Information processing apparatus, control method, and storage medium

Abstract

To promote the automation of paddy rice cultivation.SOLUTION: An information processing device includes first acquisition means for acquiring a soil image obtained by imaging soil of a paddy field, second acquisition means for acquiring an inference result indicating a degree of partial drying of the soil of the paddy field by causing inference means to perform inference based on the soil image, and execution means for executing control processing related to management of the paddy field based on the inference result.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to an information processing device, a control method, and a program, and in particular to a process management technique for rice cultivation. [Background technology]

[0002] In paddy rice cultivation, water management work in paddy fields (fields) that is tailored to the growth of the rice is important. In recent years, in order to reduce the burden of water management work, systems have been introduced that remotely and automatically control the opening and closing of water supply valves and water outlets installed in the fields by installing water level sensors in the fields. One such system determines the growth stage of rice plants based on the plant height obtained from images of the rice plants in the field, and controls the water level according to the growth stage (Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-106583 Summary of the Invention [Problem to be solved by the invention]

[0004] During the tillering stage of rice growth, the field is temporarily drained of water, a process known as "mid-season drainage." Mid-season drainage has the effect of preventing root rot by supplying oxygen to the soil, preventing excessive stem growth by suppressing nitrogen absorption, preventing lodging by hardening the soil, and improving workability for combine harvesters.

[0005] The extent of mid-drying does not depend on the water level in the field, as mentioned above. Depending on the soil type and rice variety, it may be necessary to dry the soil vigorously until it cracks, or gently so that only a small amount of water remains in the cracks. The optimal degree of mid-drying may also vary depending on weather conditions. Traditionally, the decision on the extent of mid-drying required experienced farmers to observe the condition of the field and take various factors into consideration. However, having to monitor large or multiple fields by a specific farmer placed a heavy burden on the farmer and was also a factor in the lack of new farmers interested in rice cultivation.

[0006] The present invention has been made in consideration of the above-mentioned problems, and an object of the present invention is to provide an information processing device, a control method, and a program that promote automation of paddy rice cultivation. [Means for solving the problem]

[0007] In order to achieve the above-mentioned object, the information processing device of the present invention is characterized by having a first acquisition means for acquiring a soil image of the soil of a paddy field, a second acquisition means for acquiring an inference result indicating the degree of inter-paddy field drying by having an inference means make an inference based on the soil image, and an execution means for executing control processing related to the management of the paddy field based on the inference result. [Effects of the Invention]

[0008] With this configuration, the present invention makes it possible to promote automation of rice cultivation. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 illustrates an example of a configuration of a management system according to an embodiment and a modification of the present invention. [Figure 2] FIG. 1 is a block diagram illustrating a hardware configuration of a management system according to an embodiment and a modification of the present invention. [Figure 3] FIG. 1 is a block diagram illustrating a functional configuration of a management system according to an embodiment and a modification of the present invention. [Figure 4] FIG. 1 is a diagram illustrating machine learning performed in an inference server 107 according to an embodiment and a modification of the present invention. [Figure 5] 1 is a flowchart illustrating a water supply and drainage control process executed by the control server 101 according to the first embodiment of the present invention. [Figure 6] 1 is a flowchart illustrating a management process executed by the control server 101 according to the first embodiment of the present invention. [Figure 7] 10 is a flowchart illustrating a water supply and drainage control process executed by the control server 101 according to the second embodiment of the present invention. [Figure 8] 10 is a flowchart illustrating a management process executed by the control server 101 according to the second embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0010] [Embodiment 1] Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the scope of the invention claimed. Although multiple features are described in the embodiments, not all of these multiple features are necessarily essential to the invention, and multiple features may be combined arbitrarily. Furthermore, in the accompanying drawings, the same reference numerals are used to designate the same or similar components, and redundant explanations will be omitted.

[0011] In the embodiment described below, the present invention is applied to a control server, which is an example of an information processing device, that acquires an inference result of the degree of soil drying based on a soil image of a paddy field (field) where rice is grown, and executes various processes related to field management. However, the present invention can be applied to any device that can execute various control processes based on an inference result of the degree of drying of the soil in the field based on the soil image.

[0012] Management system configuration 1 illustrates an example of the system configuration of a management system according to an embodiment of the present invention. In the following description, the management object of the management system is a field (paddy field) where rice is cultivated, and a group of devices provided for the field includes a relay base station 102, an imaging device 103, a measuring device 104, and a water supply and drainage device 105, all connected via a local network 106. These device groups may be provided for multiple fields managed by a single farmer, for example, or may be provided for each field.

[0013] The control server 101 executes various control processes related to the management of the managed farm field. By executing these control processes, the control server 101 transmits and receives information for controlling external devices via the Internet 109 as necessary.

[0014] The relay base station 102 controls various devices installed to change / monitor the state of the farm field. In the example shown in the figure, the various devices include an imaging device 103, a measuring device 104, and a water supply and drainage device 105. The relay base station 102 acquires information obtained by the various devices installed in the farm field and transmits it to the control server 101 and the like via the Internet 109. The relay base station 102 also controls the operation of the relevant devices by transmitting the information received from the control server 101 to the relevant devices.

[0015] The imaging device 103 is a camera installed to capture images of the soil in the field. The imaging device 103 is installed in a position where it can capture images of the soil in the field, and outputs soil images captured periodically or at predetermined timings to show the state of the soil. The soil images output by the imaging device 103 are transmitted to the control server 101 via the relay base station 102. In this embodiment, the imaging device 103 is configured to be able to output RGB images captured using visible light as soil images.

[0016] The measuring device 104 includes sensors for measuring various conditions of the field and outputs the measurement results. In this embodiment, the measuring device 104 includes a water level sensor and is mainly used for water supply and drainage control to control the water level in the field. In addition, the measuring device 104 can include other measuring devices such as a water temperature sensor and an air temperature sensor as needed. The water level information output by the measuring device 104 is transmitted to the control server 101 via the relay base station 102.

[0017] The water supply and drainage device 105 is a device provided to supply and drain water to and from the field. In this embodiment, the water supply and drainage device 105 manages a water supply valve that supplies water to the field and a water outlet that drains water from the field. More specifically, the water supply and drainage device 105 includes actuators for adjusting the opening and closing amounts of the water supply valves and water outlets, and supplies and drains water to and from the field by receiving control signals to control the drive of the corresponding actuators. In other words, in the management system of this embodiment, the water supply and drainage to and from the field is configured to be electronically controllable by control signals supplied from the control server 101.

[0018] The inference server 107 is a device configured to infer the state of soil in a field and output the inference results. As will be described in detail later, in this embodiment, the inference server 107 has a trained model that has been trained on interim drainage performed during the tillering stage, and is able to infer the degree of interim drainage for the soil shown in an input soil image. When the inference server 107 receives a soil image together with an inference request from the control server 101, it returns the inference results obtained for the soil image.

[0019] The client terminal 108 is a device used by users of the management system (for example, farmers, agricultural workers, etc.). The client terminal 108 may include any communication device that can connect to the Internet 109, such as a PC, smartphone, or tablet terminal. The management system of this embodiment is configured to be able to provide the client terminal 108 with a user interface for presenting information about the fields to be managed and for various settings. In one aspect, such a user interface is configured to be usable by accessing a management web page via, for example, a browsing application executed on the client terminal 108.

[0020] <Hardware configuration of various devices> Next, the hardware configuration of the main devices that make up the management system will be explained using FIG.

[0021] <Control Server> First, the hardware configuration of the control server 101 will be described.

[0022] The CPU 201 is a control device that controls the operation of each block of the control server 101. The CPU 201 can control the operation of each block by reading out the operation program of each block stored, for example, in the ROM 202 or the HDD 204, expanding it in the RAM 203, and executing it.

[0023] The ROM 202 is a non-volatile storage device such as a Flash ROM. The ROM 202 stores the operation programs of each block of the control server 101 as well as information such as parameters required for the operation of each block. On the other hand, the RAM 203 is a volatile storage device. The RAM 203 is used not only as an area for loading the operation programs of each block, but also as a storage area for temporarily storing intermediate data generated by the operation of each block.

[0024] The recording device 204 is a so-called large-capacity recording device capable of permanently storing information, such as an HDD or SSD, etc. The recording device 204 records information received via the Internet 109, for example.

[0025] The NIC 205 is a communication interface provided in the control server 101. By using the NIC 205, the control server 101 can send and receive information to and from external devices via the Internet 109.

[0026] The system bus 206 is a general-purpose bus that enables transmission and reception of information (data, signals, etc.) between blocks included in the control server 101 .

[0027] Inference Server Next, the hardware configuration of the inference server 107 will be described.

[0028] The CPU 211 is a control device that controls the operation of each block of the inference server 107. The CPU 211 can control the operation of each block by reading out the operation program of each block stored, for example, in the ROM 212 or the HDD 214, expanding it in the RAM 213, and executing it.

[0029] The ROM 212 is a non-volatile storage device such as a Flash ROM. The ROM 212 stores the operation programs of each block of the inference server 107 as well as information such as parameters required for the operation of each block. On the other hand, the RAM 213 is a volatile storage device. The RAM 213 is used not only as an area for expanding the operation programs of each block, but also as a storage area for temporarily storing intermediate data generated by the operation of each block.

[0030] The recording device 214 is a so-called large-capacity recording device capable of permanently storing information, such as an HDD or SSD. The recording device 214 records, for example, information received via the Internet 109, a trained model (or its parameters) generated by training, and the like.

[0031] The GPU 215 is a computing device configured primarily for executing rendering processing. Due to its characteristics, the GPU 215 can efficiently execute many parallel processes, making it suitable for machine learning that frequently executes repetitive processes, such as deep learning. For this reason, the inference server 107 of this embodiment is configured so that the CPU 211 and the GPU 215 cooperate to execute machine learning for constructing a trained model. The GPU 215 can operate not only during training, but also when using a trained model constructed by training.

[0032] The NIC 216 is a communication interface provided in the inference server 107. By using the NIC 216, the inference server 107 can send and receive information to and from external devices via the Internet 109.

[0033] The system bus 217 is a general-purpose bus that realizes transmission and reception of information (data, signals, etc.) between blocks that the inference server 107 has.

[0034] <Relay base station> Next, the hardware configuration of the relay base station 102 will be described.

[0035] The CPU 221 is a control device that controls the operation of each block of the relay base station 102. The CPU 221 can control the operation of each block by, for example, reading out an operation program for each block stored in the ROM 222, expanding it into the RAM 223, and executing it. The CPU 221 also controls the operation of the imaging device 103, the measuring device 104, and the water supply and drainage device 105 by outputting control signals.

[0036] The ROM 222 is a non-volatile storage device such as a Flash ROM. The ROM 222 stores information such as parameters required for the operation of each block in addition to the operation programs of each block of the relay base station 102. On the other hand, the RAM 223 is a volatile storage device. The RAM 223 is used not only as an area for loading the operation programs of each block, but also as a storage area for temporarily storing intermediate data generated by the operation of each block.

[0037] The NIC 224 is a communication interface provided in the relay base station 102. By using the NIC 224, the relay base station 102 can transmit and receive information to and from external devices via the Internet 109.

[0038] The display unit 225 is a device for displaying information that the relay base station 102 has. The display unit 225 is configured to include, for example, an LED or the like, and can notify the user of information such as the operating status of the relay base station 102. In this embodiment, the display unit 225 is described as being configured to be included in the relay base station 102, but the display unit 225 can also be included in, for example, at least one of the imaging device 103, the measuring device 104, and the water supply and drainage device 105, and can be used to notify the user of their operating status.

[0039] The system bus 226 is a general-purpose bus that enables transmission and reception of information (data, signals, etc.) between blocks that the relay base station 102 has.

[0040] <<Functional configuration of the management system>> Next, the functional configuration of the management system of this embodiment, which is realized by the above hardware configuration and program execution in each device, will be described with reference to FIG.

[0041] The imaging unit 331 of the imaging device 103 captures images of the soil in the field to generate a soil image. The soil image generated by the imaging unit 331 is output to the relay base station 102. The measurement unit 341 of the measuring device 104 measures, for example, the water level in the field and outputs the obtained measurement results (hereinafter referred to as water level information) to the relay base station 102. The relay unit 321 of the relay base station 102 transmits information related to the state of the field (hereinafter sometimes referred to as field information) output from the imaging unit 331 and the measurement unit 341 in this manner to the control server 101 via the Internet 109. The first acquisition unit 301 of the control server 101 acquires the field information transmitted via the relay unit 321.

[0042] As will be described in detail later, when paddy rice cultivation in a field reaches the tillering stage and, for example, the number of rice stalks meets a predetermined condition, the control server 101 controls the water supply and drainage of the water supply and drainage device 105 to start mid-draining in the field. In the management system of this embodiment, as described above, the degree of mid-draining of the field soil is inferred using a trained model. For this purpose, the request unit 302 of the control server 101 transmits the soil image acquired by the first acquisition unit 301 to the inference server 107 together with an inference request. When the inference unit 312 of the inference server 107 receives the soil image together with the inference request, it inputs the soil image into the trained model to obtain an inference result. The inference result is returned to the control server 101 as a response to the inference request and is acquired by the second acquisition unit 303.

[0043] The control unit 304 of the control server 101 controls the water supply and drainage operations performed by the water supply and drainage device 105 (water supply and drainage control). The control unit 304 derives the water level to be maintained in the field based on, for example, water level information and soil images from the acquired field information, and outputs control commands related to water supply and drainage control. When controlling water supply and drainage, the control unit 304 can also refer to other information such as weather information in addition to the field information. Furthermore, as will be described in detail later, when an inference result is acquired, the control unit 304 controls water supply and drainage based on the inference result.

[0044] The output control command is transmitted to the drive control unit 351 of the water supply and drainage system 105 via the relay unit 321 of the relay base station 102. Upon receiving the control command, the drive control unit 351 controls the drive of the actuator required to change the water level in the field. That is, if the control command is to raise the water level (water supply command), the drive control unit 351 controls the drive of the actuator provided for the water supply valve. On the other hand, if the control command is to lower the water level (drain command), the drive control unit 351 controls the drive of the actuator provided for the water outlet.

[0045] The communication unit 305 of the control server 101 is mainly responsible for sending and receiving information to and from the client terminal 108. In the management system of this embodiment, services are provided to users that include a function for monitoring the state of the field, including the state of mid-season drainage, and a function for remotely controlling water supply and drainage. The UI unit 361 of the client terminal 108 provides various user interfaces related to the use of the services. As described above, the functions related to the UI unit 361 are provided via a display device (not shown) of the client terminal 108 while the client terminal 108 is accessing a specific web page. Regarding operation inputs at the client terminal 108 received via the UI unit 361, information indicating the input contents is sent to the control server 101 and acquired by the communication unit 305.

[0046] <<Outline of water supply and drainage control for the mid-drying process>> Below, we will provide an overview of the water supply and drainage control of a field related to the mid-season drainage process, which is carried out in such a management system.

[0047] <Building a trained model> As described above, in the management system of this embodiment, machine learning using soil images is performed in the inference server 107 to enable inference of the degree of interim drainage of soil based on soil images of the soil in the field. During the interim drainage process, as the soil dries and loses moisture, changes occur, such as an increase in the number of cracks on the ground surface and an increase in the width of the cracks. Therefore, by machine learning the characteristics of cracks and the like that appear in the soil image, a trained model can be constructed that can infer the degree of interim drainage of the soil that appears in the image for an input soil image. This machine learning is performed using the learning unit 311 as a functional configuration and the CPU 211 and GPU 215 as a hardware configuration.

[0048] The soil images used for learning are primarily images captured after the start of the interim drainage process. These soil images may be images of soil captured (by the imaging device 103) in a single field where water supply and drainage control is planned using a trained model constructed thereafter, or images of soil captured in one or more fields, not limited to the specific field, may be used. Furthermore, to enable inference of various levels of interim drainage, it is preferable for machine learning to use multiple types of soil images in which the moisture content of the field is intentionally adjusted to vary the level of interim drainage. Furthermore, learning accuracy may vary depending on the lighting conditions when capturing the soil images and how the soil is captured in the field of view. Therefore, soil images captured by multiple imaging devices 103 installed in different locations may be used in machine learning to enable inference of an appropriate level of interim drainage from soil images captured under various lighting conditions and angles of view.

[0049] In the management system of this embodiment, in order to construct a trained model for inferring the intensity of interim drainage of soil shown in a soil image, an evaluation value (hereinafter referred to as an interim drainage index) that evaluates the intensity of interim drainage is provided as training data during machine learning. The intensity of interim drainage of soil shown in a soil image can be, for example, a value determined by a farmer with some experience in rice cultivation after viewing the soil image. The soil image and interim drainage index used in machine learning in one aspect may be set, for example, as follows: In the example table, the interim drying index is defined so that the higher the intensity of interim drying, i.e., the larger the cracks in the soil, the smaller the value. In other words, the interim drying index is defined so that the drier the soil, the smaller the value.

[0050] The learning unit 311 constructs a trained model by machine learning the soil image together with such training data. Specific algorithms that can be used for machine learning include, for example, nearest neighbor methods, naive Bayes methods, decision trees, and support vector machines. Also, a deep learning method that uses a neural network to generate features and connection weighting coefficients for machine learning can be used. The machine learning in the learning unit 311 may be performed using any of these methods.

[0051] FIG. 4(a) shows an example of machine learning according to this embodiment. The technique shown in FIG. 4(a) uses a neural network. In this technique, the learning unit 311 includes an error detection unit and an update unit (not shown). During machine learning, the error detection unit obtains the error between the training data and output data output from the output layer of the neural network in response to input data input to the input layer. The error detection unit may use a loss function to calculate the error between the output data from the neural network and the training data. Based on the error obtained by the error detection unit, the update unit updates the connection weighting coefficients between the nodes of the neural network to reduce the error. The update unit can update the connection weighting coefficients using, for example, backpropagation. The backpropagation is a technique for adjusting the connection weighting coefficients between the nodes of each neural network to reduce the error between the output data and the training data. By repeating this type of machine learning for multiple types of soil images, the learning unit 311 can construct a trained model that can infer the mid-drying index of the soil shown in the input soil image.

[0052] Use of pre-trained models Next, we will explain how the trained model constructed in this way is used to control water supply and drainage in a field. The trained model is assumed to be used after the start of the mid-season drainage process.

[0053] In the management system of this embodiment, the user can specify the target degree of interim drainage for the field. The target degree can be specified via the UI unit 361 in the client terminal 108. More specifically, for interim drainage to be performed in the field, the user can specify an interim drainage index as the target degree (target evaluation value) of the interim drainage. When the user inputs the interim drainage index on the client terminal 108, the information on the interim drainage index is transmitted to the control server 101 and becomes available for water supply and drainage control.

[0054] After the start of the interim drying process, when the request unit 302 of the control server 101 acquires a soil image captured by the imaging device 103, it transmits the soil image to the inference server 107 together with an inference request. Upon receiving the inference request, the inference unit 312 of the inference server 107 inputs the soil image into the trained model, thereby inferring the interim drying index of the soil appearing in the soil image and transmitting the inference request back to the control server 101. In this way, the request unit 302 can cause the inference server 107 to infer the interim drying index for the acquired soil image. The soil images are controlled to be acquired at a predetermined frequency, and the request unit 302 makes an inference request for the soil image in response to the first acquisition unit 301 acquiring the soil image.

[0055] The control unit 304 of the control server 101 executes various control processes related to soil management based on the interim drainage index obtained for the soil image as an inference result. As described above, in the management system of this embodiment, the user specifies the target degree of interim drainage for the field to be managed. Therefore, the control unit 304 switches control processes depending on the relationship between the interim drainage index specified as the target degree (hereinafter referred to as the target index) and the (current) interim drainage index of the field obtained as an inference result.

[0056] Specifically, the control unit 304 controls the water supply and drainage so that the interim drainage continues until the interim drainage index of the field reaches the target index. That is, the control unit 304 sends a control command to control the water supply and drainage device 105 to open the water outlet provided in the field and drain the field. The control unit 304 also performs control processing to end the interim drainage process on the condition that the interim drainage index of the field reaches the target index.

[0057] The control process to end the interim drainage process includes sending a control command (a command to stop drainage) to control the water supply and drainage device 105 to close the water outlet provided in the field and stop draining the field. At this time, the control unit 304 may send a control command (a command to start water supply) to control the water supply and drainage device 105 to open the water supply valve provided in the field and start supplying water to the field. The control process to end the interim drainage process may also include a process to notify the user's client terminal 108 that the interim drainage state corresponding to the specified interim drainage index has been reached. This notification allows the user to know that the field has reached the desired interim drainage state.

[0058] In this way, the management system of this embodiment can control water supply and drainage during the mid-drying process carried out during the tillering period until the mid-drying state desired by the user is achieved, while reducing the effort required for field management.

[0059] <Water supply and drainage control processing> The following describes in detail the water supply and drainage control processing executed by the control server 101 of the management system of this embodiment for the water supply and drainage of the managed field, using the flowchart in Figure 5. The processing corresponding to this flowchart can be realized by the CPU 201 reading out the corresponding processing program stored in, for example, the ROM 202, and then expanding and executing it in the RAM 203. This water supply and drainage control processing will be described as being started, for example, when a setting request for water supply and drainage control of the managed field is received from the client terminal 108. Hereinafter, the managed field will be referred to simply as the "target field."

[0060] In S501, the CPU 201 performs various settings for water supply and drainage control of the target field based on the received setting request. The setting request includes various setting information input via a user interface related to the management system at the client terminal 108, and the CPU 201 performs various settings based on the setting information. The setting information may include information on the time when rice seedlings were transplanted (planted) in the target field and information on various facilities in the target field (relay base station 102, imaging device 103, measuring device 104, water supply and drainage device 105). The setting information may also include information on conditions for starting the mid-drying process (hereinafter referred to as mid-drying start conditions). The mid-drying start conditions may specify, for example, conditions regarding the number of rice stalks and plant height, the number of days elapsed since transplanting, etc.

[0061] In S502, the CPU 201 acquires field information for the target field. The processing of this step may be performed in response to the field information being transmitted at a predetermined timing from the relay base station 102. Alternatively, the processing may be performed by sending a transmission request for the field information from the CPU 201 to the relay base station 102 when the predetermined timing has arrived, and receiving the field information from the relay base station 102 in response to the transmission request.

[0062] In S503, the CPU 201 determines whether the inter-season drainage start conditions for the target field are satisfied based on the acquired field information. In a mode in which the inter-season drainage start conditions are set for the number of rice stalks or plant height, the CPU 201 may make the determination in this step by analyzing images of the field, such as soil images, included in the field information. If the CPU 201 determines that the inter-season drainage start conditions for the target field are satisfied, the CPU 201 proceeds to S505; if it determines that the inter-season drainage start conditions are not satisfied, the CPU 201 proceeds to S504.

[0063] In S504, the CPU 201 sends a control signal to control water supply and drainage according to each period before the start of interim drainage, and returns the process to S502. The water supply and drainage control performed in this step is performed, for example, by referring to water level information included in the acquired field information. For example, during the rooting period, which is about two weeks after planting, a control signal is sent to the water supply and drainage device 105 to supply water in the morning and stop water supply during the daytime, so that the water level is kept shallow and the water temperature rises. Furthermore, during the early tillering period after the rooting period (the period until the interim drainage start conditions are met), a control signal is sent to the water supply and drainage device 105 to perform water supply and drainage control related to "intermittent irrigation." Here, intermittent irrigation basically refers to water management in which water is stored in the field for three days and then water is drained for two days, repeating the cycle of supplying and draining.

[0064] On the other hand, if it is determined in S503 that the interim drainage start condition is satisfied, the CPU 201 executes management processing related to farm field management in the interim drainage process in S505.

[0065] <Administrative processing> Here, the management process performed in this step will be described in detail with reference to the flowchart of FIG.

[0066] In S601, the CPU 201 acquires a target index for the interim drainage to be performed on the target field and stores it in the RAM 203. The management system of this embodiment is configured to accept a target index designation from a user when the interim drainage start condition is met. For example, when acquiring the target index, the CPU 201 may be configured to notify the user's client terminal 108 that interim drainage will begin and accept user input regarding the designation of the target index in response to the notification. When the processing of this step is executed, the user can designate a target value for the interim drainage index for the interim drainage process to be performed this year, taking into consideration, for example, the past harvest conditions in the target field, weather trends for the current year, the soil type of the target field such as drainage trends, the variety of crop to be harvested, etc.

[0067] In S602, the CPU 201 sends a control signal to start draining the target field. This control signal controls the drive of various actuators in the water supply and drainage device 105 to close the water supply valve and open the water outlet.

[0068] In S603, the CPU 201 acquires field information for the target field. The processing in this step can be performed in the same manner as S502 in the water supply and drainage control processing.

[0069] In S604, the CPU 201 acquires the interim drainage index of the target field. Specifically, the CPU 201 transmits the soil image included in the field information acquired in S603 together with an inference request to the inference server 107, and acquires the interim drainage index of the soil shown in the soil image as the inference result from the inference server 107.

[0070] In S605, the CPU 201 determines whether the interim drainage index acquired in S604 has reached the target index. In other words, the CPU 201 determines whether the soil of the target field has been drained to the extent specified by the user, based on the interim drainage index inferred from the soil image. If the CPU 201 determines that the interim drainage index has reached the target index, it proceeds to S606; if it determines that the index has not reached the target index, it returns to S603.

[0071] In S606, the CPU 201 executes a control process to end the interim drying process, thereby completing this management process. As described above, the control process to end the interim drying process may include a process to send a control command to stop drainage and a control command to start (restart) water supply to the water supply and drainage device 105. The control process may also include a process to send an interim drying end notification to the user's client terminal 108, indicating that the interim drying has progressed to the desired state.

[0072] When the management process is completed in this manner, the CPU 201 shifts the water supply and drainage control process to S506.

[0073] In S506, the CPU 201 acquires field information and sends control signals to control water supply and drainage according to the respective periods after the end of interim drainage. The water supply and drainage control performed in this step is performed with reference to water level information contained in the acquired field information and, as necessary, water temperature and air temperature information, just as it was before the start of interim drainage. For example, in the late tillering stage (the period after the end of the interim drainage process), a control signal is sent to the water supply and drainage device 105 to perform water supply and drainage control such that intermittent irrigation is performed while preventing the water temperature from rising too high at night. Although detailed explanations will be omitted, control signals are also sent to the water supply and drainage device 105 to perform predetermined water supply and drainage control based on the field information for the panicle neck differentiation period, booting period, heading period, flowering period, ripening period, and drainage period after the tillering stage.

[0074] As described above, the information processing device of this embodiment can drain the water until a predetermined mid-drying state is reached and then move on to the post-drying process, without the need for an experienced farmer to frequently visit the field to check the mid-drying state. As a result, automation of paddy rice cultivation can be promoted.

[0075] In this embodiment, the target index is specified in step S601 of the management process for the interim drying step, but the present invention is not limited to this. The target index may be specified in the setting information received together with the setting request in step S501 of the water supply and drainage control process, for example.

[0076] In this embodiment, when the interim draining index reaches the target index, a notification is given that the interim draining index specified by the interim draining completion notification has been reached. However, the notification of the interim draining index for a field is not limited to this. For example, the CPU 201 may notify information on whether the specified interim draining index has been reached each time an inference result is obtained by the inference server 107. Alternatively, the CPU 201 may notify the client terminal 108 of the interim draining index, which is the inference result, in a timely manner each time an inference result is obtained.

[0077] In addition, although the present embodiment has described an aspect in which the mid-drying index is introduced as an evaluation value for evaluating the degree of mid-drying, it goes without saying that the implementation of the present invention is not limited to this. The evaluation value for evaluating the degree of mid-drying can be any other index that can be provided as training data during machine learning of the trained model.

[0078] [Variation 1] In the above embodiment, the target index is specified by the user, but the present invention is not limited to this. The target index may be set without user input, for example, by obtaining a previously set mid-season drainage index for the target field or a mid-season drainage index specified for a field with a similar soil type.

[0079] [Embodiment 2] In the above-described embodiment, a target index is set, and the end of mid-drying is determined based on whether the mid-drying index of the soil obtained based on the soil image has reached the target index. However, in order to set an appropriate target index, the user must have knowledge, such as the extent to which mid-drying is required to achieve a favorable harvest. In this embodiment, a trained model is used to infer the appropriate degree of mid-drying for the soil shown in the input soil image, so that even users without such knowledge can enjoy favorable water supply and drainage control effects.

[0080] <Outline of water supply and drainage control for mid-season drainage> Below, an overview will be given of the water supply and drainage control of the field related to the mid-season drainage process, which is executed in the management system of this embodiment.

[0081] <Building a trained model> In the management system of this embodiment, in order to be able to infer whether mid-season drying is appropriate for the soil shown in the soil image, a trained model is constructed using not only the soil image as in embodiment 1, but also information about the soil for machine learning.

[0082] As shown in Figure 4(b), the inference server 107 of this embodiment uses as learning objects not only soil images but also information on the soil type (soil type, drainage characteristics, etc.) of the soil shown in the soil images and the rice variety grown in the soil. For example, a farmer with experience in rice cultivation in fields with similar conditions (soil type, variety) is asked to evaluate the appropriateness of inter-draining for each of the soil images captured in the same field with intentionally varying degrees of inter-draining, and the evaluations are used as training data. In one aspect, the appropriateness of inter-draining can be expressed, for example, as a percentage of the progress of inter-draining, with 100 representing an appropriate inter-draining state (the state where inter-draining should be completed).

[0083] That is, the inference server 107 performs machine learning using suitability on multiple types of soil images taken at different times during the inter-season drainage process in fields corresponding to each combination of multiple soil types and cultivated rice varieties, not limited to a specific field. In this way, the inference server 107 can build a trained model that can infer not only the characteristics of the soil, such as the state of cracks, that appear in the soil images, but also the appropriate state / time for ending inter-season drainage that is suited to the soil type of the soil and the cultivated rice variety.

[0084] In one embodiment, the input data (soil image, soil type, and variety of rice being cultivated (hereinafter sometimes simply referred to as variety)) used for machine learning and the appropriateness of mid-season drying may be set, for example, as follows. TIFF2026013282000003.tif83154 In this example table, unlike in embodiment 1, different degrees of suitability for mid-season drainage are provided as training data for combinations of soil type and variety, even for soil images with different crack conditions on the ground surface. The learning unit 311 performs machine learning on the input data together with such training data to construct a trained model.

[0085] Use of pre-trained models Next, we will explain how the trained model constructed in this way is used to control water supply and drainage in a field. The trained model is assumed to be used after the start of the mid-season drainage process.

[0086] In the management system of this embodiment, unlike in embodiment 1, the user does not need to specify the target extent of mid-drainage for the field, as long as they input information on the soil type and variety of the field. In other words, in the management system of this embodiment, after the mid-drainage process has begun, the trained model determines whether the mid-drainage state has been reached appropriately based on soil images taken of the field and information on the soil type and variety input in advance, and can then control the water supply and drainage of the field.

[0087] After the start of the interim draining process, the request unit 302 of the control server 101 acquires a soil image captured by the imaging device 103 and transmits the soil image and information on the soil type and variety to the inference server 107 along with an inference request. Upon receiving the inference request, the inference unit 312 of the inference server 107 inputs this information into the trained model, thereby inferring the suitability of interim draining for the soil shown in the soil image and transmitting the inference request back to the control server 101. In this way, the request unit 302 can cause the inference server 107 to infer the suitability of interim draining from the acquired soil image and other information. Soil images are controlled to be acquired at a predetermined frequency, and the request unit 302 makes an inference request for the soil image in response to the first acquisition unit 301 acquiring the soil image.

[0088] The control unit 304 of the control server 101 executes various control processes related to soil management based on the suitability of interim drainage obtained from the soil image as an inference result. Since interim drainage progresses over time, depending on the frequency of capturing soil images, the determination of the end of the interim drainage process based on the suitability of interim drainage does not necessarily require that the suitability of interim drainage reach 100. Therefore, the control unit 304 determines that interim drainage of the field has become appropriate when the (current) suitability of interim drainage of the field obtained as an inference result exceeds a threshold value, such as 90.

[0089] Therefore, the control unit 304 controls the water supply and drainage so as to continue the interim drainage until the suitability of the interim drainage of the field exceeds the threshold. That is, the control unit 304 sends a control command to control the water supply and drainage device 105 to open the water outlet provided in the field and drain the field. The control unit 304 also performs control processing to end the interim drainage process, provided that the suitability of the interim drainage of the field exceeds the threshold. In addition to sending the control command related to the water supply and drainage described above, the control processing to end the interim drainage process may include processing to notify the user's client terminal 108 that the interim drainage has become appropriate.

[0090] In this way, the management system of this embodiment can control water supply and drainage during the mid-drying process carried out during the tillering period until the mid-drying condition is appropriate for the soil type and variety, while reducing the effort required for field management.

[0091] <Water supply and drainage control processing> The following describes in detail the water supply and drainage control processing executed by the control server 101 of the management system of this embodiment for the water supply and drainage of the managed field, using the flowchart in Figure 7. The processing corresponding to this flowchart can be realized by the CPU 201 reading out the corresponding processing program stored in, for example, the ROM 202, and loading and executing it in the RAM 203. This water supply and drainage control processing will be described as being started, for example, when a setting request for water supply and drainage control of the managed field is received from the client terminal 108. Hereinafter, the managed field will be referred to simply as the "target field." Note that in the description of the water supply and drainage control processing in Figure 7, steps that perform processing similar to the water supply and drainage control processing of the first embodiment will be given the same reference numerals and will not be described again.

[0092] In step S701, the CPU 201 performs various settings for water supply and drainage control of the target field based on the received setting request. The setting request includes various setting information input to the client terminal 108 via a user interface related to the management system, and the CPU 201 performs various settings based on the setting information. The setting information may include information on the time when rice seedlings were transplanted (planted) in the target field and information on various facilities in the target field (the relay base station 102, the imaging device 103, the measuring device 104, and the water supply and drainage device 105). The setting information may also include information on conditions for starting the mid-drying process (hereinafter referred to as mid-drying start conditions). The mid-drying start conditions may specify, for example, conditions regarding the number of rice stalks and plant height, or the number of days elapsed since transplanting. In addition, in the management system of this embodiment, the setting information includes information on the soil type of the target field and information on the variety of rice cultivated in the target field. Hereinafter, the soil type and variety information will be referred to as inference information.

[0093] If it is determined in S503 that the interim draining start conditions are met, the CPU 201 executes management processing related to farm field management in the interim draining process in S702.

[0094] <Administrative processing> Here, the management processing performed in this step will be described in detail with reference to the flowchart in Fig. 8. In the description of the management processing in Fig. 8, steps that perform processing similar to the management processing in the first embodiment will be given the same reference numerals and will not be described again.

[0095] After acquiring the field information for the target field in S603, the CPU 201 acquires the suitability of interim drainage for the target field in S801. Specifically, the CPU 201 transmits the inference information and the soil image included in the field information acquired in S603 together with an inference request to the inference server 107, and acquires the suitability of interim drainage for the soil shown in the soil image from the inference server 107 as an inference result.

[0096] In S802, the CPU 201 determines whether the appropriateness of interim drainage obtained in S801 exceeds a threshold. In other words, the CPU 201 determines whether the soil of the target field has been drained to an appropriate state based on the appropriateness of interim drainage inferred from the soil image and the inference information. If the CPU 201 determines that the appropriateness of interim drainage exceeds the threshold, it proceeds to S803, and if it determines that it does not exceed the threshold, it returns to S603.

[0097] In S803, the CPU 201 executes a control process to end the interim draining process and completes this management process. As described above, the control process to end the interim draining process may include a process to send a control command to stop drainage and a control command to start (restart) water supply to the water supply and drainage device 105. The control process may also include a process to send an interim draining completion notification to the user's client terminal 108, indicating that the interim draining has progressed to an appropriate state for the field.

[0098] As described above, the information processing device of this embodiment can drain the water until the appropriate drainage state is achieved and then move on to the post-drying process, without the need for an experienced farmer to frequently visit the field to check the drainage state, thereby facilitating the automation of paddy rice cultivation.

[0099] [Variation 2] In the above-described embodiment and modified example, after the control process for terminating the interim drying process is executed, water supply and drainage control is performed according to the time period after the interim drying process is completed, but the implementation of the present invention is not limited to this. For example, even if the interim drying process is completed, if the degree of interim drying becomes weaker due to rainfall or if the degree of interim drying becomes too strong due to prolonged high temperatures, the water supply and drainage control for the interim drying process may be performed again.

[0100] [Variation 3] In the second embodiment described above, the soil images to be machine-learned are described as being trained using the appropriateness of mid-drying evaluated by farmers and others as training data. However, the present invention is not limited to this. That is, the appropriateness of mid-drying and the mid-drying index described above do not need to be based on the subjective evaluation of a specific farmer, but may be derived based on objective information such as the harvest results of the field. Such information may include information about the harvest history of the field, such as the past harvest volume and quality of the harvested product. In other words, by performing machine learning on the soil image (related to the mid-drying process) to be learned using the evaluation value of the soil image derived based on the harvest results of the same field in the year the soil image was captured, a trained model that infers more practical evaluation values ​​can be constructed.

[0101] [Variation 4] Factors related to harvest in a field may include, in addition to soil type and variety, the growing conditions of the crop, meteorological information, the planned harvest time in the field, the harvest conditions of the previous year, etc. Therefore, such information may be included in the data input for machine learning of soil images. That is, such information may include at least one of the following for the field shown in the soil image: soil type, the variety of the crop being grown, the growing conditions of the crop, meteorological information such as the year the image was taken, information about the planned harvest of the crop, and information about the harvest history of the field. Furthermore, in an aspect in which such data is added to the soil image and machine learning is performed, the relevant data may be provided as input data along with the soil image when inferring from the constructed trained model.

[0102] [Variation 5] In the above-described embodiment, after the start of the interim drainage process, the soil image acquired as field information is sent to the inference server 107 along with an inference request to infer the degree of interim drainage of the field. However, the present invention is not limited to this. When the interim drainage process begins and drainage of the field begins, the field is still waterlogged. Although the extent of interim drainage varies, it is basically a task of draining the field to the point where cracks appear on the ground surface. Therefore, there is little need for the inference server 107 to perform inference on soil images at the beginning of the interim drainage process. Therefore, the CPU 201 may, for example, refer to water level information included in the field information and control the inference server 107 to perform inference on the soil image when the water level in the field falls below a predetermined value. This reduces the computational load related to inference on the inference server 107.

[0103] [Variation 6] In the above-described embodiment and modified examples, the imaging device 103 captures visible light and outputs an RGB image as a soil image, but the implementation of the present invention is not limited to this. In recent years, in the field of agriculture, there has been an increasing number of cases where information other than visible light is effectively utilized, and the imaging device 103 that captures soil images used during learning and inference may be, for example, an infrared camera that captures near-infrared light or thermal infrared light. Alternatively, the imaging device 103 may be a hyperspectral camera that captures images by dispersing light into wavelengths.

[0104] Furthermore, the imaging device 103 does not need to be fixedly installed in the field, but may be attached to a remotely operated autonomous device such as a drone or a land-based machine.

[0105] [Variation 7] In the above-described embodiment and modified example, the control server 101 controls the operation of the water supply and drainage equipment 105 installed in the field by sending control commands via the Internet 109, thereby controlling the water supply and drainage of the field. However, the implementation of the present invention is not limited to this, and it goes without saying that the functions of the control server 101 may be incorporated into the relay base station 102 or the water supply and drainage equipment 105. Also, while it has been described that the relay base station 102 acquires information from the imaging device 103, the measuring device 104, and the water supply and drainage equipment 105 and transfers the information and control commands to the control server 101, it goes without saying that the relay base station 102 is not an essential component for implementing the present invention. The control server 101 may be configured to be able to communicate information with each of the imaging device 103, the measuring device 104, and the water supply and drainage equipment 105 without going through the relay base station 102.

[0106] In addition, although the present embodiment has been described in terms of an embodiment in which the control server 101 and the inference server 107 are configured as separate devices, the present invention is not limited to this. The functions of the inference server 107 may be incorporated into the control server 101. In other words, the management system shown in Fig. 1 is merely an example of a system configuration for implementing the present invention, and the various functional configurations can be integrated / distributed into any number of devices, not limited to the embodiment shown in Fig. 3.

[0107] [Other embodiments] The present invention can also be realized by supplying a program that realizes one or more functions of the above-described embodiments to a system or device via a network or a storage medium, and having one or more processors in the computer of the system or device read and execute the program.The present invention can also be realized by a circuit (e.g., ASIC) that realizes one or more functions.

[0108] [Summary of the embodiment and modifications] The disclosure of this specification includes the following information processing device, control method, and program. (Item 1) a first acquisition means for acquiring a soil image obtained by capturing an image of soil in a paddy field; a second acquisition means for acquiring an inference result indicating the degree of mid-drying of the soil in the paddy field by causing an inference means to make an inference based on the soil image; an execution means for executing a control process related to management of the paddy field based on the inference result; An information processing device comprising: (Item 2) Further, a setting means for setting a target degree of mid-drying of the soil of the paddy field is provided, The control process includes a process of determining whether the degree of mid-drying indicated by the inference result has reached the target degree. 2. The information processing device according to item 1, (Item 3) 3. The information processing device according to item 2, wherein the setting means sets the target level based on a user input. (Item 4) The degree of interim drying indicated by the inference result includes an evaluation value that evaluates the intensity of the interim drying, The setting means sets a target evaluation value as the target degree. 4. The information processing device according to item 2 or 3. (Item 5) The information processing device described in item 4 is characterized in that the inference means is a trained model constructed by learning images of the soil of one or more rice paddies using evaluation values ​​that evaluate the intensity of inter-drainage of the soil shown in the images as training data, and is a trained model that infers the intensity of inter-drainage of the soil shown in the input soil image. (Item 6) The information processing device described in any one of items 2 to 5, characterized in that the control processing includes a process of notifying that the degree of mid-drying indicated by the inference result has reached the target degree, on the condition that the degree has reached the target degree. (Item 7) The information processing device described in any one of items 2 to 6, characterized in that the control processing includes processing related to stopping drainage of the rice paddy, on the condition that the degree of mid-drying indicated by the inference result has reached the target level. (Item 8) The information processing device described in any one of items 2 to 7, characterized in that the control processing includes processing related to starting water supply to the rice paddy, on the condition that the degree of mid-drying indicated by the inference result has reached the target level. (Item 9) the inference means infers whether mid-season drainage of the soil in the paddy field is appropriate or not based on the soil image; The control process includes a process of determining whether or not the inference result indicating that mid-season drainage of the soil of the paddy field is appropriate has been obtained. 2. The information processing device according to item 1, (Item 10) The information processing device described in item 9 is characterized in that the inference means is a trained model constructed by learning images of the soil of one or more rice paddies using information on whether or not the soil shown in the images is appropriate for inter-drainage as training data, and is a trained model that infers whether or not the soil shown in the input soil image is appropriate for inter-drainage. (Item 11) The information processing device described in item 9 or 10, characterized in that the control process includes a process of notifying that inter-drying of the rice paddy soil has become appropriate, on the condition that the inference result indicating that inter-drying of the rice paddy soil has become appropriate has been obtained. (Item 12) The information processing device described in any one of items 9 to 11, characterized in that the control processing includes processing related to stopping drainage of the rice paddy, on the condition that the inference result indicating that the mid-drying of the soil of the rice paddy has become appropriate is obtained. (Item 13) The information processing device described in any one of items 9 to 12, characterized in that the control processing includes processing related to starting water supply to the rice paddy, on the condition that the inference result indicating that the mid-drying of the soil in the rice paddy has become appropriate has been obtained. (Item 14) The information processing device described in any one of items 1 to 13, characterized in that the inference means outputs the inference result based on at least one of the soil type of the paddy field, the variety of crop grown in the paddy field, the growing condition of the crop, weather information, information regarding the planned harvest in the paddy field, and information regarding the harvest history in the paddy field, and the soil image. (Item 15) The information processing device further includes a third acquisition means for acquiring information on the water level of the paddy field, The second acquisition means causes the inference means to perform inference based on the soil image on the condition that the water level of the paddy field is below a predetermined value. 15. The information processing device according to any one of items 1 to 14. (Item 16) a first acquisition step of acquiring a soil image of a paddy field; a second acquisition step of acquiring an inference result indicating the degree of mid-drying of the soil in the paddy field by causing an inference means to make an inference based on the soil image; an execution step of executing a control process related to management of the paddy field based on the inference result; 1. A method for controlling an information processing device, comprising: (Item 17) A program that causes a computer to function as each of the means of the information processing device described in any one of items 1 to 15.

[0109] The invention is not limited to the above-described embodiments, and various changes and modifications can be made without departing from the spirit and scope of the invention. Accordingly, the following claims are appended to apprise the public of the scope of the invention. [Explanation of symbols]

[0110] 101: Control server, 103: Imaging device, 105: Water supply and drainage device, 107: Inference server, 201: CPU, 202: ROM, 203: RAM, 205: NIC, 301: First acquisition unit, 302: Request unit, 303: Second acquisition unit, 304: Control unit, 311: Learning unit, 312: Inference unit, 331: Imaging unit, 351: Drive control unit

Claims

1. a first acquisition means for acquiring a soil image obtained by capturing an image of soil in a paddy field; a second acquisition means for acquiring an inference result indicating the degree of mid-drying of the soil in the paddy field by causing an inference means to make an inference based on the soil image; an execution means for executing a control process related to management of the paddy field based on the inference result; An information processing device comprising:

2. Further, a setting means for setting a target degree of mid-drying of the soil of the paddy field is provided, The control process includes a process of determining whether the degree of mid-drying indicated by the inference result has reached the target degree.

2. The information processing apparatus according to claim 1, wherein:

3. 3. The information processing apparatus according to claim 2, wherein the setting means sets the target level based on a user input.

4. The degree of interim drying indicated by the inference result includes an evaluation value that evaluates the intensity of the interim drying, The setting means sets a target evaluation value as the target degree.

3. The information processing apparatus according to claim 2, wherein:

5. The information processing device described in claim 4, characterized in that the inference means is a trained model constructed by learning images of the soil of one or more paddy fields using evaluation values ​​that evaluate the intensity of inter-drainage of the soil shown in the images as training data, and is a trained model that infers the intensity of inter-drainage of the soil shown in the input soil image.

6. The information processing device according to claim 2, characterized in that the control processing includes a processing for notifying that the degree of mid-drying indicated by the inference result has reached the target degree, on the condition that the degree of mid-drying indicated by the inference result has reached the target degree.

7. The information processing device according to claim 2, characterized in that the control processing includes processing related to stopping drainage of the paddy field on the condition that the degree of mid-drying indicated by the inference result has reached the target degree.

8. The information processing device according to claim 2, characterized in that the control processing includes processing related to starting water supply to the rice paddy, on the condition that the degree of mid-drying indicated by the inference result has reached the target level.

9. the inference means infers whether mid-season drainage of the soil in the paddy field is appropriate or not based on the soil image; The control process includes a process of determining whether or not the inference result indicating that mid-season drainage of the soil of the paddy field is appropriate has been obtained.

2. The information processing apparatus according to claim 1, wherein:

10. The information processing device described in claim 9, characterized in that the inference means is a trained model constructed by learning images of the soil of one or more rice paddies using information on whether or not the soil shown in the images is appropriate for inter-drying as training data, and is a trained model that infers whether or not the soil shown in the soil image is appropriate for inter-drying based on the input soil image.

11. The information processing device described in claim 9, characterized in that the control processing includes a processing for notifying that inter-drying of the rice paddy soil has become appropriate, on the condition that the inference result indicating that inter-drying of the rice paddy soil has become appropriate has been obtained.

12. The information processing device according to claim 9, characterized in that the control processing includes processing related to stopping drainage of the rice paddy, on the condition that the inference result indicating that the mid-drying of the soil of the rice paddy has become appropriate is obtained.

13. The information processing device according to claim 9, characterized in that the control processing includes processing related to starting water supply to the rice paddy, on the condition that the inference result indicating that the soil of the rice paddy has been properly dried out is obtained.

14. The information processing device according to claim 1, characterized in that the inference means outputs the inference result based on the soil image and at least one of the soil type of the paddy field, the variety of crop grown in the paddy field, the growing status of the crop, weather information, information regarding the planned harvest in the paddy field, and information regarding the harvest history in the paddy field.

15. The information processing device further includes a third acquisition means for acquiring information on the water level of the paddy field, The second acquisition means causes the inference means to perform inference based on the soil image on the condition that the water level of the paddy field is below a predetermined value.

2. The information processing apparatus according to claim 1, wherein:

16. a first acquisition step of acquiring a soil image of the paddy field soil; a second acquisition step of acquiring an inference result indicating the degree of mid-drying of the soil in the paddy field by causing an inference means to make an inference based on the soil image; an execution step of executing a control process related to management of the paddy field based on the inference result; 1. A method for controlling an information processing device, comprising:

17. A program that causes a computer to function as each of the means of the information processing apparatus according to any one of claims 1 to 15.

Images