JP6609681B1 - Determination system, determination apparatus, determination method, and computer program - Google Patents

Abstract

[PROBLEMS] To predict a more appropriate harvest timing of agricultural products. An environmental information sensor that is installed in a farm where a specific agricultural product is cultivated and acquires environmental information indicating an environment in the farm, and an imaging that is installed in the farm and captures an image of the farm product. Based on the apparatus, the environmental information, and the image information corresponding to the growth time of the agricultural product among the image information that is information obtained from the image of the agricultural product, a predetermined environmental value until the appropriate harvesting of the agricultural product And a determination device that determines a remaining integrated value that is an integrated value of. [Selection] Figure 1

Description

  The present invention relates to a technique for estimating the growth state of agricultural products.

  Fruit vegetables such as tomatoes have a complicated growth process. Therefore, it is difficult to accurately predict the growth state. Such a problem is not limited to fruits and vegetables, but is a problem common to all agricultural products.

  As a technique for solving such a problem, techniques disclosed in Patent Documents 1 to 3 have been proposed. In patent document 1, the system which suppresses the dispersion | variation in the quality of crops or completion | finish is proposed. Patent Document 2 proposes a technique for estimating the end with high accuracy by using a satellite image or the like. Patent Document 3 proposes a technique for predicting the amount of agricultural products that satisfy the shipping standards based on the amount of agricultural products delivered from the producer.

JP, 2006-049102, A JP2015-000049A Japanese Patent Laid-Open No. 2006-118895

However, due to factors such as changes in the weather just before the harvest time, the appropriate harvest time often varies. Therefore, it was difficult to predict an appropriate harvest date for agricultural products.
In view of the above circumstances, an object of the present invention is to provide a technique capable of predicting a more appropriate harvest timing of agricultural products.

  One aspect of the present invention is an environmental information sensor that is installed in a plantation where a specific agricultural product is cultivated, acquires environmental information indicating an environment in the plantation, and an image of the agricultural product that is installed in the plantation. Based on the imaging device for imaging, the environment information, and the image information corresponding to the growth time of the agricultural product among the image information that is information obtained from the image of the agricultural product, predetermined until the appropriate harvest of the agricultural product A determination apparatus that determines a remaining integrated value that is an integrated value of the environmental values.

  One aspect of the present invention is environmental information indicating an environment in a plantation where a specific agricultural product is cultivated, and image information corresponding to a growth period of the agricultural product among image information that is information obtained from the image of the agricultural product. And a determination unit that determines a remaining integrated value that is an integrated value of a predetermined environmental value until the agricultural product is appropriately harvested.

  One aspect of the present invention is the above-described determination device, in which the determination unit is further configured to acquire actual environment information, actual image information, and the actual image information acquired before harvesting about the agricultural products that are actually harvested at an appropriate timing. Based on the actual environment information or the actual value of the remaining integrated value from the acquisition of the actual image information to the harvest timing, the result of the learning process performed for each of the agricultural products belonging to the same growth period is used. Thus, the remaining integrated value is determined.

  One aspect of the present invention is the step in which an environmental information sensor installed in a cultivation place where a specific agricultural product is cultivated acquires environmental information indicating the environment in the cultivation place, and an imaging device installed in the cultivation place Is based on the step of capturing an image of the agricultural product, the environmental information, and image information corresponding to a growth period of the agricultural product among image information that is information obtained from the image of the agricultural product. Determining a remaining integrated value that is an integrated value of a predetermined environmental value until appropriate harvesting.

  One aspect of the present invention is environmental information indicating an environment in a plantation where a specific agricultural product is cultivated, and image information corresponding to a growth period of the agricultural product among image information that is information obtained from the image of the agricultural product. And a determination step of determining a remaining integrated value that is an integrated value of a predetermined environmental value until the agricultural product is appropriately harvested.

  One embodiment of the present invention is a computer program for causing a computer to function as the determination system.

  One embodiment of the present invention is a computer program for causing a computer to function as the determination device.

  According to the present invention, it is possible to predict a more appropriate harvest timing of agricultural products.

It is a schematic block diagram which shows the system configuration | structure of the determination system 200 of this invention. 3 is a schematic block diagram illustrating an exemplary functional configuration of a determination device 60. FIG. It is a figure which shows the concept of a remaining integrated value. It is a figure which shows the specific example of classification | category information. It is a figure which shows the outline of an integrated value. 5 is a flowchart showing a specific example of a learning process flow of the determination apparatus 60. It is a sequence chart which shows the specific example of the flow of the determination process of the determination system.

Hereinafter, specific configuration examples of the present invention will be described with reference to the drawings.
FIG. 1 is a schematic block diagram showing a system configuration of a determination system 200 of the present invention. The determination system 200 is a system for determining an appropriate harvest timing of the agricultural product to be determined.

  First, the network configuration related to the determination system 200 will be described. The determination system 200 includes a plurality of environment information sensors 10, a plurality of imaging devices 20, a control device 30, a log data storage device 40, a terminal device 50, and a determination device 60. The plurality of environment information sensors 10 and the plurality of imaging devices 20 are connected to the control device 30 so as to communicate with each other. Communication performed by the plurality of environment information sensors 10, the plurality of imaging devices 20, and the control device 30 may be realized by wireless communication or may be realized by wired communication. Control device 30 and log data storage device 40 are communicably connected. The control device 30, the terminal device 50, and the determination device 60 are connected to be communicable via the network 80. The network 80 is configured using a communication network such as the Internet or a public line communication network. The determination device 60 is communicably connected to the weather forecast information providing device 70 via the network 80.

Next, each device related to the determination system 200 will be described.
The plurality of environmental information sensors 10 and the plurality of imaging devices 20 are installed in the cultivation place 100 to be determined. In the cultivation place 100, one or more kinds of specific agricultural products are cultivated. For example, in the cultivation place 100, fruit and vegetables such as tomatoes and eggplants are cultivated. In this case, the determination device 60 of the determination system 200 determines an appropriate harvest timing for fruit and vegetables such as tomatoes and eggplants cultivated in the cultivation place 100.

  The environment information sensor 10 acquires specific information (hereinafter referred to as “environment information”) related to the environment of the farm 100. The environment information sensor 10 acquires any one or more of values such as air temperature, humidity, precipitation, sunshine duration, air volume, etc. at the cultivation place 100, for example. As the environment information sensor 10, one type of sensor may be installed in the cultivation plant 100, or a plurality of types of sensors may be installed in the cultivation plant 100. The environmental information sensor 10 may be fixedly installed at a predetermined position, or may be movably installed. Each environmental information sensor 10 transmits data indicating the acquired environmental information to the control device 30. The data indicating the environmental information transmitted from the environmental information sensor 10 may be transmitted directly to the control device 30 by wireless communication or wired communication, or may be transmitted to the control device 30 via a relay device (not shown). Good.

  The imaging device 20 photographs the agricultural products to be determined that are cultivated in the cultivation place 100. The imaging device 20 may be configured to shoot agricultural products from a predetermined imaging position in a predetermined imaging direction with predetermined imaging parameters (view angle, magnification, etc.). The imaging device 20 may be fixedly installed at a predetermined position and in a predetermined direction, may be installed at a predetermined position in a form that can change the shooting direction, or may be installed so as to be movable. . The imaging device 20 may be configured to move in the cultivation place 100 by being fixed to a moving body that moves along a predetermined guide, for example, or cultivated by being fixed to a flying body such as a drone. It may be configured to move within the location 100. The imaging device 20 transmits data of an image obtained by photographing (hereinafter referred to as “captured image”) to the control device 30. The imaging device 20 may transmit the identification information of the own device in association with the captured image data. The captured image data transmitted from the imaging device 20 may be transmitted directly to the control device 30 by wireless communication or wired communication, or may be transmitted to the control device 30 via a relay device (not shown).

  The control device 30 is configured using an information processing device such as a personal computer or a PLC (Programmable Logic Controller). The control device 30 receives environmental information data from the environmental information sensor 10. The control device 30 records the received environmental information data in the log data storage device 40 in association with the date / time information. The date and time information to be recorded may be date and time information when the environment information data is acquired by the environment information sensor 10, or date and time information when the environment information data is received by the control device 30, or log data. Date and time information recorded in the storage device 40 may be used. The control device 30 receives captured image data from the imaging device 20. The control device 30 records the received captured image data in the log data storage device 40 in association with the date / time information. The date and time information to be recorded may be date and time information when the captured image data is acquired by the imaging device 20, date and time information when the captured image data is received by the control device 30, or log data storage. Date / time information recorded in the device 40 may also be used.

  The log data storage device 40 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The log data storage device 40 stores a log of data obtained by the environment information sensor 10 and the imaging device 20 installed in the cultivation place 100.

  The terminal device 50 is a device having an information processing function such as a personal computer, a server, a smartphone, a tablet, a smart speaker, or a television receiver. The terminal device 50 is used, for example, by a user such as an administrator or manager of the cultivation plant 100. The terminal device 50 receives information regarding the harvest time of agricultural products from the determination device 60 via the network 80. The terminal device 50 outputs the received information to the user in a predetermined manner. For example, the terminal device 50 may output information on the harvest time of agricultural products by displaying an image on a display, or may output information on the harvest time of agricultural products by voice.

  The determination device 60 is configured using an information processing device such as a personal computer or a server device. FIG. 2 is a schematic block diagram illustrating a functional configuration example of the determination device 60. The determination device 60 includes a communication unit 61, a log storage unit 62, a learning information storage unit 63, a classification information storage unit 64, a learning result storage unit 65, and a control unit 66.

  The communication unit 61 is configured using a communication interface. The communication unit 61 performs data communication with other devices via the network 80. For example, the communication unit 61 receives data from the control device 30. For example, the communication unit 61 transmits data indicating the determination result to the terminal device 50.

  The log storage unit 62 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The log storage unit 62 stores environment information and captured image logs. The log storage unit 62 stores, for example, a record in which date information and environment information are associated with each other. The log storage unit 62 further stores, for example, a record in which date / time information and captured images are associated with each other.

  The learning information storage unit 63 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The learning information storage unit 63 stores learning information. The learning information is data used when the learning unit 663 of the control unit 66 performs a learning process. The learning information includes, for example, evaluation information indicating whether or not the timing of harvesting was appropriate for agricultural products that have already been harvested in the cultivating place 100, and photographed images of individual agricultural products (hereinafter “ Real image information, which is information obtained from “real individual image”), real environment information indicating environmental information acquired before harvesting, and the remaining total at the timing when the data of the actual individual image or the real environment information is acquired The information which matched the information which shows the actual value of a value may be sufficient. The real image information may be, for example, information indicating the color of the agricultural product shown in the actual individual image, or may be information indicating the size of the agricultural product on the actual individual image.

  The learning information includes, for example, real image information obtained from real individual images taken before harvesting, and real environment information acquired before harvesting, for the agricultural products that have already been harvested at an appropriate timing in the cultivation plant 100. It may be information associated with information indicating the actual value of the remaining integrated value at the timing when the data of the actual individual image or the actual environment information is acquired. The remaining integrated value indicates an integrated value of specific environmental information values from that day to the optimal harvest timing. The learning information may further be associated with weather forecast information at the time when each real environment information or real individual image is acquired. Weather forecast information is information which shows the weather forecast of the area where the cultivation place 100 is included. The weather forecast information is information indicating forecast values such as weather, temperature, humidity, and pressure after the target day.

  FIG. 3 is a diagram showing the concept of the remaining integrated value. In the example of FIG. 3, the environmental information represented by the remaining integrated value is a temperature value. A value accumulated from a certain reference date to an optimal harvesting timing when the temperature of the day (for example, maximum temperature, average temperature, minimum temperature, etc.) is integrated every day is referred to as a remaining integrated value. In the example of FIG. 3, the remaining integrated value is 0 degree on the harvest day that is the optimal timing. A remaining integrated value of 0 degrees indicates that the optimum timing for harvesting. When the remaining integrated value is 40 degrees, it indicates that the timing at which the temperature is 40 degrees when the temperature is integrated every day from that day is the optimum harvesting timing. Therefore, when the remaining integrated value is 40 degrees and the temperature of the day is 40 degrees, it indicates that the next day is the optimum harvest timing. When the remaining integrated value is 40 degrees and the temperatures on the current day and the next day (after one day) are 20 degrees, respectively, it indicates that two days later is the optimum harvest timing. FIG. 3 shows such an example. Even if the remaining integrated value is 40 degrees, if the temperature is 10 degrees every day after that day, four days later indicates the optimum harvest timing. It should be noted that, based on the value of the remaining integrated value, the harvester may appropriately change the harvesting timing to a more suitable one. For example, the harvesting timing may be changed according to circumstances such as whether the sale is performed immediately after harvesting or whether the sale is performed after being transported to a remote place.

  The correspondence relationship between each remaining integrated value and the number of days until harvest shown in FIG. 3 is only a specific example. Even if the remaining integrated value is the same, the number of days until the optimum harvesting changes depending on the value of the temperature after that day.

  Returning to the description of FIG. The learning information stored in the learning information storage unit 63 is given in advance to the determination device 60 by a predetermined means. For example, the learning information may be given by recording learning information recorded in a storage medium such as a USB memory or a CD-ROM in the learning information storage unit 63. For example, the learning information may be given to the learning information storage unit 63 of the determination device 60 by transmitting the learning information from another information processing device to the determination device 60. For example, with respect to the log information stored in the log storage unit 62, evaluation information indicating whether or not the timing of harvesting is appropriate for the user, and an actual individual image extracted from the captured image recorded as the log information Learning information is generated by adding each of the actual environment information or the actual value of the remaining integrated value corresponding to the actual individual image calculated based on the actual harvest timing, and the learning information storage unit 63 may be provided.

  The classification information storage unit 64 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The classification information storage unit 64 stores classification information. The classification information is information used to classify each captured image at which time (hereinafter referred to as “growth time”) between the flowering time and the date appropriate for harvesting. Further, the classification information may further include information indicating which learning result is used when the determination process is performed using the captured image at the time of each classification result.

  FIG. 4 is a diagram illustrating a specific example of classification information. The classification information shown in FIG. 4 includes a plurality of records having three values: a classification identifier, a classification condition, and a usage model. The classification identifier indicates the identifier of the group in the classification result. The classification condition indicates a condition for classification into a group indicated by the classification identifier. The usage model indicates the type of feature amount used for learning processing and determination processing for images classified into the group indicated by the classification identifier.

  For example, the record described in the first row has values of “A”, “H> 25”, and “R / G”, respectively, as a classification identifier, a classification condition, and a usage model. H represents the average hue of the captured image. R represents the average value of the R component of the image. G indicates the average value of the G component of the image. This record is classified into the group of identifiers A when the average value of hues of all the pixels of the photographed image is greater than 25, and the determination processing is performed for the photographed images classified into the group of identifiers A. In the case where the learning result is obtained, the learning result obtained by the learning process using the R / G value of the actual individual image as the feature amount is used. Note that the group of identifiers A is classified with the intention of being crops between just before harvesting and 4 days before harvesting.

  For example, the record described in the second row has values of “B”, “H ≦ 25”, and “R / G + integrated value” as a classification identifier, a classification condition, and a usage model, respectively. The integrated value indicates an integrated value of specific environmental information values from the time of flowering until the day when the image was taken. The integrated value may be, for example, an integrated value of the temperature from the time of flowering until the day when the image was taken. FIG. 5 is a diagram showing an outline of the integrated value. FIG. 5 shows an example where the integrated value is an integrated value of temperature. On the day of flowering, the integrated value is 0 ° C. Thereafter, the integrated value increases as the day passes. The second record in FIG. 4 is classified into the group of identifiers B and the photography classified into the group of identifiers B when the average value of the hues of all the pixels of the photographed image is 25 or less. When determination processing is performed on an image, learning obtained by learning processing using a combination of the R / G value of the actual individual image and the integrated value at the time when the actual individual image was captured as a feature amount The results are used. Note that the group of identifiers B is classified with the intention of being crops before 5 days before harvest.

Although the average value of hue is shown as a specific example of the classification condition, other conditions can be used as long as the image can be classified at which time from the flowering time to the date suitable for harvesting. May be used.
Although only R / G, R / G, and the integrated value are used as the feature amount of the use model, the learning process may be performed using another feature amount. For example, an average value of hue (H) may be used as a feature amount, or a combination of H and an integrated value may be used as a feature amount.

  The learning result storage unit 65 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The learning result storage unit 65 stores information indicating the learning result obtained by the learning process of the learning unit 663 of the control unit 66. As will be described later, the learning unit 663 performs a learning process for each group that is a classification result by the classification unit 662. Therefore, the learning result storage unit 65 stores information indicating the learning result for each group. For example, the learning result storage unit 65 stores learning results belonging to the group A and learning results belonging to the group B, respectively. The learning result stored by the learning result storage unit 65 is used by the determination unit 665.

  The control unit 66 is configured using a processor such as a CPU (Central Processing Unit) connected to the bus and a memory. The control unit 66 functions as a log control unit 661, a classification unit 662, a learning unit 663, a detection unit 664, a determination unit 665, and an information providing unit 666 when the processor executes a program.

  Note that all or a part of the log control unit 661, the classification unit 662, the learning unit 663, the detection unit 664, the determination unit 665, and the information providing unit 666 realized by the processor executing the program is ASIC (Application Specific Integrated). It may be realized by using hardware such as a circuit (PLC), a programmable logic device (PLD), or a field programmable gate array (FPGA). The program may be recorded on a computer-readable recording medium. The computer-readable recording medium is a portable medium such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, a semiconductor storage device (for example, SSD: Solid State Drive), a hard disk built in a computer system, or a semiconductor storage. A storage device such as a device. The program may be transmitted via a telecommunication line.

  The log control unit 661 receives environment information and captured image data from the control device 30. The log control unit 661 records the received data in the log storage unit 62 as a log in association with the date / time information.

  The classification unit 662 classifies the actual individual image used for the learning process and the individual image used for the determination process at which time from the flowering time to the date appropriate for harvesting. The classification unit 662 performs a classification process based on the classification condition of the classification information stored in the classification information storage unit 64. For example, the classification unit 662 classifies the image into one of a group of the classification identifier A and a group of the classification identifier B.

  The learning unit 663 uses, for each classified group, among the learning information stored in the learning information storage unit 63, only the feature amount defined in the use model of the learning information belonging to the group. Machine learning. The learning unit 663 acquires a learning result for obtaining a value of the remaining integrated value until the optimum harvest among a plurality of values included in the learning information. For the learning unit 663, for example, a machine learning technique for generating or updating a discriminator by supervised learning is applied. Specific examples of the machine learning technique include SVM (Support Vector Machine) and DL (Deep Learning). For example, when learning processing is performed using real individual images belonging to group A, the learning unit 663 uses R / G of each real individual image as a feature amount. For example, when learning processing is performed using real individual images belonging to group B, the learning unit 663 uses a combination of R / G and integrated value of each real individual image as a feature amount.

  The detection unit 664 detects an image of an individual agricultural product to be processed from the captured image. For example, when the agricultural product to be processed is a tomato and 10 tomatoes being grown are shown in the photographed image, the detection unit 664 detects each of the 10 tomatoes. The detection unit 664 extracts an image of each detected tomato (hereinafter referred to as “individual image”). The detection unit 664 determines the position of the agricultural product shown in the extracted individual image in the cultivation place 100 based on the identification information of the imaging device associated with the captured image. The detection unit 664 outputs each extracted individual image to the classification unit 662 and the determination unit 665 as an image to be processed.

  The determination unit 665 acquires current environment information from the log control unit 661. Further, the determination unit 665 obtains an individual image obtained by photographing individual agricultural products to be subjected to the determination process from the detection unit 664 and a classification result by the classification unit 662 for the individual image. The determination unit 665 acquires, as image information, a feature amount defined as a usage model corresponding to the classification result from the individual image. The determination unit 665 captures an individual image based on the acquired environment information and image information and the learning result corresponding to the group of the classification results among the learning results stored in the learning result storage unit 65. Determine the remaining integrated value of agricultural products. The determination unit 665 may further use weather forecast information for the determination process in addition to the environment information and the image information. In this case, the learning process needs to be performed so that the determination process is performed based on the weather forecast information. The weather forecast information may be acquired from the weather forecast information providing device 70 via the network 80, for example. The weather forecast information providing device 70 is a server installed by, for example, the Japan Meteorological Agency, a specific corporation, or an individual, and is a device that provides weather forecast information.

  The information provision unit 666 generates provision information including information indicating the determination result by the determination unit 665. For example, the provided information includes the individual image subjected to the determination process, the remaining integrated value indicating the result of the determination process performed by the determination unit 665 for the individual image, and the agricultural product cultivation place 100 captured in the individual image. And information indicating a place.

  FIG. 6 is a sequence chart illustrating a specific example of the flow of the learning process of the determination device 60. First, the classification unit 662 classifies the learning information stored in the learning information storage unit 63 into each group based on the classification condition of the classification information (step S101). The learning unit 663 acquires a feature amount defined in the classification information use model for each group, and performs a learning process (step S102). The learning unit 663 records the learning result obtained for each group in the learning result storage unit 65 (step S103). With the above processing, the learning processing by the determination device 60 is completed.

  FIG. 7 is a sequence chart showing a specific example of the flow of determination processing of the determination system 200. Each environmental information sensor 10 acquires environmental information at a predetermined timing and transmits it to the control device 30 (step S201). Each imaging device 20 captures an image at a predetermined timing, and transmits the captured image to the control device 30 (step S202).

  The control device 30 records the environment information received from the environment information sensor 10 and the captured image received from the imaging device 20 in the log data storage device 40 (step S203). And the control apparatus 30 transmits the environmental information received from the environmental information sensor 10, and the picked-up image received from the imaging device 20 to the determination apparatus 60 (step S204).

  The log control unit 661 of the determination device 60 receives environment information and a captured image from the control device 30. The log control unit 661 records the received environment information and captured image in the log storage unit 62 as a log in association with the time information (step S205). The detection unit 664 detects the agricultural product that is shown as a subject in the newly acquired photographed image (step S206). The detection unit 664 adds a label to the detected image (individual image) of each agricultural product (step S207). The classification unit 662 classifies the individual images with the labels into groups (step S208). The determination unit 665 stores the received environment information, the image information (feature amount) obtained according to the group of the classification result regarding the image information of the agricultural product to be determined, and the learning information storage unit 63. Based on the learning result corresponding to the group of the classification results among the learning results, the remaining integrated value of the agricultural product to be determined is determined (step S209). The information provision unit 666 generates provision information including the result of the determination process (step S210). The information provision part 666 transmits provision information with respect to the terminal device 50 matched with the cultivation place 100 used as the object of the determination process (step S211). The terminal device 50 outputs the received provision information to the user.

  According to the determination system 200 configured as described above, the remaining integrated value is determined as information indicating an appropriate timing of harvesting for the agricultural products cultivated in the cultivation place 100. The value of the remaining integrated value itself is not easily affected by changes in the weather after the determination is made. That is, although the date that is a timing suitable for harvesting changes according to the actual temperature after the determination is made, the remaining integrated value itself at the determination timing is uniquely determined. By providing such a remaining integrated value to the user as information indicating an appropriate timing of harvesting, it is possible to more accurately determine a harvesting timing more suitable for the user.

  Such a determination system 200 is particularly effective for a type of agricultural product in which the integrated value of environmental information from a predetermined event (for example, flowering) to an optimal harvest timing in the agricultural product is constant. Examples of such agricultural products include fruit vegetables such as tomatoes. In FIG. 3 described above, the temperature is described as a specific example of the remaining integrated value, but the remaining integrated value is not necessarily an integrated value related to the temperature. For example, it may be an integrated value of sunshine hours.

  In particular, in the determination system 200, in the determination process for obtaining the remaining integrated value, image information (feature amount) corresponding to the classification result of the individual image obtained at that time and learning performed using the image information The learning result of the process is used. The classification result is information indicating at which time the image is from the flowering time to the date suitable for harvesting. Therefore, the determination is made using different feature amounts depending on which time period from the flowering time to the date suitable for harvesting the crop to be determined. In other words, depending on which time period from the flowering time to the appropriate date for harvesting, the crop subject to judgment depends on the feature quantity and learning results suitable for that time, so that The integrated value can be determined.

  In the determination system 200, the remaining integrated value is determined based not only on the environment information but also on the image information of the individual images. Therefore, the remaining integrated value can be determined more accurately than when the remaining integrated value is determined based only on the environmental information.

(Modification)
Any means may be employed as means for causing the environment information to reach the determination device 60 from each environment information sensor 10. In the above description, environmental information is transmitted from each environmental information sensor 10 to the determination device 60 via the control device 30. For example, the environmental information is transmitted from each environmental information sensor 10 to the determination device 60 without passing through the control device 30. May be. Further, for example, any one or more of the environmental information sensors 10 collect environmental information from other environmental information sensors 10, and collectively transmit one or more environmental information to the control device 30 or the determination device 60. Also good. The same applies to the imaging device 20.

  The determination device 60 is not necessarily configured as a single device. For example, the determination device 60 may be configured using a plurality of information processing devices. A plurality of information processing apparatuses constituting the determination apparatus 60 may be communicably connected via a communication path such as the network 80 and may be configured as a system such as a cluster machine or a cloud.

  The determination device 60 may be configured as a device or system that does not include the learning unit 663. In this case, the processing performed by the learning unit 663 described above may be executed by another device or system as a learning device. In this case, the determination device including the determination unit 665 acquires learning result data from the learning device and performs determination processing.

  The environmental information sensor 10 may be connected to a storage battery or a charger. By being configured in this way, it is possible to reduce restrictions on the installation location of the environment information sensor 10. For example, the environment information sensor 10 can be installed in a place where electricity is not passed. The environment information sensor 10 may be connected to a device that generates power using natural energy. For example, a solar panel or a generator such as a wind power generator may be connected. The environmental information sensor 10 may operate based on electric power generated by a device that performs such power generation.

  The determination unit 665 and the learning result storage unit 65 may be provided in the control device 30. With this configuration, it is possible to more quickly determine the remaining integrated value of agricultural products in each individual image.

  The learning unit 663 may perform a learning process for each set of a plurality of pieces of learning information. For example, the learning unit 663 may perform a learning process based on each learning information prepared in advance for each type or variety of agricultural product. In this case, the learning result storage unit 65 stores the learning result obtained for each type or each kind for each group of classification results. The determination unit 665 selects an optimal learning result to be used according to the type or variety of the agricultural product shown in the individual image to be processed, and the remaining integration is performed using the selected optimal learning result. Determine the value. With this configuration, the remaining integrated value can be determined with higher accuracy. The set of a plurality of learning information is not limited to the above-described patterns (for each type and each product type). For example, a plurality of sets of learning information may be used depending on differences in cultivation methods (house hydroponic, house soil plowing, alleys, etc.) and seasons (spring, summer, autumn, winter).

  In the above-described embodiment, an example in which a suitable feature amount (use model) is known for each group of classification results is shown. However, it may be unknown which feature quantity is suitable for each group of classification results. In this case, before the learning process is performed, the use model value of the classification information may not be recorded. In this case, the learning unit 663 may perform the learning process as follows. The learning unit 663 performs a learning process for each of a plurality of use model candidates defined in advance for each group. That is, a learning result is obtained for each combination of a group and a usage model. The determination unit 665 performs determination processing using the learning information and each learning result. At this time, the learning information used for the learning process may be used for the determination process, or the learning information that is not used for the learning process (prepared for the determination process) is used for the determination process. May be. The learning unit 663 compares the obtained determination result with the actual value of the remaining integrated value of the learning information for each group, and selects the learning result with the smallest difference. The learning unit 663 records the use model used for the selected learning result as the value of the use model in the classification information storage unit 64, and records the selected learning result in the learning result storage unit 65. With this configuration, a suitable usage model is automatically selected even if it is unknown which usage model is optimal for each group.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes designs and the like that do not depart from the gist of the present invention.

DESCRIPTION OF SYMBOLS 100 ... Cultivation place, 200 ... Determination system, 10 ... Environmental information sensor, 20 ... Imaging device, 30 ... Control device, 40 ... Log data storage device, 50 ... Terminal device, 60 ... Determination device, 61 ... Communication part, 62 ... Log storage unit, 63 ... learning information storage unit, 64 ... classification information storage unit, 65 ... learning result storage unit, 66 ... control unit, 661 ... log control unit, 662 ... classification unit, 663 ... learning unit, 664 ... detection Part, 665... Determination part, 666.

Claims (7)

An environmental information sensor that is installed in a farm where a specific agricultural product is grown, and that acquires environmental information indicating the environment in the farm;
An imaging device installed in the plantation for capturing images of the agricultural products;
Based on the environmental information and image information corresponding to the growth time of the agricultural product among image information that is information obtained from the image of the agricultural product, an integrated value of a predetermined environmental value until the appropriate harvesting of the agricultural product A determination device for determining the remaining integrated value,
A determination system comprising:   Based on environmental information indicating an environment in a cultivation place where a specific agricultural product is cultivated, and image information corresponding to a growth time of the agricultural product among image information that is information obtained from the image of the agricultural product, the agricultural product A determination apparatus comprising: a determination unit that determines a remaining integrated value that is an integrated value of a predetermined environmental value until the appropriate harvesting.   Further, the determination unit is configured to acquire the real environment information, the real image information, and the real environment information or the real image information that have been acquired before harvesting about the agricultural products that are actually harvested at an appropriate timing. The remaining integrated value is determined by using a result of a learning process performed for each of the agricultural products belonging to the same growth period based on the actual value of the remaining integrated value up to the timing of Judgment device. An environmental information sensor installed in a plantation where a specific agricultural product is grown, obtaining environmental information indicating the environment in the plantation; and
An imaging device installed in the plantation, capturing an image of the agricultural product;
Based on the environmental information and image information corresponding to the growth time of the agricultural product among image information that is information obtained from the image of the agricultural product, an integrated value of a predetermined environmental value until an appropriate harvest of the agricultural product And determining a remaining integrated value.   Based on environmental information indicating an environment in a cultivation place where a specific agricultural product is cultivated, and image information corresponding to a growth time of the agricultural product among image information that is information obtained from the image of the agricultural product, the agricultural product The determination method which determines the remaining integrated value which is the integrated value of the predetermined | prescribed environmental value until appropriate harvesting of this.   A computer program for causing a computer to function as the determination system according to claim 1.   A computer program for causing a computer to function as the determination apparatus according to claim 2.

Images