Program, information processing device, method, and system

Abstract

To determine with high accuracy whether a plot in a target area is a paddy field or not. [Solution] A program to be executed by a computer having a processor and a memory. The program causes the processor to execute the following steps: acquiring multiple first images of a target area taken from the air during a first period, analyzing the time-series fluctuation pattern of vegetation indices obtained from the multiple first images to generate identification information indicating the results of identifying paddy fields, acquiring second images of the target area taken from the air during a second period, inputting the second images into a trained AI model and causing the AI ​​model to generate first polygons indicating plots, and extracting, from the multiple first polygons, those whose correspondence relationship between the first polygons and the paddy field areas indicated by the identification information satisfies a predetermined standard as second polygons indicating paddy fields.

Description

[Technical Field]

[0001] The present disclosure relates to a program, an information processing device, a method, and a system. [Background technology]

[0002] In recent years, addressing issues such as global food security, environmental monitoring, and improving agricultural productivity has become an urgent task. In particular, while paddy fields play an important role in the world's food supply, accurate understanding of their distribution and cultivation status is essential for food production planning, water resource management, and climate change countermeasures such as estimating methane emissions. Traditionally, this information has relied on field surveys or statistical data, but there are limitations to collecting information over a wide area and at high frequency.

[0003] Against this background, remote sensing technology using satellites, aircraft, drones, etc. has been attracting attention. Remote sensing technology has been used to efficiently acquire information on the earth's surface over a vast area and extract various agricultural-related information. For example, Patent Document 1 discloses a technology that classifies the uses of candidate farmland plots extracted from observation images taken from the air by a high-altitude flying object, and then calculates the amount of noise, vegetation index, and shape of each area of ​​the candidate farmland plot to determine whether the candidate farmland plot is farmland. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2008-152425 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the technology disclosed in Patent Document 1 leaves room for improvement in the accuracy of determining whether a candidate farmland plot is farmland or not.

[0006] An object of the present disclosure is to accurately determine whether a plot in a target area is a paddy field or not. [Means for solving the problem]

[0007] To solve the above-mentioned problems, one embodiment of the present disclosure provides a program for causing a computer including a processor and a memory to execute the program. The program causes the processor to execute the following steps: acquiring a plurality of first images of a target area taken from the air during a first period set to cover at least one cropping cycle of paddy fields in the target area; analyzing a time-series variation pattern of vegetation indices obtained from the plurality of first images to generate identification information indicating the results of identifying the paddy fields; acquiring one or more second images of the target area taken from the air during a second period set to include a period during which boundaries of compartments in the target area are identifiable; inputting the one or more second images into a trained AI model and causing the AI ​​model to generate first polygons representing each of the plurality of compartments; and extracting, from the plurality of first polygons, those whose correspondence relationship between the first polygon and the paddy field area indicated by the identification information satisfies a predetermined criterion as second polygons representing the paddy fields, based on the identification information. [Effects of the Invention]

[0008] According to the present disclosure, it is possible to determine with high accuracy whether a plot in a target area is a paddy field or not. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a block diagram showing an example of the overall configuration of a system 1. FIG. [Figure 2] 2 is a block diagram showing an example of the configuration of a terminal device 10 shown in FIG. 1. FIG. [Figure 3] 2 is a block diagram showing an example of the configuration of a server 20 shown in FIG. 1. FIG. [Figure 4] FIG. 4 is a diagram showing the data structure of a polygon database 2021 shown in FIG. 3. [Figure 5]10 is a flowchart showing an example of the operation of the server 20 when extracting paddy field polygons. [Figure 6] 14 is an example screen showing an example of map data 1411 on which a paddy field polygon 1412 is displayed. [Figure 7] FIG. 2 is a block diagram showing the basic hardware configuration of a computer 90. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In all drawings describing the embodiments, common components are designated by the same reference numerals, and repeated explanations will be omitted. Note that the following embodiments do not unduly limit the content of the present disclosure described in the claims. Furthermore, not all components shown in the embodiments are necessarily essential components of the present disclosure. Furthermore, each drawing is a schematic diagram and is not necessarily a precise illustration.

[0011] [1. Overview] The system according to this embodiment acquires multiple first images (for time-series data) of a target area photographed from the air during a specified first period, and analyzes this time-series data to generate identification information indicating the results of paddy field identification. The system according to this embodiment acquires one or more high-resolution second images (for polygon generation) of the target area photographed from the air during a specified second period, and inputs these second images into a trained AI model to generate compartment polygons (first polygons) indicating each of the multiple compartments present in the target area. The system according to this embodiment extracts paddy field polygons (second polygons) indicating paddy fields by comparing the identification information with the compartment polygons.

[0012] [2 System Configuration] <2-1 Overall system configuration> FIG. 1 is a block diagram showing an example of the overall configuration of system 1. System 1 is a system that provides a service (hereinafter, polygon generation service) for determining whether a target area includes a rice paddy and generating a polygon representing the rice paddy (hereinafter, rice paddy polygon). The target area is, for example, land that is the target of a rice paddy project, which is a type of carbon credit project. The rice paddy polygon is an example of a second polygon according to one embodiment of the present disclosure.

[0013] By providing polygon generation services, System 1 supports the use of data for a wide range of applications, including precision agriculture in the agricultural sector, water resource management, food production forecasting, and environmental monitoring (e.g., determining paddy field area to calculate methane emissions).

[0014] 1 includes, for example, a terminal device 10, a server 20, a satellite 30, and an AI system 40. The terminal device 10, the server 20, and the AI ​​system 40 are communicatively connected, for example, via a network 80. The satellite 30 transmits, for example, various types of satellite data to a ground station (not shown). The ground station is communicatively connected to the network 80, and, for example, receives a transmission request from the server 20 and transmits the satellite data to the server 20 via the network 80.

[0015] While FIG. 1 illustrates an example in which the system 1 includes one terminal device 10, the system 1 may include two or more terminal devices 10, for example. When multiple terminal devices use a service simultaneously, the server 20 can appropriately manage requests from each terminal device and process them in parallel. While FIG. 1 illustrates an example in which the system 1 includes one server 20, for example, a collection of multiple devices may be used as a single server 20. The manner in which the multiple functions required to realize the server 20 are allocated to one or more pieces of hardware can be determined appropriately depending on the processing capacity of each piece of hardware and / or the specifications required for the server 20. For example, the execution of an AI model with a high data processing load can be offloaded to a dedicated GPU server.

[0016] The artificial satellite 30 acquires satellite images, for example, images of a target area taken from the air, and transmits them to a ground station. In this embodiment, the artificial satellite 30 acquires mainly two types of satellite images. One is a first image with low resolution (e.g., 10 m / pixel) and high frequency (e.g., once every few days to once a week) for capturing time-series change patterns in rice paddies, such as satellite images from Sentinel-2 satellites. The other is a second image with high resolution (e.g., 30 cm / pixel) and short-term or low frequency for capturing detailed shapes of the field, such as satellite images from commercial satellites like the WorldView series or GeoEye-1. The first and second images are not limited to satellite images and may be, for example, images taken from the air by an aircraft or drone. In this case, the system 1 includes an aircraft or drone instead of the artificial satellite 30. Aerial images taken by a drone can be captured on demand at higher resolution, making them suitable for use in specific small areas or in urgent situations.

[0017] Here, "low resolution" and "high resolution" in this specification do not simply refer to absolute high and low resolution, but are defined according to the role of each image in System 1. Specifically, "low resolution" refers to a resolution that is insufficient to identify the detailed boundaries of individual fields, but is suitable for time-series analysis that captures seasonal changes in vegetation indices specific to paddy fields over a wide area. On the other hand, "high resolution" refers to a resolution that is suitable for the AI ​​model described below to identify the boundaries of individual plots, including fields, with high accuracy and generate polygons.

[0018] Therefore, the relationship between the resolution of the first image and the resolution of the second image is such that the second image has a higher resolution than the first image. For example, the resolution of the first image is preferably in the range of several meters per pixel to several tens of meters per pixel, with a specific example being 10 meters per pixel, as in the Sentinel-2 satellite image. In contrast, the resolution of the second image is preferably 1 meter per pixel or less, which allows field boundaries to be clearly identified, and more preferably 50 cm per pixel or less, with a specific example being 30 cm per pixel, as in the WorldView series.

[0019] If the target area shown on the satellite image is covered by clouds, the image does not contain information about the ground surface and is therefore unsuitable for extracting paddy field polygons. In such cases, System 1 substitutes a low-resolution image (described below as the third image) or performs processing to supplement the data for that period. This ensures data continuity and maintains the accuracy of the extracted paddy field polygons.

[0020] 1 shows an example in which the system 1 includes one satellite 30, but the system 1 may include, for example, two or more satellites 30. When two or more satellites are included, the types of the satellites 30 may be the same or different. Combining different types of satellites enables multifaceted data acquisition, improving the robustness or analytical capabilities of the system.

[0021] The AI ​​system 40 is a system that includes a trained AI model. The AI ​​model implements algorithms specialized for tasks such as image recognition, semantic segmentation, and object detection.

[0022] Examples of types of AI models include convolutional neural networks (CNNs), recurrent neural networks (RNNs), transformers, and deep learning models that are a combination of these.

[0023] The AI ​​model may also be a generative AI model. The generative AI model may be a single-modal generative AI model that can output polygon data, or a multi-modal generative AI model that can process multiple different types of data (images, text, etc.) in an integrated manner. The generative AI model may be built on a large language model (LLM). An LLM is a model pre-trained with large amounts of text data and is specialized for tasks such as natural language understanding, generation, translation, and summarization.

[0024] The training data for an AI model consists of, for example, aerial images of the target area (satellite images, aerial photographs, drone images, etc.) and polygon data in which paddy field and non-paddy field (e.g., field, forest) sections within the image are accurately annotated (by manually drawing boundaries, etc.). This gives the AI ​​model the ability to automatically identify fields and other sections within the image and generate polygons that indicate their boundaries when an untrained aerial image is input.

[0025] The preferred architecture for the AI ​​model is U-Net, which is widely used in the field of image segmentation, or a derivative model thereof. The training data used for training preferably consists of thousands to tens of thousands of image patches taken in various regions, cultivation seasons, and weather conditions, along with the corresponding annotation data. Annotation is performed by accurately tracing the boundaries of plots of land shown in the images, separated by farm ridges, waterways, roads, etc. Hyperparameters used during training, such as the learning rate, batch size, and number of epochs, can be adjusted as appropriate based on performance evaluation of the validation dataset.

[0026] The AI ​​system 40 receives the second image (or the third image described below) transmitted from the server 20, inputs it into the AI ​​model, and causes the AI ​​model to output a partition polygon. The AI ​​system 40 transmits the partition polygon output from the AI ​​model to the server 20.

[0027] A parcel polygon is a polygon that represents each of multiple parcels that exist in a target area, and is an example of a first polygon according to one embodiment of the present disclosure. In this specification, a "parcel" refers to a unit area of ​​land that can be visually identified on an aerial image, and includes, for example, a single farm field physically separated by a footpath, road, waterway, etc., as well as a mass of forest, residential area, etc. A "parcel" does not necessarily coincide with an administrative boundary such as a registered land address. Parcels represented by a parcel polygon may include paddy field parcels as well as non-paddy field parcels.

[0028] The AI ​​model may be stored, for example, in the storage unit 202 of the server 20. Alternatively, the AI ​​model may be externally deployed as a cloud-based AI service and accessed from the server 20 via an API (Application Programming Interface).

[0029] If the AI ​​model is a generative AI model, a prompt instructing the AI ​​model to generate a first polygon is also input to the AI ​​model. The prompt is a text input by the user to give instructions to the AI ​​model in natural language, and may include specific instructions such as, for example, "Please identify the paddy field plots from this image and generate polygons." The prompt may be hard-coded into a program stored in the AI ​​system 40. Alternatively, for example, an input device (not shown) provided in the server 20 may accept an input operation of the prompt, and the server 20 may transmit the prompt to the AI ​​system 40. Alternatively, the prompt may be pre-stored in the storage unit 202 or the AI ​​system 40.

[0030] There may be one or more AI models. When multiple AI models are used, they may be the same type of AI model, or different types of AI models, such as one being a semantic segmentation model and the other being an object detection model.

[0031] Each information processing device, such as the terminal device 10, the server 20, and the AI ​​system 40, is configured by a computer 90 (see FIG. 7) equipped with an arithmetic unit and a storage device. The basic hardware configuration of the computer 90 and the basic functional configuration of the computer 90 realized by the basic hardware configuration will be described later. Note that for each of the terminal device 10, the server 20, and the AI ​​system 40, descriptions that overlap with the basic hardware configuration and basic functional configuration of the computer 90 will be omitted.

[0032] <2-2 Terminal Device Configuration> Fig. 2 is a block diagram showing an example configuration of the terminal device 10 shown in Fig. 1. The terminal device 10 provides an interface for a user to use the polygon generation service. As shown in Fig. 2, the terminal device 10 includes a communication unit 120, an input device 13, an output device 14, an audio processing unit 170, a microphone 171, a speaker 172, a camera 160, a position information sensor 150, an acceleration sensor 155, a storage unit 180, and a control unit 190. The blocks included in the terminal device 10 are electrically connected by, for example, a bus or the like.

[0033] The communication unit 120 performs processes such as modulation and demodulation for communication between the terminal device 10 and an external device (for example, the server 20). The communication unit 120 performs transmission processing on signals generated by the control unit 190 and transmits the signals to the external device. The communication unit 120 performs reception processing on signals received from the external device and outputs the signals to the control unit 190. This allows the user to specify a target area or send a polygon generation request to the server 20, or receive the generated paddy field polygons or map data.

[0034] The input device 13 is a device for a user to input instructions or information. The input device 13 is realized, for example, by a touch-sensitive device 131 that inputs instructions by touching the operation surface. If the terminal device 10 is a PC or the like, the input device 13 may be realized by a reader, keyboard, mouse, or the like. The input device 13 may include a receiving port that receives instructions input by the user as electrical signals. The user can specify, via this input device, position information of the target area (target area information described below), a period for polygon generation, etc.

[0035] The output device 14 is a device for presenting information to the user. The output device 14 is realized, for example, by a display 141. The display 141 displays various information under the control of the control unit 190. The display 141 is realized, for example, by an LCD (Liquid Crystal Display) or an organic EL (Electro-Luminescence) display. The generated paddy field polygons are displayed on this display superimposed on map data, allowing the user to visually confirm the results.

[0036] The audio processing unit 170 performs, for example, digital-to-analog conversion processing of an audio signal. The audio processing unit 170 converts a signal provided from the microphone 171 into a digital signal and provides the converted signal to the control unit 190. The audio processing unit 170 also provides the audio signal to the speaker 172. The audio processing unit 170 is realized, for example, by a processor for audio processing. The microphone 171 receives an audio input and provides an audio signal corresponding to the audio input to the audio processing unit 170. The speaker 172 converts the audio signal provided from the audio processing unit 170 into audio and outputs the audio to the outside of the terminal device 10. This makes it possible to realize a variety of user interfaces, such as audio operations or notifications of processing completion.

[0037] The camera 160 is an imaging device that captures images using visible light. In other words, the camera 160 is a device that receives visible light using a light-receiving element and outputs image data as an image capture signal. The camera 160 captures an image of a subject in a certain direction and within a certain imaging range relative to the terminal device 10, and outputs image data as the image capture result. If the camera 160 has a function that allows the imaging range, or more precisely, the angle of view, to be adjustable, the camera 160 also outputs information regarding this angle of view. This function is known as a zoom function. In the future, it is conceivable that images captured by the camera of the terminal device may be used as auxiliary information.

[0038] The position information sensor 150 is a sensor that detects the position of the terminal device 10 and is generally a GNSS device, such as a GPS module. The GPS module is a receiving device used in a satellite positioning system. The satellite positioning system receives signals from at least three or four satellites and detects the current position of the terminal device 10 equipped with the GPS module as coordinate values ​​based on the received signals. The position information sensor 150 may detect the current position of the terminal device 10 from the position of a wireless base station to which the terminal device 10 connects via the communication unit 120. This allows the user to request polygon generation for a field around the current location and to check the generated polygon on site.

[0039] The acceleration sensor 155 is a sensor that detects acceleration applied to the terminal device 10. Preferably, the acceleration sensor 155 has a function of detecting tilt around each axis (X-axis, Y-axis, Z-axis) of a three-dimensional coordinate system with the position of the terminal device 10 as the origin. The acceleration sensor 155 having such a function can detect the attitude of the terminal device 10, that is, the direction with respect to the X-axis, Y-axis, and Z-axis, by detecting the gravitational acceleration of the gravitational force with respect to the earth.

[0040] The storage unit 180 is realized by the memory 15 and storage 16 shown in FIG. 1, and stores data and programs used by the terminal device 10. The programs include application programs such as a web browser application. This allows the user to use the polygon generation service via a web browser or to install and use a dedicated application.

[0041] The control unit 190 is realized, for example, by the processor 19 reading a program stored in the storage unit 180 and executing instructions included in the program. The control unit 190 controls the operation of the terminal device 10. The control unit 190 performs the functions of an operation reception unit 191, a transmission / reception unit 192, and a presentation control unit 193 by operating in accordance with the read program.

[0042] The operation reception unit 191 performs processing for receiving instructions or information input from the input device 13. Specifically, the operation reception unit 191 receives instructions or information input from the touch-sensitive device 131. The transmission / reception unit 192 performs processing for the terminal device 10 to transmit and receive data to and from an external device in accordance with a communication protocol. Specifically, the transmission / reception unit 192 transmits instructions or information input by the user to the server 20. The transmission / reception unit 192 receives information transmitted from the server 20. The presentation control unit 193 controls the output device 14 to present various information to the user. These functional collaborations allow the user to intuitively use services.

[0043] <2-3 Server configuration> Fig. 3 is a block diagram showing an example of the configuration of the server 20 shown in Fig. 1. The server 20 is an information processing device that serves as the core of the polygon generation service, and executes a series of processes such as acquisition of the first and second images, analysis, polygon generation, and final extraction of the paddy field polygon. As shown in Fig. 3, the server 20 performs the functions of a communication unit 201, a storage unit 202, and a control unit 203.

[0044] The communication unit 201 performs processes such as modulation and demodulation for communication between the server 20 and an external device (e.g., the terminal device 10, a ground station, or another data provider). The communication unit 201 performs transmission processing on a signal generated by the control unit 203 and transmits the signal to the external device. The communication unit 201 performs reception processing on a signal received from the external device and outputs the signal to the control unit 203. This enables the transmission of a request to acquire the first and second images, the reception of the first and second images, the transmission of generated polygons, and the like. The storage unit 202 is realized by the memory 25 and the storage 26, and stores data and programs used by the server 20. The programs include application programs such as a web browser application. The storage unit 202 stores, for example, a polygon database 2021 and an app 2022.

[0045] The polygon database 2021 stores various data necessary for the polygon generation service. The various data include, for example, information about the first image, information about the second image, identification information, information about division polygons, and information about paddy field polygons. The polygon database 2021 and the identification information will be described in detail later.

[0046] The application 2022 is an application for managing the use of the polygon generation service by the user. The application 2022 runs, for example, in the background of other applications installed on the server 20, and monitors the processes executed by the user. The user can access the application 2022 in the server 20 using a web browser application installed on the terminal device 10.

[0047] The server 20 may, for example, grasp / manage the usage status of the app 2022 and execute a predetermined analysis process. This allows the usage status or issues of the service to be grasped and used for improvements. Also, for example, the app 2022 may be installed in the terminal device 10 and stored in the storage unit 180. In this case, some functions can be used offline or faster response can be expected.

[0048] The control unit 203 is realized by the processor 29 reading a program stored in the storage unit 202 and executing instructions included in the program. The control unit 203 controls the operation of the server 20. The control unit 203 operates in accordance with the read program to fulfill the functions of a reception control module 2031, a transmission control module 2032, a presentation control module 2033, and a generation processing module 2034.

[0049] The reception control module 2031 controls the process in which the server 20 receives signals from external devices in accordance with a communication protocol. The reception control module 2031 receives the target area (e.g., location information, name, etc.) transmitted from the terminal device 10. The reception control module 2031 receives a plurality of first images (low resolution, for time-series data) of the target area photographed from the air during a first period. The first period is a period sufficient to cover the growth cycle of a rice paddy, which is set to cover at least one cropping cycle (e.g., flooding, transplanting, growing, harvesting, draining) in the rice paddy, and includes, for example, one year's worth of data.

[0050] The reception control module 2031 also receives one or more second images (high resolution, for polygon generation) of the target area taken aerial photographs during a second period. The second period is a period necessary to generate the first polygon, set to include a period during which the boundaries of the plots in the target area are identifiable, and may be the same as or different from the first period. For example, the first period may be one year, while the second period may be a short period such as a few weeks that includes a specific growth stage (e.g., immediately after rice planting). Both the first period and the second period can be set arbitrarily. To generate the first polygon (a polygon that includes non-paddy fields), image data from a single period when there is no cloud cover is generally sufficient.

[0051] Here, the second image is primarily intended to accurately extract the detailed shape of the plot. Therefore, the second image is taken from the air once or several times during specific periods when the boundaries of plots, such as fields, can be clearly distinguished from surrounding features, such as during the flooding period (immediately after rice planting) when the ground is flooded with water, or before harvest when the rice matures and turns golden brown. For example, if the second image is a satellite image such as the WorldView series (high-resolution commercial satellite), it is often taken from the air several times during a specific project period. Furthermore, if the second image is an aerial image taken by an aircraft or drone, high-resolution imagery can be captured on demand, so it is often taken on a specific day (i.e., only once).

[0052] The transmission control module 2032 controls the process of transmitting signals to external devices by the server 20 in accordance with a communication protocol. The presentation control module 2033 controls the process of presenting various types of information to the user.

[0053] The generation processing module 2034, in cooperation with the AI ​​system 40, analyzes time-series data in which the acquired multiple first images are arranged in chronological order, and generates identification information indicating the results of identifying paddy fields present in the target area. Specifically, for example, the generation processing module 2034 calculates water-related vegetation indices such as NDWI (Normalized Difference Water Index), EVI (Enhanced Difference Vegetation Index), and LSWI (Land Surface Water Index) for each pixel from each of the multiple first images. These vegetation indices clearly represent seasonal fluctuation patterns such as waterlogging, vegetation growth, and drainage that are specific to paddy fields.

[0054] The generation processing module 2034 identifies pixels that show patterns specific to paddy fields by tracking seasonal fluctuations in the calculated vegetation index for each first image over time (such as a sudden rise in water level during the puddling season, an increase in EVI due to rice growth, and a decrease in NDWI due to water drainage), and generates a low-resolution paddy field / non-paddy field map as identification information. In other words, the generation processing module 2034 identifies pixels that show patterns specific to paddy fields by analyzing seasonal fluctuation patterns in the vegetation index obtained from the multiple calculated first images, and generates a low-resolution paddy field / non-paddy field map as identification information.

[0055] The following method, for example, can be used to analyze the seasonal variation patterns described above. First, preprocessing is performed, such as removing noise such as clouds contained in the time-series data using cloud mask information and applying a smoothing filter (e.g., moving average method). Next, the similarity between a typical vegetation index variation pattern (template) predefined based on the growth cycle of rice paddies and the time-series data of each pixel is calculated using an algorithm such as dynamic time warping (DTW). Pixels whose similarity is equal to or greater than a predetermined threshold are identified as "paddy fields." Alternatively, a machine learning classifier such as a random forest, support vector machine, or recurrent neural network (RNN) can be trained using known time-series data of rice paddies and non-paddy fields as training data, and the time-series data of each pixel can be input to this trained classifier to classify whether the pixel is a rice paddy or not.

[0056] The specific information is information that indicates the results of identifying paddy fields that exist in the target area. The paddy field / non-paddy field map is, for example, a map composed of binary information (after threshold processing) that indicates whether each pixel is a paddy field or not. The specific information may be information other than the paddy field / non-paddy field map. Examples of specific information other than the paddy field / non-paddy field map include a probability map and growth stage estimation information. The probability map is a map that indicates the probability that each pixel is a paddy field. The growth stage estimation information is information that indicates the results of estimating the rice growth stage at each pixel.

[0057] The generation processing module 2034 inputs the acquired one or more second images into an AI model and causes the AI ​​model to generate partition polygons representing each of the multiple partitions present in the target area. When multiple second images are acquired, they may be input to the AI ​​model sequentially or all at once. The second image has a higher resolution than the first image and is suitable for capturing the detailed shape of paddy fields. However, it is not necessary for the second image to have a higher resolution than the first image; the first and second images may have the same resolution. Furthermore, the resolution of each of the first and second images can be set arbitrarily.

[0058] The generation processing module 2034 inputs, for example, a high-resolution (e.g., 30 cm) satellite image (second image) acquired separately from the satellite image (first image) for time-series data into a semantic segmentation AI model that has learned the shape of rice paddies. This generates parcel polygons. The AI ​​model may distinguish parcels as either rice paddies or non-rice paddies (e.g., fields, forests, residential areas, etc.), but the parcel polygon generation stage primarily extracts the shape of the parcels.

[0059] The generation processing module 2034 uses the identification information to extract paddy field polygons from among the plurality of division polygons. There are several possible modes for this extraction processing.

[0060] In a first mode, the generation processing module 2034 overlays the division polygons with the paddy field / non-paddy field map (specific information) on a geographic information system (GIS). If the overlapping area between an area indicating paddy fields in the paddy field / non-paddy field map (paddy field area indicated by the specific information) and an individual division polygon is equal to or greater than a predetermined percentage (e.g., 80%) of the area of ​​the division polygon, the generation processing module 2034 extracts the division polygon as a paddy field polygon. This spatial filtering process (equivalent to a logical AND operation) can exclude paddy fields that appear to be paddy fields but are actually non-paddy fields, such as fields, and generate paddy field polygons that accurately indicate only paddy fields as the final product.

[0061] In a second aspect, when the identification information is a probability map indicating the likelihood of each pixel being a paddy field, the extraction process is performed in the following steps. First, the generation processing module 2034 overlays this probability map on the division polygon generated by the AI ​​model on a geographic information system (GIS). The area defined as a set of pixels on the probability map whose probability of being a paddy field exceeds a predetermined first threshold (e.g., 0.5) is the "paddy field area indicated by the identification information" according to one aspect of the present disclosure. Next, the generation processing module 2034 calculates the average or median probability value of all pixels contained in each division polygon. If the calculated value exceeds a predetermined second threshold, the division polygon is extracted as the final paddy field polygon. The predetermined second threshold is a threshold set in advance according to the required accuracy as a criterion for finally identifying the entire division polygon as a paddy field. This enables more flexible and quantitative determination based on the confidence level that the area is a paddy field.

[0062] In other words, based on the identification information, the generation processing module 2034 extracts, from among a plurality of division polygons, those for which the correspondence between the division polygon and the paddy field area indicated by the identification information satisfies a predetermined criterion as paddy field polygons. Here, in the first mode, "the overlapping area is a predetermined percentage (e.g., 80%) or more of the area of ​​the division polygon" and in the second mode, "when the average or median of the probability values ​​of all pixels exceeds a predetermined second threshold" each correspond to the requirement that "the correspondence between the first polygon and the paddy field area indicated by the identification information satisfies a predetermined criterion."

[0063] [3 Data Structure] 4 is a diagram showing the data structure of tables stored in the server 20. Note that FIG. 4 is merely an example and does not exclude data that is not listed. Furthermore, even data listed in the same table may be stored in separate storage areas in the storage unit 202.

[0064] Fig. 4 is a diagram showing the data structure of the polygon database 2021. The polygon database 2021 shown in Fig. 4 is a table having columns of target area information, first image information, second image information, specific information, division polygon information, paddy field polygon information, and processing log, with the case ID as a key.

[0065] The item "Project ID" is an identifier for uniquely identifying a project to be implemented in the target area.

[0066] The "Target Area Information" item stores basic information about the area that is the target for polygon generation (e.g., target area ID, name, geographic coordinates, area, user ID, etc.). The target area ID is an identifier that uniquely identifies the target area. The name is the proper name of the target area or the name of the project that is planned to be carried out in the target area. The geographic coordinates are geospatial data such as latitude and longitude information that indicate where the target area is located on Earth. The area is the area of ​​the target area. The user ID is an identifier that uniquely identifies the user.

[0067] The "Primary Image Information" item stores metadata (e.g., image ID, shooting date and time, resolution, image path, sensor type, band information, etc.) about the low-resolution, high-frequency primary image used in time-series data analysis. The image ID is an identifier for uniquely identifying the primary image. The shooting date and time is the date and time the primary image was taken from the air. The resolution is information that indicates the spatial resolution of the primary image, that is, the image detail. The image path is information that indicates the location on storage where the primary image is actually stored (e.g., file path, URL, etc.). In other words, the primary image itself is stored in separate storage and referenced by the image path. The sensor type is the type of sensor that acquired the primary image (e.g., optical sensor, SAR (synthetic aperture radar) sensor, etc.). The band information is information that indicates the type and number of spectral bands the primary image has (e.g., visible light, near infrared, shortwave infrared, etc.).

[0068] The "Second Image Information" item stores metadata (e.g., image ID, shooting date and time, resolution, image path, sensor type, band information, etc.) about the high-resolution, single-period second image used for polygon generation by the AI ​​model. The image ID is an identifier that uniquely identifies the second image. The shooting date and time is the date and time the second image was taken from the air. The resolution is information that indicates the spatial resolution of the second image, that is, the level of detail of the image. The image path is information that indicates the location on storage where the second image is actually stored. The sensor type is the type of sensor that acquired the second image. The band information is information that indicates the type and number of spectral bands that the second image has.

[0069] The "Specific Information" item stores specific information about paddy fields (paddy field / non-paddy field map) generated from the time series analysis of the first image, as well as related information about the specific information. Specifically, the "Specific Information" item stores the specific information ID for uniquely identifying the specific information, the date and time the specific information was generated, the start and end dates and times of the analysis period for the time series data, paddy field / non-paddy field map data (e.g., polygon data in GeoJSON format or raster data, etc.), the version of the analysis model used in the time series analysis, the threshold value used to determine paddy fields (or the identification results by an AI model, etc.), etc.

[0070] The "Division Polygon Information" item stores information about division polygons. Specifically, the "Division Polygon Information" item includes a polygon ID for uniquely identifying the division polygon, a target area ID, the date and time the division polygon was generated, polygon coordinate data (e.g., GeoJSON format) indicating the coordinates of the division polygon, and the generation model version of the AI ​​model used to generate the division polygon. The polygon coordinate data is stored for each of the multiple division polygons generated.

[0071] The "paddy field polygon information" item stores information about paddy field polygons. Specifically, the "division polygon information" item includes a polygon ID for uniquely identifying the paddy field polygon, a target area ID, the extraction date and time of the paddy field polygon, polygon coordinate data (e.g., GeoJSON format) indicating the coordinates of the paddy field polygon, a reliability score, and the extraction method from the division polygon. Polygon coordinate data is stored for each of the multiple extracted paddy field polygons.

[0072] The reliability score is a numerical value or index that indicates how accurate or reliable the paddy field polygon is as an indication of a paddy field. The generation processing module 2034 calculates the reliability score by comprehensively considering, for example, the degree of agreement between the specific information (paddy field / non-paddy field map) and the division polygon, the generation accuracy of the AI ​​model, etc.

[0073] The "Processing Log" item stores log information such as the execution date and time of each processing step (image acquisition, time series analysis, AI model processing, paddy field polygon extraction, etc.), the success / failure of each processing step, error messages during processing, processing time, etc. This makes it possible to monitor the system's operational status or to investigate the cause of any problems that may occur.

[0074] These pieces of information are stored in the corresponding items of the polygon database 2021 as follows. That is, information relating to the first image and the second image is stored in the corresponding items of the polygon database 2021 when the server 20 receives satellite images transmitted from the artificial satellite 30 via a ground station. The specific information, division polygon information, and paddy field polygon information are sequentially accumulated in the corresponding items of the polygon database 2021 by the generation processing module 2034.

[0075] [4 actions] An example of the operation of the server 20 when extracting paddy field polygons will be described with reference to Fig. 5. Fig. 5 is a flowchart showing an example of the operation of the server 20 when extracting paddy field polygons.

[0076] In step S11 shown in FIG. 5, the server 20 acquires a plurality of first images of the target area taken from the air during a predetermined first period.

[0077] Specifically, for example, after receiving a polygon generation request from a user, the transmission control module 2032 identifies a target area for polygon generation and a first period (e.g., the past year) required for time-series analysis of paddy fields. The transmission control module 2032 identifies the target area by, for example, referring to information for identifying the target area included in the polygon generation request (e.g., geographic information such as address, name, map data, etc.). Also, for example, information indicating the first period is pre-stored in the storage unit 202, and the transmission control module 2032 identifies the first period by referring to this information indicating the first period.

[0078] The transmission control module 2032 then transmits to the ground station a transmission request for first images (e.g., Sentinel-2 satellite images) that are taken frequently and at low resolution (e.g., 10 m / pixel) corresponding to the first period and target area. The reception control module 2031 receives the first images from the ground station that accepted the transmission request, and stores their reference information (e.g., image path, metadata) in the "first image information" item of the polygon database 2021. At this time, the first images may be acquired together with quality information such as cloud mask information.

[0079] In step S12, the server 20 analyzes time-series data in which a plurality of first images are arranged in time series, and generates identification information indicating the results of identifying paddy fields present in the target area.

[0080] Specifically, for example, the generation processing module 2034 reads out a plurality of first images (time-series data) acquired from the polygon database 2021. Then, from this time-series data, it calculates water-related indices such as NDWI (Normalized Difference Water Index), EVI (Enhanced Difference Vegetation Index), and LSWI (Land Surface Water Index) for each pixel.

[0081] In step S13, the server 20 acquires a plurality of second images of the target area taken from the air during a predetermined second period.

[0082] Specifically, for example, the transmission control module 2032 identifies a second period (e.g., the past six months) required to generate the division polygon. For example, information indicating the second period is pre-stored in the storage unit 202, and the transmission control module 2032 identifies the second period by referring to the information indicating the second period. The transmission control module 2032 then transmits to the ground station a transmission request for a second image (e.g., a WorldView satellite image) with a high resolution (e.g., 30 cm / pixel) required to generate the division polygon and corresponding to the second period. The second period may be the same as the first period, or may be a short period limited to a specific period (e.g., immediately after rice planting or before harvest) when the shape of the field can be clearly identified. The reception control module 2031 receives the second image from the ground station that accepted the transmission request and stores the reference information (e.g., image path, metadata) in the "second image information" field of the polygon database 2021. At this time, the second image may be acquired together with quality information such as cloud mask information.

[0083] In step S14, the server 20 inputs one or more second images into the AI ​​model, and causes the AI ​​model to generate partition polygons.

[0084] Specifically, for example, the generation processing module 2034 reads one or more second images acquired from the polygon database 2021. Then, the generation processing module 2034 inputs these second images into an AI model included in the AI ​​system 40. The AI ​​model generates division polygons based on the input image data. The generated division polygons include, for example, the shape of each division and information indicating the possibility that it is a field (e.g., paddy field, farmland, forest, etc.). Furthermore, the generation processing module 2034 stores information about the generated division polygons in the polygon database 2021 in the item "division polygon information."

[0085] In step S15, the server 20 uses the specific information to extract paddy field polygons from among the plurality of division polygons.

[0086] Specifically, for example, the generation processing module 2034 reads paddy field / non-paddy field map data and polygon coordinate data for multiple division polygons from the polygon database 2021. Then, these data are overlaid on a geographic information system (GIS). The generation processing module 2034 extracts the overlapping portions of each division represented by multiple division polygons with areas determined to be paddy fields in the specific information. This eliminates non-paddy fields that appear to be farmland but are actually fields, and generates paddy field polygons that accurately represent only paddy fields as the final output. This integration using logical AND operations clarifies the distinction between paddy fields and non-paddy fields, which is difficult to distinguish using high-resolution images alone, significantly improving the accuracy of paddy field polygons. The generation processing module 2034 stores information about the extracted paddy field polygons in the "paddy field polygon information" item of the polygon database 2021.

[0087] In step S16, the server 20 presents the extracted paddy field polygons to the user.

[0088] Specifically, for example, the presentation control module 2033 controls the presentation control unit 193 to present the paddy field polygons to the user. The transmission control module 2032, for example, transmits information for displaying the paddy field polygons (e.g., GeoJSON data or an image file) to the terminal device 10. The transmission / reception unit 192, for example, receives information for displaying the paddy field polygons from the server 20. For example, upon receiving a presentation request from the user, the presentation control unit 193 displays the paddy field polygons on the display 141, superimposed on the map data (including the target area). This allows the user to visually confirm where the paddy fields are located within the target area and what shape they have. Note that the processing of step S16, i.e., presentation of the paddy field polygons, is not essential.

[0089] [5 Screen Examples] An example of the screen of the display 141 when map data showing paddy field polygons is presented to the user will be described below with reference to Fig. 6. Fig. 6 is a schematic diagram showing an example of map data 1411 showing paddy field polygons 1412.

[0090] In the example screen of Fig. 6, map data 1411 on which paddy field polygons 1412 are displayed is displayed on the display 141. In the map data 1411, paddy field areas are accurately divided by the paddy field polygons 1412, and the boundaries are clearly visualized. This allows the user to grasp the exact position or shape of the paddy fields at a glance. Note that this example screen is merely an example, and various variations can be adopted for the display content and display mode of the paddy field polygons.

[0091] [6 Summary] As described above, in this embodiment, the server 20 acquires multiple low-resolution, high-frequency first images of the target area taken from the air during a predetermined first period, analyzes the time-series data from these images, and generates identification information indicating the results of paddy field identification. The server 20 also acquires multiple high-resolution second images of the target area taken from the air during a predetermined second period, inputs these second images into an AI model included in the AI ​​system 40, and causes the AI ​​model to generate multiple division polygons. The server 20 then combines the generated identification information with the multiple division polygons to accurately extract paddy field polygons.

[0092] This makes it possible to clearly distinguish between paddy and non-paddy fields, which is difficult to distinguish using high-resolution, single-temporal satellite images alone, by combining them with seasonal patterns specific to paddy fields obtained from low-resolution, high-frequency time-series data such as Sentinel-2. This integrated approach significantly improves the accuracy of distinguishing between paddy and non-paddy fields, even in areas where it is difficult to distinguish between visually similar agricultural land uses.

[0093] [7 Variations] The polygon generation service described in this embodiment can be modified in various ways, and various applications are possible without departing from the technical scope of this disclosure. In particular, the cloud mask processing and substitute image use technologies are extremely important in feature detection or classification using remote sensing data, and can also be applied to the extraction of paddy field polygons according to one aspect of this disclosure. Below, application examples of cloud mask processing and substitute image use are described.

[0094] <7-1 First Modification> This first modification provides a detailed explanation of cloud mask processing, which is a useful preprocessing step for extracting areas where the ground surface is clearly visible from a high-resolution image.

[0095] That is, the server 20 may, for example, apply cloud mask processing to each of one or more second images, and extract second images showing the ground surface from the one or more second images. Furthermore, for example, the server 20 may input the second images showing the ground surface to an AI model. The server 20 may employ, for example, the following method as an extraction method using cloud mask processing.

[0096] Specifically, for example, the generation processing module 2034 may perform threshold processing as a pixel-based cloud detection method, utilizing the characteristic that clouds have high reflectance in the blue band or shortwave infrared band in the image. In other words, because clouds have high reflectance in the visible light (especially blue) and shortwave infrared regions, the generation processing module 2034 may use a method of identifying pixels whose reflectance in these bands exceeds a certain threshold as clouds. Note that the above characteristics are purely relative. For example, when compared with water bodies, it is expected that clouds will have higher reflectance than water bodies, but when compared with dry soil, soil may have higher reflectance than clouds.

[0097] For example, the generation processing module 2034 may employ a detection method that utilizes the fact that vegetation and water indexes such as NDVI or NDWI show abnormal values ​​(extremely low or high values) due to clouds. For example, if the NDVI is significantly low in an area with active vegetation, the generation processing module 2034 determines that the pixel is likely to be covered by clouds.

[0098] Furthermore, for example, the generation processing module 2034 may use an object-based cloud detection method to identify cloud regions using not only individual pixels but also the shape, texture, shadow characteristics, etc. of the entire cloud mass. For example, by detecting that the cloud shape is irregular, that the cloud edge is clear, or that a shadow cast by the cloud on the ground is present, it becomes possible to generate a cloud mask with higher accuracy.

[0099] Furthermore, the generation processing module 2034 may perform cloud mask processing using a semantic segmentation model using a convolutional neural network (CNN). This method is effective because it can classify clouds at the pixel level with high accuracy. The AI ​​model used in this method is trained, for example, using pairs of a large number of cloud-embedded images and corresponding manually annotated cloud masks as training data. This enables robust cloud detection even in the presence of complex cloud shapes, various lighting conditions, and diverse ground surfaces. By adopting or combining any of these methods, the generation processing module 2034 can achieve more robust cloud mask processing and maximize the quality of the input image (second image) to the AI ​​model.

[0100] In this way, according to this first variant, it is possible to reduce the risk that the AI ​​model provided in the AI ​​system 40 will mistakenly recognize clouds as features, and improve the accuracy of the generated division polygons.

[0101] <7-2 Second Modification> This second modification provides a detailed explanation of the use of alternative images. It is an important means of compensating for data loss when high-resolution images cannot be used due to clouds.

[0102] That is, when performing cloud mask processing to extract a second image showing the ground surface (first modified example), the server 20 may acquire a third image of the target area photographed from the air during a predetermined second period. Also, for example, the server 20 may replace the second image not showing the ground surface with a third image and input the third image to the AI ​​model.

[0103] Specifically, for example, the generation processing module 2034 may acquire third images with a medium resolution (e.g., SPOT images at 1.5 m / pixel) or a low resolution (e.g., Sentinel-2 images at 10 m / pixel), which are taken from the air more frequently than the second images, from the satellite 30 via a ground station during periods or locations when the second images with high resolution (e.g., 30 cm / pixel) are obscured by clouds. Then, the generation processing module 2034 may input the acquired third images into the AI ​​model as a substitute, for example. The third images serve to complement the state of the Earth's surface during periods when the second images with high resolution cannot be acquired.

[0104] Here, the purpose of the third image is to substitute (supplement data) when the high-resolution second image cannot be acquired due to clouds. This ensures data continuity and makes it possible to track changes in rice paddies over time. For example, if the third image is a satellite image acquired from the Sentinel-2 satellite, it is taken from the air at a high frequency, from once every few days to once a week, so there is a high probability that the satellite image will be free from cloud cover. Also, if the third image is an SAR image, for example, it uses microwaves, so it can be taken from the air without being affected by clouds at all. Furthermore, data can be reliably acquired even in bad weather, making it a strong candidate for a substitute image.

[0105] It is not essential that the second image has a high resolution and the third image has a low resolution. For example, the third image may have a high resolution, or the second image and the third image may have approximately the same resolution.

[0106] More specifically, the generation processing module 2034 may, for example, input a low-resolution, high-frequency image (third image) from a period when a high-resolution image (second image) is missing, thereby estimating the probability of the presence of paddy fields during that period and ensuring continuity in the identification of paddy fields over time. For example, multi-temporal data such as Sentinel-2 data shows clear patterns of change in spectral reflectance during the flooding, growing, and harvesting periods of paddy fields, making it extremely useful for determining the presence or absence of paddy fields even at low resolution. This makes it possible to provide more reliable data for applications such as year-round management of paddy fields or estimation of past cropping history, even when images from a specific period are unavailable.

[0107] Furthermore, the generation processing module 2034 may also use data interpolation techniques that statistically or machine learningly interpolate information about missing portions using, for example, paddy field growth pattern information obtained from time-series data analysis or data about similar paddy fields located near the target area. That is, the generation processing module 2034 may use, for example, a statistical method such as linear interpolation or spline interpolation to estimate spectral values ​​for the missing period in the second image, or a deep learning model such as GAN (Generative Adversarial Networks) to generate a third image corresponding to the missing area. Furthermore, for example, the generation processing module 2034 may estimate missing information using healthy pixel information around the missing area or data about similar paddy fields located near the target area.

[0108] Thus, according to this second modification, when the second image contains many clouds, the second image that does not show the ground surface can be substituted with satellite data (third image) that is taken more frequently, preventing a decrease in analytical accuracy due to missing data. Furthermore, by using the high-frequency data (third image) to confirm the existence of paddy fields during the target period using an AI model, the presence of paddy fields can be estimated with high accuracy even during periods or locations where low-frequency data (second image) is lacking. This overcomes issues specific to remote sensing data, such as differences in the quality or characteristics of aerial images (second image) and the influence of clouds, dramatically improving the accuracy of paddy field polygons.

[0109] Here, "high frequency of capture" refers to a frequency that reduces the risk of observations being unavailable due to clouds or other factors at certain times and ensures that at least one clear image showing the ground surface can be obtained within a certain period. Specifically, while secondary imagery (e.g., the WorldView series) is captured once every few weeks to several months, high-frequency data refers to data captured every few days to once a week or more. This allows the AI ​​model to confirm the presence of rice paddies during the target period using high-frequency data (tertiary imagery), enabling highly accurate estimation of the presence of rice paddies even in periods or locations where low-frequency data (secondary imagery) is lacking. This overcomes challenges specific to remote sensing data, such as differences in the quality or characteristics of aerial imagery (secondary imagery) and the influence of clouds, dramatically improving the accuracy of rice paddy polygons.

[0110] The paddy field polygon generation service described in this embodiment presents generated paddy field polygons to the user as an example, but various other application examples are conceivable. For example, paddy field area can be automatically calculated based on the generated paddy field polygons and provided as basic data for carbon credit applications or water resource management. Furthermore, by combining paddy field polygons with other geospatial information (e.g., soil data, meteorological data), the service can be developed into precision agriculture support services such as crop growth forecasts, yield estimates, and pest and disease risk assessments. Furthermore, wide-area paddy field distribution maps or data on their changes over time can be provided to local governments or research institutions, contributing to regional development plans or environmental policy formulation.

[0111] [8 Basic Computer Hardware Configuration] 7 is a block diagram showing the basic hardware configuration of a computer 90. The computer 90 represents a general-purpose hardware platform for configuring an information processing device (terminal device 10, server 20, etc.) according to the present disclosure. The computer 90 includes at least a processor 901, a main memory device 902, an auxiliary memory device 903, and a communication IF 991 (interface). These are electrically connected to each other by a communication bus 921.

[0112] The processor 901 is hardware for executing an instruction set written in a program. The processor 901 is composed of an arithmetic unit, a register, a peripheral circuit, etc., and controls various calculation processes, data analysis, execution of AI models, etc. in this disclosure. It may also include an accelerator such as a multi-core processor with multiple cores or a GPU (Graphics Processing Unit).

[0113] The main memory device 902 is used to temporarily store programs and data processed by the programs. For example, it is a volatile memory such as a DRAM (Dynamic Random Access Memory). It allows high-speed data access and temporarily stores program code currently being executed, image data currently being processed, intermediate results, etc.

[0114] The auxiliary storage device 903 is a storage device for permanently storing data and programs. For example, it may be a flash memory, a hard disk drive (HDD), a magneto-optical disk, a CD-ROM, a DVD-ROM, or a semiconductor memory (solid state drive (SSD)). Large amounts of image data or analysis results, such as the aerial image database, specific information database, and polygon information database in this disclosure, are stored here.

[0115] The communication IF 991 is an interface for inputting and outputting signals for communication with other computers over a network using wired or wireless communication standards, enabling data transmission and reception between a terminal device and a server, between a server and a ground station, and between a server and other data providers.

[0116] The network is composed of the Internet, a LAN, various mobile communication systems constructed by wireless base stations, etc. For example, the network includes 3G, 4G, and 5G mobile communication systems, LTE (Long Term Evolution), and wireless networks (e.g., Wi-Fi (registered trademark)) that can connect to the Internet via a predetermined access point. In the case of a wireless connection, communication protocols include, for example, Z-Wave (registered trademark), ZigBee (registered trademark), and Bluetooth (registered trademark). In the case of a wired connection, the network also includes a direct connection using a USB (Universal Serial Bus) cable, etc.

[0117] Note that the computer 90 can be virtually realized by distributing all or part of each hardware configuration across multiple computers 90 and interconnecting them via a network. In this way, the computer 90 is a concept that includes not only a computer 90 housed in a single housing or case, but also a virtualized computer system such as a virtual server or distributed computing system in a cloud computing environment.

[0118] [9 Basic Functional Configuration of Computer 90] The following describes the functional configuration of a computer realized by the basic hardware configuration (FIG. 7) of the computer 90. The computer includes at least the functional units of a control unit, a storage unit, and a communication unit.

[0119] The functional units of the computer 90 can also be realized by distributing all or part of the functional units among multiple computers 90 interconnected via a network. The computer 90 is a concept that includes not only a single computer 90 but also a virtualized computer system such as a cloud environment or a distributed system.

[0120] The control unit is realized by the processor 901 reading out various programs stored in the auxiliary storage device 903, expanding them in the main storage device 902, and executing processing in accordance with the programs. The control unit can realize functional units that perform various types of information processing depending on the type of program. In this way, the computer is realized as an information processing device that performs information processing.

[0121] The storage unit is realized by a main storage device 902 and an auxiliary storage device 903. The storage unit stores data, various programs, and various databases. Furthermore, the processor 901 can allocate a storage area corresponding to the storage unit in the main storage device 902 or the auxiliary storage device 903 in accordance with the programs. Furthermore, the control unit can cause the processor 901 to execute processes for adding, updating, and deleting data stored in the storage unit in accordance with the various programs.

[0122] A database refers to a relational database, which manages data sets called masters and tables in a tabular format structurally defined by rows and columns, by relating them to each other. In a database, a table is called a table, a master, a column in a table is called a column, and a row in a table is called a record. In a relational database, relationships between tables and masters can be set and associated.

[0123] Typically, each table and each master has a column set as a primary key to uniquely identify a record, but setting a primary key to a column is not essential. The control unit can cause the processor 901 to add, delete, or update records in specific tables and masters stored in the storage unit according to various programs.

[0124] Furthermore, by storing data, various programs, and various databases in the storage unit, it can be considered that the information processing device and information processing system according to the present disclosure have been manufactured.

[0125] Note that the databases and masters in this disclosure may include any data structure (list, dictionary, associative array, object, etc.) in which information is structurally defined. The data structure also includes data that can be considered as a data structure by combining data with functions, classes, methods, etc. written in any programming language.

[0126] The communication unit is realized by the communication IF 991. The communication unit realizes a function of communicating with other computers 90 via a network. The communication unit can receive information transmitted from other computers 90 and input the information to the control unit. The control unit can cause the processor 901 to execute information processing on the received information in accordance with various programs. In addition, the communication unit can transmit information output from the control unit to other computers 90.

[0127] Furthermore, the above-described configurations, functions, processing units, processing means, etc. may be partially or entirely implemented in hardware, for example, by designing them as integrated circuits. The present invention can also be realized by software program code that implements the functions of the embodiments. In this case, a storage medium on which the program code is recorded is provided to a computer, and a processor included in the computer reads the program code stored in the storage medium. In this case, the program code itself read from the storage medium implements the functions of the above-described embodiments, and the program code itself and the storage medium on which it is stored constitute the present invention. Examples of storage media for providing such program code include flexible disks, CD-ROMs, DVD-ROMs, hard disks, SSDs, optical disks, magneto-optical disks, CD-Rs, magnetic tape, non-volatile memory cards, and ROMs.

[0128] Furthermore, the program code that realizes the functions described in this embodiment can be implemented in a wide range of program or script languages, such as assembler, C / C++, perl, Shell, PHP, Java (registered trademark), JavaScript, and TypeScript.

[0129] Furthermore, the program code of the software that realizes the functions of the embodiments may be distributed via a network and stored in a storage means such as a computer's hard disk or memory, or in a storage medium such as a CD-RW or CD-R, and the processor of the computer may read and execute the program code stored in the storage means or the storage medium.

[0130] The functions performed by the components described herein may be implemented in circuitry or processing circuitry, including general-purpose processors, application-specific processors, integrated circuits, ASICs (Application Specific Integrated Circuits), CPUs (Central Processing Units), conventional circuits, and / or combinations thereof, programmed to perform the described functions. A processor includes transistors or other circuits and is considered to be circuitry or processing circuitry. A processor may also be a programmed processor that executes a program stored in a memory.

[0131] In this specification, a circuitry, unit, or means is hardware that is programmed to realize or performs the described functions, which may be any hardware disclosed herein or any hardware known to be programmed to realize or perform the described functions.

[0132] If the hardware is a processor that is considered to be a type of circuitry, the circuitry, means, or unit is a combination of the hardware and the software used to configure the hardware and / or processor.

[0133] Although several embodiments of the present disclosure have been described above, these embodiments can be embodied in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and modifications thereof are intended to be included in the scope of the inventions and their equivalents as defined in the claims, as well as within the scope and spirit of the inventions.

[0134] [10 Appendix] The matters described in the above embodiments will be supplemented below.

[0135] <Appendix 1> A program to be executed by a computer having a processor and a memory, The program causes a processor to execute the following steps: acquiring a plurality of first images of a target area taken from the air during a first period set to cover at least one cropping cycle of the rice paddies in the target area; analyzing the time-series pattern of fluctuations in vegetation indices obtained from the plurality of first images to generate identification information indicating the results of identifying the rice paddies; acquiring one or more second images of the target area taken from the air during a second period set to include a period when the boundaries of the plots in the target area can be identified; inputting the one or more second images into a trained AI model and causing the AI ​​model to generate first polygons representing each of the plurality of plots; and extracting, from the plurality of first polygons, those whose correspondence between the first polygons and the rice paddy areas indicated by the identification information meets predetermined criteria, as second polygons representing the rice paddies, based on the identification information.

[0136] <Appendix 2> A program described in Appendix 1, which further causes the processor to execute a step of applying cloud mask processing to each of one or more second images and extracting second images showing the ground from the one or more second images, and in the generating step, inputting the second images showing the ground into an AI model.

[0137] <Appendix 3> A program described in Appendix 2, in which, in the step of extracting the second image, a third image of the target area taken from the air during a predetermined second period is obtained, and the second image that does not show the ground surface is replaced with the third image, and in the step of generating the third image, the third image is input into an AI model.

[0138] <Appendix 4> An information processing device comprising a control unit and a storage unit, wherein the control unit executes all steps in the program according to any one of (Supplementary Note 1) to (Supplementary Note 3).

[0139] <Appendix 5> A method executed by a computer having a processor and a memory, wherein the processor executes all steps in the program described in any one of (Appendix 1) to (Appendix 3).

[0140] <Appendix 6> A system comprising one or more processors that execute all steps in the program described in any one of (Appendix 1) to (Appendix 3). [Explanation of symbols]

[0141] 1. System 10...Terminal device 12...Communication IF 13...Input device 14...Output device 15...Memory 16…Storage 19...Processor 20...Server 22...Communication IF 23...Input / output IF 25…Memory 26…Storage 29...Processor 30…Artificial satellite 80…Network 90...Computer 901...processor 902…Main storage device 903…Auxiliary storage device 991…Communication IF

Claims

1. A program to be executed by a computer having a processor and a memory, The program causes the processor to: acquiring a plurality of first images of the target area taken from the air during a first period set to cover at least one cropping cycle of paddy fields present in the target area; a step of analyzing a time-series variation pattern of a vegetation index obtained from a plurality of the first images and generating identification information indicating the result of identifying the paddy field; acquiring one or more second images of the target area taken from the air during a second period set to include a period during which boundaries of sections present in the target area are identifiable; inputting one or more of the second images into a trained AI model and causing the AI ​​model to generate first polygons representing each of the plurality of sections; extracting, from among the plurality of first polygons based on the specified information, a polygon whose correspondence relationship between the first polygon and the paddy field area specified by the specified information satisfies a predetermined standard as a second polygon indicating the paddy field; A program that executes the following.

2. The processor further executes a step of applying a cloud mask process to each of the one or more second images to extract the second images showing the ground surface from the one or more second images; The program according to claim 1 , wherein in the generating step, the second image showing the ground surface is input to the AI ​​model.

3. In the step of extracting the second image, a third image of the target area photographed from the air during the predetermined second period is acquired, and the second image in which the ground surface is not captured is replaced with the third image; The program according to claim 2 , wherein in the generating step, the third image is input to the AI ​​model.

4. An information processing apparatus comprising a control unit and a storage unit, wherein the control unit executes all steps of the program according to any one of claims 1 to 3.

5. A method implemented on a computer having a processor and a memory, wherein the processor executes all the steps of the program of any one of claims 1 to 3.

6. A system comprising one or more processors that execute all steps in the program according to any one of claims 1 to 3.

Images