JP6307680B1 - Plant health diagnosis system - Google Patents

Abstract

The present invention provides a plant health diagnosis system capable of easily and appropriately grasping various health states of plants. A plant health diagnosis system for diagnosing a health condition from an image of a plant, the learning target image being a plant image to be learned, and the learning target at the time of shooting the learning target image A learning means 105 for learning a method for diagnosing the health state of the plant from an image of the plant to be diagnosed, and receiving an input of a diagnostic object image obtained by photographing the plant to be diagnosed. And a diagnosis unit 107 that outputs a diagnosis result of a health condition of a diagnosis target based on the method. [Selection] Figure 2

Description

  The present invention relates to a system for diagnosing a health condition of a plant, and more particularly to a system for diagnosing a health condition of a plant from an image obtained by photographing a plant to be diagnosed.

  When cultivating plants such as crops, it is necessary to judge the health of the plant and take appropriate measures based on it. For example, if there is an excess or deficiency in the amount of water in the plant body, the countermeasure of increasing or decreasing the amount of water to be given is taken. It is necessary to determine and deal with the situation such as providing fertilizer to compensate for the situation, such as removing the part if the plant is in a diseased state. However, when judging the state of a plant from its appearance or the like, it is difficult to make an appropriate judgment unless the plant is cultivated for many years and is skilled.

  On the other hand, a system that assists the judgment of the state of the plant without being a skilled person has been proposed. For example, in Patent Document 1, a user who cultivates a plant transmits an image of a plant to the plant image analysis device at the center by the image reading device, and the plant image analysis device uses the color of the leaf surface or the like. An automatic plant diagnosis system is described that performs diagnosis and transmits the results and countermeasures to a user's image reader.

  Patent Document 2 discloses real-time plant nutrition, in which a sample is taken from the leaf stem and soil of a plant, and components contained in the sap and soil of the plant are analyzed using a test paper and an LED reflected light clock. Diagnostic methods are described.

Japanese Patent Laid-Open No. 9-262027 JP-A-11-289870

  According to the system described in Patent Document 1, the user can know the state of the plant and how to deal with it by transmitting an image of the plant. However, in order to properly determine the state of the plant from the image of the plant, information such as what features appear on the appearance and how it changes over time It is necessary to register in advance information such as such characteristics and changes indicating what state the plant is in. Assuming various states of plants, it is assumed that a great deal of labor is required to comprehensively input the appearance features that appear. Also, assuming that it is used for various types of plants, it is not realistic to register all information in advance such as appearance characteristics that differ from plant to plant and diseases specific to each plant.

  If the direct analysis of the components contained in the sap or soil of the plant as described in Patent Document 2 is performed, the state of the plant can be determined without performing a large amount of information accumulation in advance as described above. It can be carried out. However, for that purpose, an instrument for analysis and complicated procedures are required, and it is difficult to carry out easily. In addition, the analysis as shown in Patent Document 2 is simple, and in order to obtain more detailed and reliable analysis results, samples are taken from plants and soil and sent to specialized institutions. Then, the procedure of requesting the analysis and waiting for the result is necessary. Such detailed analysis at specialized institutions is recommended to determine the state of the plant as needed from the viewpoints of cost, labor, and the period until the analysis results are obtained, and take appropriate measures against it. Use in is difficult.

  Therefore, an object of the present invention is to provide a plant health diagnosis system that can easily and appropriately grasp various health states of plants.

In order to solve the above problems, a plant health diagnosis system according to the present invention comprises:
A plant health diagnosis system for diagnosing a health condition from an image of a plant,
With reference to the learning target image that is an image of the plant to be learned and the health state of the learning target at the time of shooting the learning target image, the health state of the plant from the image of the plant to be diagnosed Learning means for learning diagnostic methods;
Diagnostic means for receiving an input of a diagnostic target image obtained by photographing a plant that is a diagnostic target, and outputting a diagnosis result of a health condition of the diagnostic target based on the diagnostic method.

  In this way, by learning the diagnosis method of the health condition from the photographed plant image and the health condition at that time, and performing a health check based on it, it is possible to perform a large amount of input work of the reference for diagnosis in advance Even if not performed, the health condition can be diagnosed from the image of the plant to be diagnosed.

In a preferred embodiment of the present invention, the learning target image includes at least an image obtained by photographing the learning target veins,
The learning means learns the diagnostic method based on the form of the veins to be learned,
The diagnostic target image includes at least an image of the diagnostic target vein image;
The diagnosis means outputs the diagnosis result based on the form of the vein of the diagnosis object ;
The form of the vein of the learning object and the form of the vein of the diagnosis object include at least one of the thickness of the vein and the continuity of the vein .
Thus, in learning of the diagnostic method and in the health examination based on the learning result, the health condition can be diagnosed with high accuracy by paying attention to the leaf vein form of the plant.

In a preferred aspect of the present invention, the health condition of the learning target includes at least an analysis result of the sap of the learning target.
Thus, by including the analysis result of sap as a learning target, a more accurate health state of the learning target can be grasped, and a diagnostic method using the health condition can be learned.

In a preferred embodiment of the present invention, the learning means further refers to information related to the growth environment of the learning target at the time of shooting the learning target image.
As described above, by using the information about the growth environment to be learned for the learning of the diagnosis method, it is possible to learn the diagnosis method with higher accuracy.

In a preferred embodiment of the present invention, the learning means further learns a method of determining a countermeasure according to the health condition of the learning target,
The diagnosis result includes the countermeasure based on a health condition of the diagnosis target.
In this way, by adopting a configuration that outputs a coping method based on the diagnosis result, it is possible to provide information for performing an appropriate coping when the health condition of the plant is poor or when the plant is sick.

In a preferred aspect of the present invention, the learning means performs machine learning.
Thus, by using machine learning for learning of the diagnostic method, learning of the diagnostic method can be performed effectively.

The plant health check program according to the present invention includes:
A plant health diagnosis program for diagnosing a health condition from an image of a plant, comprising a computer device,
With reference to the learning target image that is an image of the plant to be learned and the health state of the learning target at the time of shooting the learning target image, the health state of the plant from the image of the plant to be diagnosed Learning means for learning diagnostic methods;
Receiving an input of a diagnostic object image obtained by photographing a plant that is a diagnostic object, and operating as a diagnostic means for outputting a diagnosis result of a health condition of the diagnostic object based on the diagnostic method ;
The learning target image includes at least an image of the learning target vein image;
The learning means learns the diagnostic method based on the form of the veins to be learned,
The diagnostic target image includes at least an image of the diagnostic target vein image;
The diagnosis means outputs the diagnosis result based on the form of the vein of the diagnosis object;
The form of the vein of the learning object and the form of the vein of the diagnosis object include at least one of the thickness of the vein and the continuity of the vein .

The plant health diagnosis method according to the present invention comprises:
A plant health diagnosis method using a computer device that diagnoses a health condition from an image of a plant,
The computer device refers to a learning target image that is an image of a plant that is a learning target and a health state of the learning target at the time of shooting the learning target image, and the image from the plant image that is a diagnosis target Learning how to diagnose plant health,
The computer device receives an input of a diagnostic target image obtained by photographing a plant that is a diagnostic target, and outputs a diagnosis result of a health condition of the diagnostic target based on the diagnostic method ,
The learning target image includes at least an image of the learning target vein image;
In the step of learning the diagnostic method, the diagnostic method is learned based on the morphology of the veins to be learned,
The diagnostic target image includes at least an image of the diagnostic target vein image;
In the step of outputting the diagnosis result, the diagnosis result based on a form of a vein of the diagnosis target is output,
The form of the vein of the learning object and the form of the vein of the diagnosis object include at least one of the thickness of the vein and the continuity of the vein .

  It is possible to provide a plant health diagnosis system capable of diagnosing and outputting various health states of a plant without carrying out exhaustive input such as criteria for judging the health state in advance.

1 is a configuration diagram of a plant health diagnosis system according to an embodiment of the present invention. It is a functional block diagram of the medical examination server device concerning one embodiment of the present invention. It is a flowchart which shows the flow of the recording process of the data made into learning object in one Embodiment of this invention. It is a flowchart which shows the flow of the diagnostic process in one Embodiment of this invention.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a configuration diagram of a plant health diagnosis system according to the present embodiment. As shown here, the plant health diagnosis system according to the present embodiment stores data to be learned by a diagnostic method, learns a diagnostic method using the data, accepts input of information to be diagnosed, and responds thereto It is a diagnosis target for the health diagnosis server device 1 that performs output of diagnosis results, the expert terminal device 2 that is used by experts who have a wealth of knowledge about the growth of plants to be diagnosed, and the health diagnosis server device 1. User terminal device 3a to 3d that transmits an image of a plant, receives a request for diagnosis of its health condition, and receives and displays the result (hereinafter simply referred to as user terminal device 3 unless there is a need for distinction). Are configured to be communicable via a network NW such as the Internet.

  FIG. 2 is a functional block diagram of the medical examination server device 1 according to the present embodiment. As shown here, the health diagnosis server device 1 according to the present embodiment includes a learning target receiving unit 101 that receives an image of a plant that is a learning target, its health state, and growing environment information from the expert terminal device 2. An image storage unit 102 that stores the learning target image, a health state storage unit 103 that stores the health state of the learning target, an environment information storage unit 104 that stores the environment information on which the learning target is growing, and an image storage Unit 102, health state storage unit 103, and environment information storage unit 104, based on the image of the plant, the health state, and the environment information, the determination of the health state of the plant from the image and appropriate measures based on it A learning means 105 for learning a diagnosis method for determination, a diagnosis method storage unit 106 for storing the diagnosis method learned by the learning means 105, and a plant to be diagnosed by the user terminal device 3 Is received, a diagnosis based on the diagnosis method recorded in the diagnosis method storage unit 106 is performed, and the health condition of the diagnosis target and appropriate measures for the diagnosis target are transmitted to the user terminal device 3 as a diagnosis result. Diagnostic means 107.

  The image storage unit 102 stores a leaf surface / leaf vein image storage unit 1021 that stores an image of a leaf surface or vein of a plant to be learned, and an image of a whole image that includes parts other than leaves. And an entire image storage unit 1022. In addition, the health state storage unit 103 is input by an expert using the expert terminal device 2 and stores a tree force information storage unit 1031 that stores information on a tree to be learned at the time of shooting, and an analysis result of the sap collected at the time of shooting. And a sap analysis result storage unit 1032. The environment information storage unit 104 includes information on the growth environment at the time of taking an image of a plant image to be learned, a soil analysis result storage unit 1041 that stores an analysis result of the soil on which the plant grows, Atmosphere / weather information storage unit 1042 that stores information on the atmosphere and weather, and a treatment information storage unit 1043 that stores information on treatments performed on plants to be learned, such as the amount of moisture and the type and amount of fertilizer applied. And having.

  The medical examination server device 1 includes an arithmetic device such as a CPU, a main storage device such as a RAM (Random Access Memory), an HDD (Hard Disk Drive), an SSD (Solid State Drive), an auxiliary storage device such as a flash memory, and a network NW. A general computer device such as a server device provided with various input / output devices including a connection means to the server can be used. More specifically, programs for operating the health diagnosis server device 1 as each means as described above are stored in the auxiliary storage device in advance or by the operation of the administrator of the health diagnosis server device 1. The computer program is used as the health diagnosis server device 1 in the electronic certification system according to the present embodiment by performing the above-described program on the main storage device, performing calculations by the calculation device, and controlling the input / output means. Can do. In addition, in this embodiment, although the structure which implement | achieves the medical examination server apparatus 1 by a single computer apparatus was shown, this invention is not limited to this. For example, a storage unit such as the image storage unit 102, the health state storage unit 103, and the environment information storage unit 104 is provided in another computer device, and a configuration for referring to it, a configuration in which the learning unit 105 is provided in another computer device, etc. The health diagnosis server device 1 may be realized by a plurality of computer devices. Further, as will be described later, when the learning means 105 is configured to perform machine learning, a configuration capable of implementing General-Purpose computing on Graphics Processing Units (GPGPU) using GPU (Graphics Processing Unit), etc. It is preferable that

  As the expert terminal device 2, a general computer device including various arithmetic devices, a main storage device, an auxiliary storage device, various input / output devices including means for connecting to the network NW, and the like can be used. Here, as will be described later, in order to capture an image of a plant to be learned, input information about the tree vigor, etc., it is equipped with a camera function and can be easily brought into a plant growing place. And a tablet-type terminal can be suitably used. In FIG. 1, only one expert terminal device 2 is shown, but the present invention is not limited to this, and there are a plurality of knowledgeable plants having the same knowledge or different kinds of plants. It is good also as a structure provided with the some expert terminal device 2 which each expert knows. Further, in addition to the expert terminal device 2 or instead of the expert terminal device 2, the health diagnosis server device 1 may directly receive an input of an image, a health condition, and environmental information about a learning target.

  As the user terminal device 3, a general computer device including various arithmetic devices, a main storage device, an auxiliary storage device, various input / output devices including means for connecting to the network NW, and the like can be used. In FIG. 1, as an example, a smartphone 3a, a tablet-type terminal 3b, a notebook computer 3c, and an agricultural robot 3d are shown. However, the present invention is not limited to this, and other various computer devices can also be used as user terminals. It can be used as the device 3. Here, the smartphone 3a, the tablet terminal 3b, and the notebook computer 3c are such that a user who cultivates a plant captures an image of a plant cultivated by himself or captures it from a digital camera, etc. 1, by receiving and displaying diagnosis results such as health status and countermeasures based on it, it is possible to perform appropriate treatments on plants cultivated by the user. Further, the agricultural robot 3d is photographed by taking an image of a plant or taking it from another photographing device by an operation by the administrator or a periodic automatic process, and transmitting it to the medical examination server device 1 The diagnosis result is received, and the countermeasures based on the health condition of the plant can be automatically performed.

<Accumulation of learning target data>
Next, a flow of processing for recording and storing data to be learned in the plant health diagnosis system according to the present embodiment will be described with reference to FIG. This process is performed by communication using HTTP (Hyper Text Transfer Protocol) between the two programs, the health check server device 1 having a Web server program and the expert terminal device 2 having a Web browser program. Alternatively, the expert terminal device 2 may be provided with a dedicated program for inputting data to be learned, and by using the program, the information of the information using an arbitrary communication protocol with the health diagnosis server device 1 may be used. It is good also as a structure which transmits / receives.

  In this process, first, in step S101, an input of an image obtained by photographing the veins and leaf surfaces of a plant to be learned is received from the expert terminal device 2. In the case where the expert terminal device 2 includes a camera, the image here may be configured to prompt the user to capture and transmit an image, or may be captured in advance and the auxiliary storage device included in the expert terminal device 2. It may be configured to prompt selection and transmission of images stored in the

  Next, in step S102, an image of the whole plant image to which the image of leaf veins and leaf surfaces was transmitted in step S101 is received. Here, similarly to step S101, the configuration may be such that transmission of an image taken by the expert terminal device 2 or an image stored in the auxiliary storage device of the expert terminal device 2 is urged. Here, for the sake of simplicity, an image of the whole image is used for convenience. However, the image is not necessarily an image including the whole plant, that is, received in step S101 including a plurality of leaves and stems. Any image that includes a wider area than an image that includes a leaf vein or leaf surface may be used. Moreover, it is good also as a structure which receives several images, such as an image including the upper part of a plant, and a root image.

  In step S103, an input of tree vigor information at the time of photographing the image received in steps S101 and S102 of the plant to be learned is accepted. Examples of the tree vigor information include information on the appearance of the plant such that the color of the leaf surface looks bad, and tactile information such as the degree of leaf surface tension.

  In step S104, the input of the analysis result of the sap collected at the time of imaging | photography of the image received by step S101, S102 of the plant made into learning object is received. Examples of the analysis result of the sap here include the moisture content in the sap and the contents of various nutrient components. The analysis of sap may be requested from a specialized institution and may take a certain period of time. Therefore, it is good also as a structure which receives an input separately about the analysis result of a sap at a later date.

  In step S105, the input of the analysis result of the soil at the time of imaging | photography of the image received in step S101, S102 of the plant made into learning object is received. Examples of soil analysis results include soil temperature and humidity, mineral values, fertilizer components, changes in PH values, and bacterial group analysis results. As for the analysis result of soil, as with the analysis result of sap, it may take a period of time until a request or result is obtained from a specialized organization. It is good also as a simple structure.

  In step S106, the input of the information regarding the air | atmosphere and the weather at the time of imaging | photography of the image received in step S101, S102 of the plant made into learning object is received. Examples of the information on the atmosphere and weather here include temperature, humidity, ultraviolet light amount, and carbon dioxide concentration. In addition, the information regarding the atmosphere and the weather receives the input of the growing point of the plant to be learned by the expert terminal device 2, and acquires the weather information of the growing point from an external server device (not shown). You may acquire by a process.

  In step S107, the input of the information regarding the treatment performed with respect to the plant made into learning object is received. Examples of the information relating to the treatment here include information such as the amount of moisture given to the plant, the type and amount of fertilizer given, and the like. Information about the amount of water given to the plant and fertilizer can also be inferred from the analysis result of the soil that was previously input in step S105. Therefore, it may be configured to omit receiving a separate input here. In addition, when receiving input of information at a plurality of points in time with respect to a plant to be learned as will be described later, the information for receiving input is to receive input of information regarding treatment performed after the previous input of information. What is necessary is just to make it a structure. In addition, if the plant to be learned is a plant that blooms flowers, it may be configured to accept input of additional information such as information about flowering, or information about harvesting if it is a crop.

  After the reception of the image of the plant to be learned and the reception of the input of information are completed as described above, in step S108, each of the image storage unit 102, the health state storage unit 103, and the environment information storage unit 104 is set. Images and information are recorded, and the process ends.

  Here, for the sake of simplicity of explanation, a process of sequentially receiving an image of a plant to be learned and receiving an input of health status and environmental information is shown, but this is merely an example, and the present invention is This is not a limitation. The order of image reception and information input may be changed from the processing illustrated in FIG. 3, or information input may be received by other methods such as a configuration that receives batch input. Further, a configuration may be adopted in which input of some information is omitted within a range that does not hinder learning by the later learning unit 105.

  In addition, here, a configuration is shown in which an input of an image, a health state, and environmental information at a certain point in time for a plant to be learned is received. However, it is more preferable to receive input of a plurality of information over time. For this purpose, that is, a daily image together with an image of a plant at a plurality of dates and times, a configuration that collectively accepts input of health status and environmental information at each time point, and information for identifying an individual plant It may be configured to accumulate information by accepting input of health information and environmental information.

  In the plant health diagnosis system according to the present embodiment, in order to diagnose a health condition for a plurality of varieties of plants, when inputting the information to be learned as described above, inputting the target plant varieties Are preferably received and accumulated together.

<Learn how to diagnose and decide how to cope>
In this way, the learning means 105 learns the method for diagnosing the health state of the plant using the information as the learning target accumulated in the image storage unit 102, the health state storage unit 103, and the environment information storage unit 104. . This learning is preferably performed by machine learning such as deep learning or reinforcement learning. In other words, the output data is estimated from the input data using the veins and leaves of the plant to be learned at a certain point in time or the whole plant image as input data, and the health state at the time of capturing each image as the output data. By performing a process such as generating a model for the purpose, a health condition can be diagnosed later from the image of the plant to be diagnosed by the diagnosis unit 107 using this model.

  More detailed processing of learning is performed by calculating the feature amount from each of the learning target images stored in the image storage unit 102 and the correspondence relationship between the calculation and the information stored in the health state storage unit 103 and the environment information storage unit 104. A model is generated. Here, as a feature amount, it is preferable to calculate a value related to the form of a vein to be learned from at least an image stored in the leaf surface / leaf vein image storage unit 1021 and use it. A leaf vein is one in which sap flows, and its thickness and the presence or absence (continuity) of a cut portion are very relevant to the health of the plant. Therefore, a diagnostic method for performing a more accurate diagnosis can be learned by using a feature amount including a value calculated from the form of such a vein.

  Here, in addition to the plant image stored in the image storage unit 102, a more detailed diagnosis method can be performed by adding information on the growth environment of the plant to be learned stored in the environment information storage unit 104. In addition, it is possible to learn how to determine a recommended action for the result. For example, if there is a problem with the health condition of the plant, where is the cause of the problem with respect to soil analysis results, information about the atmosphere and weather, actions taken on the plant, and how to deal with it? It is possible to learn a method for performing a more detailed diagnosis, such as whether or not to take appropriate measures. In addition, when information related to the harvest of a plant to be learned and flowering information are recorded, it is possible to learn the optimal time for harvesting and a method for predicting the flowering time based on the information.

  In addition, by using time-lapse information about the plants to be learned, it is possible to conduct integrated learning based on image information, health status, soil and air information at each point in time, and treatments performed between them. It can be carried out.

  The diagnostic method for determining the health state of the plant learned in this way and the recommended countermeasure according to the health state and environmental information is stored in the diagnostic method storage unit 106. When learning the diagnosis method, in addition to the information stored in the image storage unit 102, the health state storage unit 103, and the environment information storage unit 104, other storage units (not shown), external server devices, and the like are referred to. It is good also as a structure. For example, by referring to a storage unit that stores information on the symptoms and countermeasures of plant diseases and an external server device, it can be expected to derive a health condition of the plant and a diagnosis method for dealing with it more appropriately. In addition, if it is configured to refer to a storage unit that stores information on each component of a commercially available fertilizer product or an external server device, what kind of fertilizer is used when the plant lacks specific nutrients, etc. It is possible to present more specific countermeasures such as how much should be given.

<Diagnosis>
After the diagnostic method is recorded in the diagnostic method storage unit 106 by the learning unit 105 as described above, the information is used to determine the health state of the plant by the diagnostic unit 107 and a countermeasure based on the determination. Can make suggestions. FIG. 4 is a flowchart showing the flow of the health check process by the diagnosis unit 107. Note that this process is performed by communication using the HTTP (Hyper Text Transfer Protocol) between the programs, in which the medical examination server device 1 has a Web server program and the user terminal device 3 has a Web browser program. It is good also as a structure, or the user terminal device 3 is equipped with the program for exclusive use for performing the transmission of the image of the plant made into a diagnosis object, and the display of a diagnostic result, and is used with the health diagnostic server apparatus 1 using it A configuration may be adopted in which information is transmitted and received by an arbitrary communication protocol.

  In the health check process, first, in step S201, an image of a leaf vein or leaf surface of a plant to be diagnosed is received from the user terminal device 3. When the user terminal device 3 includes a camera, the image here may be configured to prompt the user to capture and transmit an image, or may be captured in advance and the auxiliary storage device included in the user terminal device 3. It may be configured to prompt selection and transmission of images stored in the

  In subsequent step S202, an image of the whole plant image that has received the leaf vein and leaf surface images in step S201 is received. Here, similarly to step S201, it may be configured to prompt transmission of an image photographed by the user terminal device 3 or an image stored in the auxiliary storage device of the user terminal device 3. Here, for the sake of simplicity of explanation, an image of the whole image is used. However, the image may not necessarily include the entire plant, that is, the veins received in step S201 that include a plurality of leaves, stems, and the like. As long as the image includes a wider area than an image including a leaf surface. Moreover, it is good also as a structure which receives several images, such as an image including the upper part of a plant, and a root image.

  Note that the images of the veins and leaves of the diagnosis target received in step S201 and the whole image of the diagnosis target received in step S202 correspond to the learning target images received in steps S101 and S102, respectively. It is preferable that the image is taken under conditions. By doing so, the diagnosis by the diagnosis method learned by the learning means 105 can be performed more effectively.

  After receiving the image of the plant to be diagnosed as described above, the health condition is determined by the diagnostic means 107 in step S203. In this process, for example, the feature value to be noted about the diagnosis target image is identified with reference to the diagnosis method storage unit 106, and the feature value is calculated by analyzing the image received in steps S201 and S202. Thus, it can be performed by a process of deriving the health state of the plant to be diagnosed from the relationship between the feature amount recorded in the diagnosis method storage unit 106 and the health state. Note that the feature amount used here corresponds to the feature amount calculated from the image stored in the image storage unit 102 by the learning unit 105 as described above. That is, in particular, it is preferable to use a feature amount including a value related to a leaf vein form of a plant to be diagnosed.

  In subsequent step S204, it is determined whether or not there is a problem in the health condition of the plant to be diagnosed with respect to the result obtained in step S203. Examples of the problem here include an excess or deficiency of water content, an excess or deficiency of specific nutrients, and a possibility of suffering from a disease.

  If it is determined in step S204 that there is a problem with the health condition, the process proceeds to step S205, and the diagnosis method storage unit 106 is referred to confirm the countermeasure against the problem. Here, for example, there are measures such as reducing the amount of water to be supplied if the amount of water is too much, and measures such as providing fertilizer to supplement it if any nutrients are insufficient.

  As described above, after the diagnosis related to the diagnosis target that has received the image from the user terminal device 3 is completed, the diagnosis result is transmitted to the user terminal device 3 in step S206, and the process ends. The user refers to the diagnosis result displayed on the user terminal device 3 and if there is no problem in the health condition of the plant to be diagnosed, and what kind of countermeasure is recommended when there is a problem, Can be obtained.

  In addition, although the process which shows the countermeasure with respect to when there exists a problem in the health state of a plant was shown here, it is good also as a structure which presents information, such as a countermeasure, other than that. For example, the diagnosis unit 107 may obtain and output information such as prediction of the optimal time for harvesting the fruits and leaves of the plant to be diagnosed and prediction of flowering time.

  In addition, when the health check system for plants according to the present embodiment is configured to be able to make a diagnosis of health conditions for a plurality of varieties of plants, It is preferable to receive input of varieties.

  As described above, it is possible to appropriately determine the health state of a plant by receiving an input of an image of a plant to be learned and a health state at the time of shooting and learning a health condition diagnosis method based on the input. By transmitting an image of the plant to be diagnosed, the user can receive an indication of the health state of the plant and appropriate countermeasures against it. In addition, by receiving input of environmental information together with the health state of the plant to be learned, it can be expected to learn a method for diagnosing the health state with higher accuracy.

  In particular, in the learning of diagnostic methods, by using features related to leaf vein morphology such as thickness and continuity, learning of effective diagnostic methods, and performing highly accurate diagnosis using the learning results Can do.

DESCRIPTION OF SYMBOLS 1 Health diagnostic server apparatus 101 Learning object receiving means 102 Image storage part 1021 Leaf / leaf image storage part 1022 Whole image storage part 103 Health state storage part 1031 Tree vigor information storage part 1032 Sap analysis result storage part 104 Environmental information storage part 1041 Soil Analysis result storage unit 1042 Atmosphere / weather information storage unit 1043 Treatment information storage unit 105 Learning unit 106 Diagnostic method storage unit 107 Diagnosis unit 2 Expert terminal device 3 User terminal device NW network

Claims (7)

A plant health diagnosis system for diagnosing a health condition from an image of a plant,
With reference to the learning target image that is an image of the plant to be learned and the health state of the learning target at the time of shooting the learning target image, the health state of the plant from the image of the plant to be diagnosed Learning means for learning diagnostic methods;
Diagnostic means for receiving an input of a diagnostic target image obtained by photographing a plant that is a diagnostic target, and outputting a diagnosis result of a health condition of the diagnostic target based on the diagnostic method ,
The learning target image includes at least an image of the learning target vein image;
The learning means learns the diagnostic method based on the form of the veins to be learned,
The diagnostic target image includes at least an image of the diagnostic target vein image;
The diagnosis means outputs the diagnosis result based on the form of the vein of the diagnosis object;
The plant health diagnosis system, wherein the learning object vein form and the diagnosis object leaf form include at least one of the thickness of the vein and the continuity of the vein . The plant health diagnosis system according to claim 1, wherein the health condition of the learning target includes at least an analysis result of the sap of the learning target. The plant health diagnosis system according to claim 1 , wherein the learning unit further refers to information related to a growth environment of the learning target at the time of photographing the learning target image. . The learning means further learns a method of determining a countermeasure according to the health condition of the learning target,
The plant health diagnosis system according to any one of claims 1 to 3 , wherein the diagnosis result includes the countermeasure based on a health condition of the diagnosis target. The plant health diagnosis system according to any one of claims 1 to 4 , wherein the learning means performs machine learning. A plant health diagnosis program for diagnosing a health condition from an image of a plant, comprising a computer device,
With reference to the learning target image that is an image of the plant to be learned and the health state of the learning target at the time of shooting the learning target image, the health state of the plant from the image of the plant to be diagnosed Learning means for learning diagnostic methods;
Receiving an input of a diagnostic object image obtained by photographing a plant that is a diagnostic object, and operating as a diagnostic means for outputting a diagnosis result of a health condition of the diagnostic object based on the diagnostic method ;
The learning target image includes at least an image of the learning target vein image;
The learning means learns the diagnostic method based on the form of the veins to be learned,
The diagnostic target image includes at least an image of the diagnostic target vein image;
The diagnosis means outputs the diagnosis result based on the form of the vein of the diagnosis object;
The plant health diagnosis program characterized in that the form of the veins of the learning object and the form of the veins of the diagnosis object include at least one of the thickness of the veins and the continuity of the veins . A plant health diagnosis method using a computer device that diagnoses a health condition from an image of a plant,
The computer device refers to a learning target image that is an image of a plant that is a learning target and a health state of the learning target at the time of shooting the learning target image, and the image from the plant image that is a diagnosis target Learning how to diagnose plant health,
The computer device receives an input of a diagnostic target image obtained by photographing a plant that is a diagnostic target, and outputs a diagnosis result of a health condition of the diagnostic target based on the diagnostic method ,
The learning target image includes at least an image of the learning target vein image;
In the step of learning the diagnostic method, the diagnostic method is learned based on the morphology of the veins to be learned,
The diagnostic target image includes at least an image of the diagnostic target vein image;
In the step of outputting the diagnosis result, the diagnosis result based on a form of a vein of the diagnosis target is output,
The plant health diagnosis method, wherein the learning object vein form and the diagnosis object leaf form include at least one of the thickness of the vein and the continuity of the vein .