Agricultural work record creation device, agricultural work record creation program, and agricultural work record creation method

Abstract

To provide an agricultural work record creating device capable of easily recording agricultural work without using a special movement detection unit. [Solution] This agricultural work record creation device creates agricultural work records based on information from an information terminal and includes a movement trajectory acquisition unit 11 that acquires movement trajectories including date and time information from the information terminal, a feature extraction unit 12 that extracts work feature amounts based on the movement trajectory, a work type estimation unit 13 that estimates the work type from the work feature amounts, and an agricultural work record creation unit 14 that creates an agricultural work record by associating the movement trajectory with the work type. The work feature amounts include the speed of movement of parts that can be considered as straight lines extracted from the movement trajectory, the width which is the distance between the straight lines, crop information based on position information of the movement trajectory, and time period information based on the date and time the movement trajectory was recorded.

Description

[Technical Field]

[0001] The present invention relates to an agricultural work record creation device, an agricultural work record creation program, and an agricultural work record creation method. [Background technology]

[0002] Various methods have been used to record agricultural work. For example, Patent Document 1 describes a work recording device that includes: a behavior record data acquisition unit that acquires movement data related to the location and movement of a worker; a candidate field selection unit that selects a candidate field where the worker is working based on the location of the worker acquired by the behavior record data acquisition unit; a feature extraction unit that extracts features from the movement data of the worker acquired by the behavior record data acquisition unit; a work type identification unit that identifies a type of work performed by the worker using the features extracted by the feature extraction unit; a field identification unit that identifies a field where the worker worked from the candidate field selection unit based on the work type identified by the work type identification unit; and a work recording unit that associates the work type identified by the work type identification unit with the field identified by the field identification unit and records the associated work record. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-161991 Summary of the Invention [Problem to be solved by the invention]

[0004] In the case of Patent Document 1, to acquire movement data, a movement detection unit detects the movement of the farmworker using an acceleration sensor or the like, and notifies a storage unit of the detected data, along with the ID of the owner of the behavior recording device, as movement data, for recording. For example, the movement detection unit uses a triaxial acceleration sensor on the arm and waist. Examples of how the behavior recording device is worn (see Figure 2) include a waist-mounted behavior recording device worn on the farmworker's waist and an arm-mounted behavior recording device worn on the farmworker's arm. However, using a waist-mounted behavior recording device or an arm-mounted behavior recording device can cause problems when performing farm work, and there has been a strong demand for an easier way to record farm work.

[0005] The present invention has been made to solve the above-mentioned problems, and aims to provide an agricultural work record creation device, an agricultural work record creation program, and an agricultural work record creation method that can easily record agricultural work without using a special motion detection unit. [Means for solving the problem]

[0006] In order to achieve the above object, the agricultural work record creation device of the present invention is an agricultural work record creation device that creates an agricultural work record based on information from an information terminal, and includes: a movement trajectory acquisition unit that acquires a movement trajectory including date and time information from the information terminal; a feature amount extraction unit that extracts a work feature amount based on the movement trajectory; a work type estimation unit that estimates a work type from the work feature amount; and an agricultural work record creation unit that associates the movement trajectory with the work type to create an agricultural work record. The task feature amount includes the speed of movement of a portion that can be considered as a straight line extracted from the movement trajectory, a width that is the distance between the straight lines, crop information based on position information of the movement trajectory, and time information based on the recording date and time of the movement trajectory. Other aspects of the present invention will be described in the following embodiments. [Effects of the Invention]

[0007] According to the present invention, farm work can be easily recorded without using a special motion detection unit. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a diagram illustrating a farm work record creation device according to an embodiment of the present invention. [Figure 2] FIG. 2 is a diagram illustrating an example of a data structure of user information. [Figure 3] FIG. 2 is a diagram illustrating an example of a data structure of movement trajectory information. [Figure 4] FIG. 2 is a diagram illustrating an example of a data structure of farm field information. [Figure 5] FIG. 4 is a diagram illustrating an example of a data structure of task feature amount information. [Figure 6] FIG. 10 is a diagram illustrating an example of a data structure of work type information. [Figure 7] FIG. 4 is a diagram illustrating an example of a data structure of work record information. [Figure 8] 10 is a flowchart showing the processing of the agricultural work record creation device. [Figure 9] FIG. 10 is a diagram illustrating the processing of a task type estimation unit. [Figure 10] FIG. 10 is a diagram illustrating an outline of a method for calculating a working width and a working speed. [Figure 11] FIG. 10 is a diagram illustrating a method for determining whether a line segment intersects with another line segment; [Figure 12] FIG. 10 is a diagram illustrating a method for calculating a working width. [Figure 13] FIG. 2 is a diagram showing a home screen of the mobile terminal. [Figure 14] FIG. 10 is a diagram showing a calendar screen of a mobile terminal. [Figure 15] FIG. 10 is a diagram showing a map screen of a mobile terminal. [Figure 16] FIG. 10 is a diagram showing a work trajectory displayed on a mobile terminal in the first work example. [Figure 17] FIG. 10 is a diagram illustrating processing of a work trajectory by a feature amount extraction unit in the first work example. [Figure 18] FIG. 10 is a diagram showing work types displayed on a mobile terminal in the first work example. [Figure 19] FIG. 10 is a diagram showing a work trajectory displayed on a mobile terminal in the second work example. [Figure 20] FIG. 10 is a diagram illustrating processing of a work trajectory by the feature amount extraction unit in the second work example. [Figure 21] FIG. 10 is a diagram showing work types displayed on a mobile terminal in a second work example. [Figure 22]FIG. 10 is a diagram showing a work trajectory displayed on a mobile terminal in the third work example. [Figure 23] FIG. 11 is a diagram illustrating processing of a work trajectory by the feature amount extraction unit in the third work example. [Figure 24] FIG. 10 is a diagram showing the work types displayed on the mobile terminal in the third work example. [Figure 25] FIG. 10 is a diagram showing a work trajectory displayed on a mobile terminal in the fourth work example. [Figure 26] FIG. 13 is a diagram illustrating processing of a work trajectory by the feature amount extraction unit in the fourth work example. [Figure 27] FIG. 10 is a diagram showing work types displayed on a mobile terminal in a fourth work example. [Figure 28] FIG. 1 is a diagram illustrating an overview of a software framework. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. 1 is a diagram showing an agricultural work record creation device 100 according to an embodiment of the present invention. The agricultural work record creation device 100 is a web server and an agricultural work record creation device that creates agricultural work records based on information from a mobile terminal 200 (information terminal). The agricultural work record creation device 100 includes a processing unit 10 that processes the agricultural work record creation, a database 20 that stores data used when processing the agricultural work record creation, and a communication unit 50 that communicates via the Internet 300 with the mobile terminals 200 (information terminals) of users who perform agricultural work and users' administrators (contractors who use the agricultural work record creation device).

[0010] The database 20 is configured by a hard disk drive (HDD) device or the like. The processing unit 10 is realized by a central processing unit (CPU) expanding a program stored in a read only memory (ROM), HDD, or the like into a random access memory (RAM) and executing the program. The communication unit 50 exchanges various data and commands with the mobile terminal 200 or the like via the Internet 300. The processing of the processing unit 10 may be performed by a graphics processing unit (GPU).

[0011] The processing unit 10 has multiple programs, including a movement trajectory acquisition unit 11 that acquires a movement trajectory including date and time information from a mobile terminal 200 (information terminal), a feature extraction unit 12 that extracts work features based on the movement trajectory, a work type estimation unit 13 (see Figure 9) that estimates the work type from the work feature, an agricultural work record creation unit 14 that associates the movement trajectory with the work type to create an agricultural work record, a prediction model creation unit 15 used when the work type estimation unit 13 predicts the work type, and an input / output processing unit 16 that processes input and output with the information terminal.

[0012] The database 20 stores user information 21 (see FIG. 2) which is information about the contractor and its user, movement trajectory information 22 (see FIG. 3) which records the movement trajectory from the mobile terminal 200, field information 23 (see FIG. 4) which indicates the relationship between the field and the crop, work feature information 24 (see FIG. 5) which indicates the work feature extracted by the feature extraction unit 12, work type information 25 (see FIG. 6) which indicates the relationship between the crop and the work type, work record information 26 (see FIG. 7) which indicates the work record from the start to the end of the work, and a trained model 27 created by the prediction model creation unit 15. Details of each piece of information will be described later with reference to FIGS. 2 to 7.

[0013] The task features include the speed of movement of the straight line segments extracted from the trajectory, the width of the segments, which is the distance between the straight lines, crop information (crop name) based on the position information of the trajectory, and time information (task time) based on the recording date and time of the trajectory (see Figure 5). Details will be described later with reference to Figures 8, 10 to 12.

[0014] The mobile terminal 200 has functions such as notifying the start and end of transmission of location information, transmitting the date and time of agricultural work and location information based on signals from a positioning satellite 400 to the agricultural work record creation device 100, and viewing and modifying agricultural work records (such as work type and work trajectory). The mobile terminal 200 receives signals from a positioning satellite 400 that constitutes a satellite positioning system. The satellite positioning system is, for example, a Global Navigation Satellite System (GNSS).

[0015] The prediction model creation unit 15 uses past work records, i.e., past work feature quantities and work types, to learn through machine learning and deep learning and create a trained model 27 (machine learning model).The work type estimation unit 13 then uses the machine learning model learned from past work records to estimate the work type.The work feature quantities, such as the crop name, work time, work width, and work speed, are explanatory variables, and the work type, etc., are target variables.

[0016] The task type estimation unit 13 performs task type AI prediction (see FIG. 9, process S73). To do this, it is necessary to prepare a sufficient amount of training data (sets of task feature amounts generated from movement trajectory information and data on the task type at that time) in advance and train the AI. AI training is performed using movement trajectory information for the time periods at which tasks start and end, recognized during work record generation, and AI input data (task feature amounts, task type) generated from the task type for those time periods. AI training can be performed by batch processing in the AI ​​execution environment, and can also be performed at any timing. It is recommended to perform training approximately once a month using a task scheduler.

[0017] The prediction model creation unit 15 periodically checks whether task features corresponding to a work record exist, and if not, generates task features from movement trajectory information for the time period of the work record and registers them in the database. It periodically checks whether an AI task type prediction result has been set for the work record corresponding to the task feature, and for work records for which this has not been set, it performs an AI task type prediction using the AI ​​input data (task feature) and registers the result in the database. Since task type prediction only needs to be performed for those for which task feature values ​​have been generated, it is preferable to perform the generation of task feature values ​​and the prediction of task type asynchronously.

[0018] The farm work record creation device 100 of this embodiment is characterized by determining the type of work using a movement trajectory including date and time information from a mobile terminal 200 carried by the farmworker, and automatically creating a farm work record. Note that the device does not use a special motion detection unit such as a waist-mounted action recording device or arm-mounted action recording device as in Patent Document 1. Furthermore, farm work records can be created automatically without operating the mobile terminal 200 during farm work.

[0019] Next, each piece of information will be described with reference to FIGS. 2 is a diagram showing an example of the data structure of the user information 21. The user information 21 includes a contract ID, a contractor's name, a user ID under the contractor's control, a user name, and the like.

[0020] 3 is a diagram showing an example of the data structure of movement trajectory information 22. Movement trajectory information 22 includes work record FID (File ID), positioning date and time received from mobile terminal 200, longitude, latitude, speed, direction, etc. The user's trajectory can be understood from the data in each row.

[0021] FIG. 4 is a diagram showing an example of the data structure of the field information 23. The field information 23 includes the field name, crop name, location information, and time information. Since the crops grown in the same field may differ depending on the time of year, the time information is associated with the field information. The field location and crop can be identified using the work start date and time and location information from the mobile terminal 200. FIG. 4 shows that soybeans are grown in field 5.

[0022] FIG. 5 is a diagram showing an example of the data structure of the work feature information 24. The work feature information 24 includes the user ID, crop name, work period, work width, work speed, work record FID, etc. The work width and work speed are values ​​calculated by the feature extraction unit 12 based on the movement trajectory (work trajectory). When the work record FID is 4845, it can be seen that the user U0001 is working on soybeans with a work width of 3.5896 m and a work speed of 7.2401 km / h.

[0023] Fig. 6 is a diagram showing an example of the data structure of the work type information 25. The work type information 25 includes the crop name, work type, etc. In the example of Fig. 6, the work types for soybeans include plowing, sowing, pest control, fertilization, intertillage / weeding, ..., harvesting, and other types.

[0024] In agriculture, plowing refers to digging up and turning over the soil. Sowing refers to sowing seeds. Pest control refers to preventing and eliminating pests and diseases in farming. Fertilizing refers to applying fertilizer to improve crop growth and increase yields. Inter-row tillage refers to shallowly plowing the surface of the ridges or plowing the spaces between the ridges (pathways). Figure 6 shows an example of soybeans, but the database 20 also has registered information 25 on a variety of other types of work, such as paddy rice.

[0025] FIG. 7 is a diagram showing an example of the data structure of work record information 26. Work record information 26 includes a user ID, a field name, a crop name, a work start date and time, a work end date and time, a work type, a work record FID, and the like. Work record information 26 includes work record information 26A for cases where the work type has not been estimated, and work record information 26B for cases where the work type has been estimated or identified. According to work record information 26B in FIG. 7, when work record FID is 4845, it can be seen that user U0001 is working on soybeans in field 5, with the work start date and time being 2024 / 05 / 03 11:16:00, the work end date and time being 2024 / 05 / 03 13:01:00, and the work type being plowing.

[0026] Figure 28 is a diagram showing an overview of the software framework. The agricultural work record creation device 100 uses a three-tier system for online processing of agricultural work record creation. In a three-tier system, the entire system is divided into three tiers: P tier (presentation tier), F tier (function tier), and D tier (data tier), and is implemented on the web server side. By separating and arranging processing into multiple tiers, it is possible to flexibly respond when changes to a certain tier become necessary.

[0027] The P layer processes requests from the browser of the mobile terminal 200 and responds to the results. It processes requests from the browser and invokes logic in the F layer corresponding to the request (event). It also responds to the browser with the results from the F layer.

[0028] The F layer uses classes that are not web-dependent and can be executed as a library from outside the P layer. It is designed to have one logic for each request from the browser. The D layer sets the parameters received from the F layer in SQL, executes it, and then returns the results to the F layer.

[0029] Next, the detailed procedure of the farm work record creation process will be described with reference to FIG. Figure 8 is a flowchart showing the farm work record creation process S60. The description will be made with reference to Figure 1 as appropriate. When the mobile terminal 200 notifies the user that transmission of position information has ended (the timing when the user stops recording the positioning points), the farm work record creation unit 14 acquires movement trajectory information (step S61), recognizes one work section from the field information and work trajectory (i.e., determines the work start date and time and end date and time), and registers this in the work record information 26A (step S62). At this point, the work type has not been registered.

[0030] The start and end dates and times of work registered in the work record are prerequisite information for extracting work feature quantities. The start and end dates and times of work can be determined by the program (for example, by the date and time of entering and exiting the field frame) or by the user (for example, by the date and time of the work start and end actions). The program periodically checks for new work records and work feature quantity registrations, and when each registration is detected, it executes the next work feature quantity registration process and work type estimation process.

[0031] When the feature extraction unit 12 detects that a new work record has been registered, it extracts the work trajectory between the work start date and time and the work end date and time, extracts and calculates work feature amounts (crop name, work period, work width, work speed), and registers them in the work feature amount information 24 (see FIG. 5) (step S63: work feature amount extraction).

[0032] The extraction of task feature amounts (see task feature amount information 24 in FIG. 5) in step S63 is shown below. (a) The name of the crop is determined by comparing the position information (longitude, latitude) and work start date and time (positioning date and time) of the movement trajectory information 22 with the position information and time information of the field information 23 and determining whether they match. Note that if time information is not registered in the field information 23, the name of the crop is determined by comparing the position information (longitude, latitude) of the movement trajectory information 22 with the position information of the field information 23 and determining whether they match.

[0033] (b) The work time is determined from the work start date and time (positioning date and time) in the movement trajectory information 22. In this embodiment, the work period is divided into the early, middle, and late parts of each month, resulting in 36 divisions from early January to late December. The work period is converted into a work period code (numerical value) and becomes learning data for the AI. In reality, agricultural work cannot always be performed exactly at the same time and date as in the past, and if the granularity of the work period is set to date and time, the granularity is too fine for learning data, and slight differences in date and time are likely to affect the estimation results. For this reason, the granularity is set to early, middle, and late parts of the month. (c) The working width and working speed are calculated using the calculation method described later in Figure 10.

[0034] 8, when the work type estimation unit 13 detects that new work feature values ​​have been registered, it uses the trained model 27 to predict the work type using AI from the work feature value information 24 and work type information 25, which is a list (candidates) of work types for the crop (step S64: work type estimation). The work type estimation unit 13 estimates the work type using AI (DNN: Deep Neural Network) based on the work feature values.

[0035] The work type estimation unit 13 determines the work type based on the estimated probability and registers the result in the work record information 26 (step S65: determine work type). The registered result becomes work record information 26B (see FIG. 7).

[0036] The user can view the work record created by the system from the mobile terminal 200. If the prediction result has been registered, the predicted work type is displayed on the mobile terminal 200. If not, the predicted work type is displayed. The user can confirm and correct the displayed work record, and once they perform the confirmed operation, the work type is confirmed and the work record is complete (step S66: setting the work record result). The confirmed work type can be used as future training data.

[0037] FIG. 9 is a diagram showing the processing (process S70) of the task type estimation unit 13. The task type estimation unit 13 generates AI input data based on the task feature amount information 24 (process S71) and stores the AI ​​input data in a memory unit (process S72). Generating AI input data means converting it into code (quantified data) that can be processed by an AI. The task type estimation unit 13 then performs task type AI prediction based on the already trained model 27, task type information 25, and AI input data (process S73), and outputs the prediction result (process S74).

[0038] (Calculation method for working width and working speed) Figure 10 shows an overview of the calculation method for the working width and working speed. Fig. 10 shows the measured positioning points (circles) in the field and the detected regression line. The working width is considered to be the distance between adjacent straight line segments that can be considered parallel to each other on the working trajectory during reciprocal movement. The straight line portion of the work trajectory is detected as follows. (1) The positioning points are traced in chronological order, and the point just before the point where the azimuth difference from the starting point, which is considered to be a straight line, is taken as the final point, and the consecutive positioning points between them are extracted. (2) The positioning points in (1) are approximated by a regression line to obtain the slope and intercept, and the latitude (y coordinate) of the starting point and the end point of the line segment is calculated by multiplying the slope by the longitude (x coordinate) + the intercept.

[0039] The working width and working speed are defined as follows: (3) The detected line segments are extracted in chronological order, and for two line segments that are within a certain range of inclination error between the extracted line segment 1 and the subsequent line segment 2 (a range that can be considered parallel), a perpendicular line can be drawn from the midpoint of line segment 2 to line segment 1. (4) If the perpendicular line does not intersect with any other detected line segments, the length of the perpendicular line shall be the working width. However, if the working width exceeds a certain size, it shall not be used as the working width. In addition, the working speed of the line shall be the average speed of the positioning points included in (1).

[0040] Regarding (3), whether or not a perpendicular line can be drawn from a point on a line segment to another line segment is determined by whether or not the dot product of the two line segments is negative. Regarding (4), whether two line segments intersect can be determined by checking whether the cross product of the two line segments is negative. In other words, it is good to use the intersection test between line segments.

[0041] 11 is a diagram showing an example of determining whether or not the work width is acceptable. When the user moves in the order of line segment 1 → line segment 2 → line segment 3 → line segment 4, the dashed arrow is not considered to be part of the work width because it intersects with other line segments.

[0042] FIG. 12 is a diagram showing a method for calculating the working width. The working width is estimated to be several tens of meters, so we consider the working width as the distance between a point on a plane and a straight line. In this case, the working width can be calculated as follows: The distance between the line and point Q, which is the intersection point with the regression line n, is the working width. If the slope of the regression line n is a and the intercept is b, Point P(x on regression line n+1 n+1 , y n+1 ) and the point Q, which is the intersection with a line perpendicular to the regression line, is given by equation (1).

number

[0043] The coordinates of points P and Q are latitude and longitude on the Earth, so the working width is calculated using a highly accurate calculation method that takes this into account. For example, the distance between two points on Earth can be calculated as follows: If the radius of the Earth is r = 6378137.0 [m], the distance [m] between two points P and Q is given by equation (2).

number

[0044] (Example of display on a mobile device) 13 is a diagram showing the home screen of the mobile terminal 200. The display screen of the mobile terminal 200 is configured to include an upper menu section 201, a lower menu section 202, and a main display section 203. The upper menu section 201 has selection buttons such as all crops, all work, and all fields. The lower menu section 202 has a home button 202a, a calendar button 202b, a map button 202c, and a settings button 202d. The main display section 203 of the home screen of the mobile terminal 200 is configured to include a location information communication status 210, a work start button 211, a work end button 212 (work stop button), and a location information transmission restriction setting area 213.

[0045] When the user enters the field and starts work, he / she presses the work start button 211, and when he / she finishes work, he / she presses the work end button 212. This allows the mobile terminal 200 to transmit information on the movement trajectory including date and time information to the agricultural work record creation device 100.

[0046] 14 is a diagram showing the calendar screen of the mobile terminal 200. When the calendar button 202b in the lower menu section 202 is pressed, a calendar showing the types of work is displayed. For example, it can be seen that the work of "plowing" was performed on April 6th and the work of "sowing" was performed on April 18th.

[0047] 15 is a diagram showing the map screen of the mobile terminal 200. When the map button 202c in the lower menu section 202 is pressed, the field and movement trajectory (work trajectory) are displayed. The main display section 203 has a trajectory display button 230 and a zoom button 235, and the field is displayed as indicated by reference numeral 231. When the trajectory display button 230 is pressed, the movement trajectory is displayed on the field as indicated by reference numeral 232. This allows the user to check the movement trajectory on the field.

[0048] Next, a working example will be described. (First work example) FIG. 16 is a diagram showing the work trajectory displayed on the mobile terminal 200 in the first work example. FIG. 17 is a diagram showing the processing of the work trajectory by the feature amount extraction unit 12 in the first work example. FIG. 17 shows an example in which the feature amount extraction unit 12 calculates the work width and work speed based on data received from the mobile terminal 200. FIG. 17 shows the positioning points (◯) in the field and the detected regression line. As a result, the work width is calculated to be 3.59 m and the work speed is calculated to be 7.24 km / h.

[0049] The work type estimation unit 13 performs AI prediction of the work type based on the work feature amount, and calculates that plowing is 73.6%, sowing is 15.9%, pest control is 0.9%, fertilization is 4.8%, and inter-cultivation / weeding is 3.4%, etc. As a result, the work type is estimated to be "plowing". 18 is a diagram showing the work type displayed on the mobile terminal 200 in the first work example. The mobile terminal 200 displays the field name, crop name, work type, etc. The user checks the work type estimation result. If corrections are necessary, they can be made by pressing the pen mark.

[0050] (Second example of work) FIG. 19 is a diagram showing the work trajectory displayed on the mobile terminal 200 in the second work example. FIG. 20 is a diagram showing the processing of the work trajectory by the feature amount extraction unit 12 in the second work example. FIG. 20 shows an example in which the feature amount extraction unit 12 calculates the work width and work speed based on data received from the mobile terminal 200. FIG. 20 shows the field positioning points (◯) and the detected regression line. As a result, the work width is calculated to be 3.17 m and the work speed is calculated to be 4.15 km / h.

[0051] The work type estimation unit 13 performs AI prediction of the work type based on the work feature amount, and calculates that plowing is 10.4%, sowing is 66.7%, pest control is 11.3%, fertilization is 6.0%, and inter-cultivation / weeding is 3.5%, etc. As a result, the work type is estimated to be "sowing". 21 is a diagram showing the work type displayed on the mobile terminal 200 in the second work example. The mobile terminal 200 displays the field name, crop name, work type, etc. The user checks the work type estimation result. If corrections are necessary, they can be made by pressing the pen mark.

[0052] (Third work example) FIG. 22 is a diagram showing the work trajectory displayed on the mobile terminal 200 in the third work example. FIG. 23 is a diagram showing the processing of the work trajectory by the feature amount extraction unit 12 in the third work example. FIG. 23 shows an example in which the feature amount extraction unit 12 calculates the work width and work speed based on data received from the mobile terminal 200. FIG. 23 shows the positioning points (◯) in the field and the detected regression line. As a result, the work width is calculated to be 24.11 m and the work speed is calculated to be 4.54 km / h.

[0053] The work type estimation unit 13 performs AI prediction of the work type based on the work feature amount, and predicts that plowing is 1.3%, sowing is 1.0%, pest control is 94.0%, fertilization is 1.2%, and inter-cultivation / weeding is 1.0%. As a result, the type of work was estimated to be "pest control." 24 is a diagram showing the work type displayed on the mobile terminal 200 in the third work example. The mobile terminal 200 displays the field name, crop name, work type, etc. The user checks the work type estimation result. If corrections are necessary, they can be made by pressing the pen mark.

[0054] (Example 4) FIG. 25 is a diagram showing the work trajectory displayed on the mobile terminal 200 in the fourth work example. FIG. 26 is a diagram showing the processing of the work trajectory by the feature amount extraction unit 12 in the fourth work example. FIG. 26 shows an example in which the feature amount extraction unit 12 calculates the work width and work speed based on data received from the mobile terminal 200. FIG. 26 shows the positioning points (◯) in the field and the detected regression line. As a result, the work width is calculated to be 18.05 m and the work speed is calculated to be 4.39 km / h.

[0055] The work type estimation unit 13 performs AI prediction of the work type based on the work feature amount, and calculates that plowing is 0.9%, sowing is 0.7%, pest control is 95.9%, fertilization is 1.2%, and inter-cultivation / weeding is 0.5%, etc. As a result, the work type is estimated to be "pest control". 27 is a diagram showing the work type displayed on the mobile terminal 200 in the fourth work example. The mobile terminal 200 displays the field name, crop name, work type, etc. The user checks the work type estimation result. If corrections are necessary, they can be made by pressing the pen mark.

[0056] The farm work record creation device 100, farm work record creation program, and farm work record creation method of the present embodiment described above have the following features. (1) A farm work record creation device that creates a farm work record based on information from an information terminal (mobile terminal 200), comprising: a movement trajectory acquisition unit 11 that acquires a movement trajectory including date and time information from the information terminal; a feature extraction unit 12 that extracts a work feature based on the movement trajectory; a work type estimation unit 13 that estimates a work type from the work feature; and a farm work record creation unit 14 that associates the movement trajectory with the work type to create a farm work record. This allows for easy farm work recording without using a special motion detection unit. In this embodiment, the farm work record is created based on information from the information terminal, so the device may be carried by the user or may be installed on a tractor, etc.

[0057] (2) In (1) above, the agricultural work record creation device includes the work feature values ​​extracted from the movement trajectory, including the speed of movement of the portion that can be considered as a straight line, the width that is the distance between the straight lines, crop information based on the position information of the movement trajectory, and time information based on the date and time when the movement trajectory was recorded.

[0058] (3) In (1) above, the work type estimation unit 13 is a farm work record creation device that uses a machine learning model learned from past work records to estimate the work type.

[0059] (4) An agricultural work record creation program that creates an agricultural work record based on information from an information terminal, which causes a computer to execute a movement trajectory acquisition process (mainly step S61) that acquires a movement trajectory including date and time information from the information terminal, a feature extraction process (step S63) that extracts work features based on the movement trajectory, a work type estimation process (steps S64, S65) that estimates the work type from the work features, and an agricultural work record creation process (mainly step S66) that associates the movement trajectory with the work type to create an agricultural work record.

[0060] (5) In (4) above, the agricultural work record creation program includes the work features extracted from the movement trajectory, including the speed of movement of the portion that can be considered a straight line, the width that is the distance between the straight lines, crop information based on the position information of the movement trajectory, and time information based on the date and time of recording of the movement trajectory.

[0061] (6) In (4) above, the work type estimation process is an agricultural work record creation program that uses a machine learning model learned from past work records to estimate the work type.

[0062] (7) A method for creating an agricultural work record of an agricultural work record creation device that creates an agricultural work record based on information from an information terminal, wherein the processing unit 10 of the agricultural work record creation device executes a movement trajectory acquisition process that acquires a movement trajectory including date and time information from the information terminal, a feature extraction process that extracts work features based on the movement trajectory, a work type estimation process that estimates a work type from the work features, and a farm work record creation process that associates the movement trajectory with the work type to create an agricultural work record. [Explanation of symbols]

[0063] 10 Processing section 11 Movement trajectory acquisition section 12 Feature extraction unit 13 Work type estimation unit 14. Agricultural Work Recording Department 15 Prediction Model Creation Department 16 Input / output processing section 20 databases 21 User Information 22 Movement trajectory information 23 Field Information 24 Work feature information 25 Work type information 26 Work Record Information 27 Pre-trained Models 50 Communications Department 100 Agricultural work record creation device 200 Mobile terminals (information terminals) 300 Internet 400 positioning satellites

Claims

1. An agricultural work record creation device that creates an agricultural work record based on information from an information terminal, a movement trajectory acquisition unit that acquires a movement trajectory including date and time information from the information terminal; a feature extraction unit that extracts task feature amounts based on the movement trajectory; a task type estimation unit that estimates a task type from the task feature amount; a farm work record creation unit that creates a farm work record by associating the movement trajectory with the work type, The task feature amount includes: The speed of movement of a portion that can be considered as a straight line extracted from the movement trajectory, and the width that is the distance between the straight lines, Crop information based on position information of the movement trajectory; Including time information based on the recording date and time of the movement trajectory A farm work record creation device characterized by:

2. The task type estimation unit estimates the task type using a machine learning model learned from past task records.

2. The agricultural work record creation device according to claim 1.

3. A farm work record creation program for creating a farm work record based on information from an information terminal, On the computer, a movement trajectory acquisition process for acquiring a movement trajectory including date and time information from the information terminal; a feature extraction process for extracting task feature amounts based on the movement trajectory; a task type estimation process for estimating a task type from the task feature amount; and a farm work record creation process for creating a farm work record by associating the movement trajectory with the work type, The task feature amount includes: The speed of movement of a portion that can be considered as a straight line extracted from the movement trajectory, and the width that is the distance between the straight lines, Crop information based on position information of the movement trajectory; Including time information based on the recording date and time of the movement trajectory A farm work record creation program characterized by:

4. The work type estimation process uses a machine learning model learned from past work records to estimate the work type.

4. The agricultural work record creation program according to claim 3.

5. A farm work record creation method for a farm work record creation device that creates a farm work record based on information from an information terminal, comprising: The processing unit of the agricultural work record creation device a movement trajectory acquisition process for acquiring a movement trajectory including date and time information from the information terminal; a feature extraction process for extracting task feature amounts based on the movement trajectory; a task type estimation process for estimating a task type from the task feature amount; a farm work record creation process for creating a farm work record by associating the movement trajectory with the work type; The task feature amount includes: The speed of movement of a portion that can be considered as a straight line extracted from the movement trajectory, and the width that is the distance between the straight lines, Crop information based on position information of the movement trajectory; Including time information based on the recording date and time of the movement trajectory A method for creating agricultural work records.

Images