particular distinction between target images and specimen images, images containing plants may simply be referred to as plant images.

[0026] The specimen image is an image of the specimen plant taken using the same method as the target image taken of the target plant. For example, if the target image was taken with an optical camera, the specimen image may also be an image taken with an optical camera. Alternatively, if the target image was taken with an infrared camera, the specimen image may also be an image taken with an infrared camera. Furthermore, the specimen plant may, but is not limited to, the same species as the target plant. For example, the specimen plant may be a plant whose changes before and after contracting a predetermined plant disease are similar to the changes in the target plant before and after contracting the same predetermined plant disease.

[0027] The memory unit 19 is a storage device such as a memory that temporarily stores various types of information, for example, a parameter set that defines one or more learning models, and images captured by the imaging device 21.

[0028] The imaging device 21 is a device that captures images of plants, etc., and is implemented as a digital camera or the like. However, if, for example, the information processing device 10 is implemented as a smartphone, then the imaging device 21 is usually the camera built into the smartphone. Similarly, the display device 22 may also be integrated with the information processing device 10. The plant images captured by the imaging device 21 are supplied to the information processing device 10 via a wired or wireless connection.

[0029] When the imaging device 21 images plants, it is desirable to perform imaging in a way that provides sufficient spatial resolution to identify disease symptoms, and imaging may be performed using a sunshade to suppress overexposure due to shadows or reflection of direct light. In addition, when imaging indoors, diffused light may be used by attaching tracing paper or the like to the light source.

[0030] The display device 22 is a display that shows images or text, etc. For example, the display device 22 displays a target image supplied from the information processing device 10, and information indicating the progression of a plant disease in the plant.

[0031] The above describes an example of the configuration of Information Processing System 1. In addition, each part included in Information Processing System 1 has the function of executing the processes described below.

[0032] [2. Examples of information processing systems] Next, we will explain the processing flow performed by the information processing system 1. Figure 3 is an example of a flowchart showing the processing flow.

[0033] In step S1, the imaging device 21 captures a target image. The acquisition unit 12 acquires the image captured by the imaging device 21.

[0034] In S2, the estimation unit 13 inputs the target image acquired by the acquisition unit 12 into the learning model and estimates the value of the first probability output by the learning model as the progression of the plant disease in the target plant.

[0035] The estimation unit 13 may also estimate the progression of the plant disease based on at least one of the values ​​of the target probability information and the target classification information output by the learning model. For example, the estimation unit 13 may estimate the progression of the plant disease as the value of the first correct answer rate, or the average or weighted sum of the first probability and the first correct answer rate, or it may perform a process such as adding a predetermined shift value to either value.

[0036] In step S3, the estimation unit 13 estimates whether the target plant is infected with a plant disease based on a comparison between the progression of the plant disease in the target plant and a threshold value. Typically, the estimation unit 13 estimates that the target plant is infected with a plant disease if the progression is equal to or greater than the threshold value.

[0037] In S4, the control unit 11 displays the target image, the estimated result of whether or not the plant is suffering from a plant disease, and the value of the first probability on the display device 22.

[0038] As described above, the information processing method performed by the information processing device 10 includes an acquisition step of acquiring a target image, and an estimation step of estimating the progression of a predetermined plant disease in the target plant based on at least one of the target probability information and the target precision information output by the learning model. With the configuration of this example, it is possible to estimate the progression of a plant disease that shows symptoms on its appearance at a lower cost and in a simpler manner.

[0039] [3. Additional Notes] The machine learning methods executed by the estimation unit 13 and the learning unit 14 are not limited to any particular method, and may include, for example, CNNs (Convolutional Neural Networks) or RNNs (Recurrent Neural Networks). Furthermore, either classification or regression models may be used. When using neural networks such as CNNs or RNNs, the input data may be pre-processed for use as input to the neural network. Such processing may include two-dimensional or multi-dimensional data arrangement, as well as various data augmentation techniques, adjustments to brightness, color tone, image quality, plant angles, object detection for extracting plant regions, and segmentation.

[0040] Furthermore, when using a CNN, one or more layers in the neural network may be convolutional layers that perform convolutional operations, and a configuration may be used in which filtering operations (sum-accumulate operations) are performed on the input data input to that layer. When performing filtering operations, processing such as padding may be used in combination, or an appropriately set stride width may be adopted.

[0041] Alternatively, the system may employ one or a combination of the following machine learning techniques: Support Vector Machine (SVM) Clustering • Inductive Logic Programming (ILP) • Genetic Programming (GP) • Bayesian Network (BN) • Autoencoder Furthermore, the information processing system 1 may be configured to estimate and display the progression of plant disease for all plant individuals included in one or more target images. In the above configuration, the acquisition unit 12 acquires one or more target images containing multiple target plants. Subsequently, the estimation unit 13 estimates the progression of plant disease for all target plants included in one or more target images by dividing the sum of the first probability or first correct answer rate estimated for each of the target plants by the number of plant individuals included in one or more target images.

[0042] Furthermore, the information processing system 1 may be configured to estimate and display the proportion of target plants that are infected with a plant disease among multiple target plants included in one or more images. In the above configuration, the acquisition unit 12 acquires one or more target images that include multiple target plants. Subsequently, the estimation unit 13 estimates whether or not each of the target plants is infected with a plant disease according to the progression of the plant disease in each target plant, and estimates the proportion of target plants infected with a plant disease by dividing the number of target plants infected with the plant disease by the number of target plants included in one or more target images.

[0043] Furthermore, the images used by the learning unit 14 as training data may include a first specimen image of a specimen plant whose disease progression is greater than or equal to a first standard value, and a second specimen image of a specimen plant whose disease progression is less than or equal to a second standard value. The above configuration also includes cases where "greater than or equal to the first standard value" is read as "greater than the first standard value," and "less than or equal to the second standard value" is read as "less than the second standard value." In the above configuration as well, the estimation unit 13 can estimate the numerical value of the disease progression as a value greater than the second standard value and less than the first standard value. This means that the specimen images used by the learning unit 14 as training data may include specimen images that include specimen plants whose disease progression is very high and are clearly diseased, and specimen images that include specimen plants whose disease progression is very low and are clearly healthy. Even in this case, the estimation unit 13 relating to this disclosure can also perform estimations targeting target plants with an intermediate degree of plant disease progression, depending on at least one of the target probability information and the target accuracy information. For example, the images used by the learning unit 14 as training data may be a first specimen image including a specimen plant with a plant disease progression of 90% or more, and a second specimen image including a specimen plant with a plant disease progression of 10% or less. Even if specimen images of specimen plants with a plant disease progression of more than 10% and less than 90% are not used as training data, the estimation unit 13 can estimate that the plant disease progression in the target plant is either more than 10% and less than 90%. This makes it possible to use specimen images including specimen plants that can be clearly estimated as diseased or healthy individuals by human visual inspection as training data, and contributes to improving the accuracy of plant disease progression estimation.

[0044] Furthermore, the configuration of the information processing system 1 disclosed herein is also applicable when the target object is something other than a plant. When the above configuration is generalized, the acquisition unit 12 acquires a target image of the target object. The estimation unit 13 takes the target image as input and estimates a predetermined quantitative value in the target object based on at least one of the target probability information and target precision information output by a learning model that outputs target classification information indicating whether the target object is classified into a first class or a second class, along with at least one of the target probability information and target precision information. The progression of the plant disease mentioned above is an example of the predetermined quantitative value. The target probability information includes at least one of the first probability that the target object is classified into the first class and the second probability that it is classified into the second class. The target precision information includes at least one of the first correct answer rate when the target object is classified into the first class and the second correct answer rate when it is classified into the second class. The aforementioned disease class and healthy class are examples of the first and second classes, respectively, which are classified qualitatively. The learning unit 14 trains its learning model using a pair of sample images and classification information indicating whether the object (sample) contained in the sample image belongs to the first or second class as training data. The sample images used by the learning unit 14 as training data may also include sample images containing objects whose quantitative value is greater than or equal to the first reference value, and sample images containing objects whose quantitative value is less than or equal to the second reference value. In this case as well, the estimation unit 13 can estimate the quantitative value of the object as a value greater than the second reference value and less than the first reference value. Furthermore, the object is not limited to plants, but may be any organic or inorganic matter, and the quantitative value may be, for example, the degree to which the plant is withered, the degree of disease resistance, the degree of pesticide control effectiveness, the degree of insect damage, the degree of aging, or the degree of contamination, in addition to the progression of disease, etc.

[0045] Furthermore, the configuration of this disclosure, which estimates the progress based on a first probability or a first correct answer rate, is also applicable to object detection and segmentation processes that may include classification as part of the process.

[0046] Furthermore, the information processing system 1 may simultaneously estimate and display the progression of multiple types of plant diseases, or it may perform classification into three or more classes. For example, the estimation unit 13 may estimate the progression of mosaic symptoms and wilting on the leaves of the same plant, according to at least one of the target probability information and the target classification information. In this case, the estimation unit 13 may classify plants into three classes: a healthy class, a mosaic symptom class, and a wilting class, or it may classify plants into four classes: a healthy class, a mosaic symptom class, a wilting class, and a class with both mosaic symptoms and wilting. In the former three-class classification, plants exhibiting both mosaic symptoms and wilting are classified into the class corresponding to the more severe symptom.

[0047] [4. Examples of implementation using software] The functions of the information processing device 10 (hereinafter referred to as "the device") are programs that cause the device to function as a computer, and these programs can be realized by programs that cause each control block of the device (particularly each part included in the control unit 11) to function as a computer.

[0048] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.

[0049] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0050] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0051] Furthermore, each process described in the above embodiments may be performed by AI (Artificial Intelligence). In this case, the AI ​​may operate on the control device described above, or it may operate on other devices (for example, an edge computer or a cloud server).

[0052] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. [Examples]

[0053] A first embodiment of the present invention will be described below. In this example, a process will be described in which an implementer uses a system equivalent to the information processing system 1 to estimate the progression of mosaic disease symptoms originating from a certain plant disease.

[0054] First, the implementer defined five stages of progression for the mosaic disease symptoms in the target plants, as follows. The 80% below is an example of the first standard value mentioned above, and the 20% below is an example of the second standard value. • Mosaic disease symptoms (severe): A condition where the progression is less than 100% but more than 80%. • Mosaic disease symptoms (moderate): A condition where the progression is 80% or less but greater than 60%. • Mosaic symptoms (mild): A condition where the progression is 60% or less but greater than 40%. • Healthy (Weak): A state where the progression is 40% or less but more than 20%. • Healthy (Strong): A state where the progression is 20% or less.

[0055] Furthermore, the implementer stipulated that plant individuals exhibiting mosaic symptoms of (severe), (moderate), or (severe) should be classified as diseased individuals, and that plant individuals exhibiting healthy (slight) and (severe) conditions should be classified as healthy individuals. Here, plant individuals exhibiting mosaic symptoms of (severe) were clearly diseased even by visual inspection, and plant individuals exhibiting healthy (severe) conditions were clearly healthy even by visual inspection. In contrast, it was not clear by visual inspection whether plant individuals in other conditions, particularly those exhibiting mosaic symptoms of (slight), were diseased or healthy.

[0056] Figure 4 shows examples of plant images of diseased individuals exhibiting mosaic symptoms at each stage. In Figure 4, the numbers below the plant images indicate the percentage progression of mosaic symptoms. Specifically, the four plant images corresponding to the numbers 100 and 90 represent plants with severe mosaic symptoms, the four plant images corresponding to the numbers 80 and 70 represent plants with moderate mosaic symptoms, and the four plant images corresponding to the numbers 60 and 50 represent plants with mild mosaic symptoms.

[0057] The researchers then trained a learning model using multiple randomly selected plant images that could be visually distinguished, specifically images of plants with severe mosaic disease symptoms and images of healthy plants with severe symptoms, as specimen images, i.e., training data and validation data. Images of plants with moderate and mild mosaic disease symptoms, as well as healthy plants with mild symptoms, were not used as training or validation data.

[0058] Next, the implementer used the system, which had a sufficiently trained learning model, to perform a two-class classification of target plants included in randomly selected target images as test data. The classification was divided into two classes: a disease class exhibiting mosaic symptoms and a healthy class. The test data used were plant images of the five stages of disease as described above, specifically images of plants with (severe), (moderate), and (slight) mosaic symptoms, as well as healthy plants with (slight) and (severe) symptoms.

[0059] Figure 5 shows an example of a table that organizes the values ​​of the first correct answer rate and the first probability according to the progression of mosaic disease symptoms in the target plant, based on the two-class classification described above. The average value of the first probability is also listed. For example, the average first probability of a target plant with severe mosaic disease symptoms being classified into the disease class was 91.7%. The system estimates that the target plant is diseased if the first correct answer rate or first probability value is above 40%, and healthy if it is 40 or less.

[0060] As shown in Figure 5, regardless of whether the first probability or the first correct answer rate is used to estimate the progression of mosaic disease symptoms, it was demonstrated that it is possible to quantify and estimate the progression corresponding to plant images with moderate and mild mosaic disease symptoms, as well as healthy plants with mild symptoms, which were not used as training or validation data. [Examples]

[0061] A second embodiment of the present invention will be described below. In this example, a process will be described in which an implementer uses a system equivalent to Information Processing System 1 to estimate the progression of wilting symptoms caused by a certain plant disease.

[0062] First, the implementer defined five stages of progression for the wilting symptoms in the target plants, as follows. The 80% below is an example of the first standard value mentioned above, and the 20% below is an example of the second standard value. • Wilting symptoms (severe): A condition where the progression is 100% or less but more than 80%. • Wilting symptoms (moderate): A condition where the disease has progressed by 80% or less but more than 60%. • Wilting symptoms (mild): A condition where the progression is 60% or less but more than 40%. • Healthy (Weak): A state where the progression is 40% or less but more than 20%. • Healthy (Strong): A state where the progression is 20% or less.

[0063] Furthermore, the implementer stipulated that plant individuals exhibiting (severe), (moderate), or (weak) wilting symptoms should be classified as diseased individuals, and plant individuals in a healthy (weak) or (severe) state should be classified as healthy individuals. Here, plant individuals exhibiting (severe) wilting symptoms were clearly diseased even by visual inspection, and plant individuals in a healthy (strong) state were clearly healthy even by visual inspection. In contrast, it was not clear by visual inspection whether plant individuals in other states, particularly those exhibiting (weak) wilting symptoms, were diseased or healthy.

[0064] Figure 6 shows examples of plant images of diseased individuals exhibiting wilting symptoms at each stage. In Figure 6, the numbers below the plant images indicate the percentage progression of wilting symptoms. Specifically, the four plant images corresponding to the numbers 100 and 90 represent plants with severe wilting symptoms, the four plant images corresponding to the numbers 80 and 70 represent plants with moderate wilting symptoms, and the four plant images corresponding to the numbers 60 and 50 represent plants with mild wilting symptoms.

[0065] The researchers then trained a learning model using multiple randomly selected plant images that could be visually distinguished, specifically images of plants with severe wilting symptoms and images of healthy plants with severe symptoms, as specimen images, i.e., training and validation data. Images of plants with moderate and mild wilting symptoms, as well as healthy plants with mild symptoms, were not used as training or validation data.

[0066] Next, the implementer used the system, which had a sufficiently trained learning model, to classify target plants included in randomly selected target images as test data into two classes: a disease class exhibiting wilting symptoms and a healthy class. The test data used were plant images of the five stages of disease as described above, specifically plant images with wilting symptoms (severe), (moderate), and (slight), and healthy plants (slight) and (severe).

[0067] Figure 7 shows an example of a table that organizes the values ​​of the first correct answer rate and the first probability according to the progression of wilting symptoms in the target plant, based on the two-class classification described above. The average value of the first probability is also listed. For example, the average first probability of a target plant with severe wilting symptoms being classified into the disease class was 93.4%. The system estimates that a target plant is diseased if the first correct answer rate or first probability value is above 40%, and healthy if it is 40 or less.

[0068] As shown in Figure 7, regardless of whether the first probability or the first correct answer rate is used to estimate the progression of wilting symptoms, it was demonstrated that it is possible to quantify and estimate the progression corresponding to plant images with moderate and mild wilting symptoms, as well as healthy plants with mild wilting symptoms, which were not used as training or validation data. [Explanation of Symbols]

[0069] 1. Information Processing System 10 Information Processing Devices 11 Control Unit 12 Acquisition Department 13 Estimation part 14. Learning Department 19 Memory section 21 Imaging device 22 Display device

Claims

1. An acquisition unit that acquires a target image of the target plant, An estimation unit using a learning model that takes the aforementioned target image as input and outputs target classification information indicating whether the target plant is classified into a first class indicating that it is suffering from a predetermined plant disease, or into a second class indicating that it is not suffering from the predetermined plant disease, together with at least one of the following: (1) target probability information including at least one of a first probability that the target plant is classified into the first class and a second probability that it is classified into the second class, and (2) target precision information including at least one of a first correct answer rate when the target plant is classified into the first class and a second correct answer rate when it is classified into the second class, wherein the estimation unit estimates the progression of the predetermined plant disease in the target plant based on at least one of the target probability information and the target precision information, An information processing device equipped with the following features.

2. The aforementioned learning model, Training data comprising a first specimen image of a specimen plant whose progression of a predetermined plant disease is equal to or greater than a first reference value, and a second specimen image of the specimen plant whose progression of the predetermined plant disease is less than or equal to a second reference value which is less than the first reference value, wherein the first specimen image is associated with classification information indicating the first class, and the second specimen image is associated with classification information indicating the second class, and is generated by machine learning using this training data. The information processing apparatus according to claim 1.

3. The acquisition unit is, Obtain one or more target images that include multiple individual target plants. The estimation unit, The information processing apparatus according to claim 1 or 2, wherein the sum of the first probability or first correct answer rate estimated for each of the target plants is divided by the number of target plants included in the one or more target images, and the resulting value is estimated as the progression of the plant disease for all the target plants included in the one or more target images.

4. The estimation unit, An information processing device according to claim 1 or 2, which estimates whether or not the target plant is suffering from the plant disease based on the result of comparing the progression of the disease with a threshold value.

5. The acquisition unit is, Obtain one or more target images that include multiple individual target plants. The estimation unit, The information processing device according to claim 4, which estimates whether each of the target plants is infected with the plant disease, and estimates the proportion of target plants infected with the plant disease by dividing the number of target plants infected with the plant disease by the number of target plants included in the one or more target images.

6. An information processing method performed by a device, The acquisition step involves obtaining a target image of the target plant, An estimation step using a learning model that takes the aforementioned target image as input and outputs target classification information indicating whether the target plant is classified into a first class indicating that it is suffering from a predetermined plant disease, or into a second class indicating that it is not suffering from the predetermined plant disease, together with at least one of the following: (1) target probability information including at least one of a first probability that the target plant is classified into the first class and a second probability that it is classified into the second class, and (2) target precision information including at least one of a first correct answer rate when the target plant is classified into the first class and a second correct answer rate when it is classified into the second class, wherein the estimation step estimates the progression of the predetermined plant disease in the target plant based on at least one of the target probability information and the target precision information. Information processing methods including

7. A control program for causing a computer to function as an information processing device according to claim 1, wherein the computer functions as the acquisition unit and the estimation unit.

8. An acquisition unit that acquires an image of the target object, An estimation unit using a learning model that takes the aforementioned target image as input and outputs target classification information indicating whether the object is classified qualitatively into a first class or a second class, along with at least one of the following: (1) target probability information including at least one of a first probability that the object is classified into the first class and a second probability that it is classified into the second class, and (2) target precision information including at least one of a first correct answer rate when the object is classified into the first class and a second correct answer rate when it is classified into the second class, wherein the estimation unit estimates a predetermined quantitative value in the object based on at least one of the target probability information and the target precision information. An information processing device equipped with the following features.

Images