Method for predicting optimum corn harvesting stage
The method employs a color tone change model for corn tassels and cumulative temperature calculations to accurately predict the silk thread extraction date, addressing inefficiencies in existing harvest time prediction methods and enhancing harvest time determination for sweet corn.
Patent Information
- Application Number
- JP2023097434
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-06-14
- Publication Date
- 2024-12-26
- Estimated Expiration
- 2043-06-14
AI Technical Summary
Existing methods for predicting corn harvest time are inefficient and inaccurate, particularly for sweet corn, leading to repeated field visits and difficulty in determining the appropriate harvest time due to reliance on small samples.
A method using a flowering stage estimation model, based on a color tone change model of corn tassels, to predict the silk thread extraction date, combined with cumulative temperature calculations, to accurately determine the appropriate harvest time.
Enables precise prediction of the corn harvest time, reducing the need for repeated field visits and improving the accuracy of harvest timing decisions.
Smart Images

Figure 2024178980000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to a method for predicting the optimum harvest time of corn. [Background technology]
[0002] Sweet corn is a highly profitable crop, but it is also known that its quality deteriorates significantly after harvest, so planned harvesting is desirable. Currently, producers touch the ears with their hands just before harvesting and decide whether to harvest them or not based on their swelling. However, this method requires farmers to visit the field multiple times during the harvest season, which hinders efficient cultivation. In addition, farmers are forced to judge the optimum harvest time from a small sample of a vast field, making it difficult to accurately judge the optimum harvest time and to plan harvesting. Incidentally, Patent Document 1 proposes a method for predicting corn yield early and with high accuracy. In addition, Patent Document 2 proposes a device that estimates the harvest time when agricultural crops will have the best taste, and Patent Document 3 proposes an information processing device, a harvest time prediction program, and a harvest time prediction method that can predict the harvest time of cultivated crops with greater accuracy. [Prior art documents] [Patent documents]
[0003] [Patent Document 1] Patent Publication No. 2022-64327 [Patent Document 2] JP 2016-19512 A [Patent Document 3] JP 2013-42668 A Summary of the Invention [Problem to be solved by the invention]
[0004] Patent Document 1 is not intended to predict the optimum harvest time for corn, and Patent Documents 2 and 3 are not suitable for predicting the optimum harvest time for corn.
[0005] Therefore, an object of the present invention is to provide a method for predicting the optimum harvest time for corn, which can accurately predict the optimum harvest time by calculating the predicted silk extraction date from image data of corn during growth. [Means for solving the problem]
[0006] The method for predicting the optimum harvest time of corn according to the present invention as described in claim 1 is a method for predicting the optimum harvest time of corn using a flowering stage estimation model machine-learned by teacher data in which flowering stage information of corn tassels is added to learning image data of corn tassels, the flowering stage estimation model being a color change model in the center of the tassels, and the method includes a data acquisition step 10 in which a computer acquires growing corn image data in which the corn is imaged during growth, and a step of acquiring the color change model of the corn tassels imaged in the growing corn image data acquired in the data acquisition step 10. The method is characterized in that it executes a determination step 20 in which, for each growing tassel, the flowering stage estimation model is used to determine whether the tassel is in at least three stages, i.e., unflowered, early flowering, and late flowering, a silking extraction date prediction step 30 in which a predicted silk extraction date is calculated using the number of unflowered tassels of the growing tassels determined as unflowered, the number of early flowering tassels of the growing tassels determined as early flowering, and the number of late flowering tassels of the growing tassels determined as late flowering, and a suitable harvest time prediction step 40 in which the suitable harvest time is predicted using an accumulated temperature with the predicted silk extraction date calculated in the silking extraction date prediction step 30 as the starting date. The present invention described in claim 2 is characterized in that, in the method for predicting the optimum harvest time for corn described in claim 1, the central part of the tassel is yellow in the early flowering stage and brown in the late flowering stage. The present invention as described in claim 3 is characterized in that, in the method for predicting the optimum harvest time for corn as described in claim 1, a flowering status model based on the ratio of flowered and unflowered parts for one tassel is used as the flowering stage estimation model, together with the color change model, and in the judgment step 20, the stage is judged to be four or more stages, and in the silk extraction date prediction step 30, the predicted silk extraction date is calculated using the number of tassels judged to be in four or more stages. The present invention as described in claim 4 is characterized in that, in the method for predicting the optimum harvest time of corn as described in claim 1, in the optimum harvest time predicting step 40, the accumulated temperature is calculated using the predicted air temperature after the predicted silk extraction date. The present invention as set forth in claim 5 is the method for predicting the optimum corn harvest time as set forth in claim 1, characterized in that in the optimum harvest time predicting step 40, the accumulated temperature is set according to the variety of the corn currently growing. The present invention as described in claim 6 is characterized in that, in the method for predicting the optimum harvest time for corn as described in any one of claims 1 to 5, the image data of the growing corn is imaged from above the corn during growth. Effect of the Invention
[0007] According to the method for predicting the optimum harvest time of corn of the present invention, a model of color change in the center of the tassel can be used to calculate the predicted silk extraction date, thereby making it possible to predict the optimum harvest time. [Brief description of the drawings]
[0008] [Figure 1] FIG. 2 is a flow chart showing the data acquisition step and the judgment step in the method for predicting the optimum harvest time of corn according to an embodiment of the present invention. [Diagram 2] A flow chart showing the steps of predicting the silk extraction date in the method for predicting the optimum harvest time of corn. [Diagram 3] FIG. 2 is a flow chart showing a step of predicting the optimum harvest time and an output step in the method for predicting the optimum harvest time of corn. [Figure 4]Demonstration photos explaining predictions by the color change model of this embodiment [Diagram 5] A diagram showing a demonstration example of the method for predicting the optimum harvest time for corn [Figure 6] A photo explaining the flowering status model used to estimate the flowering stage in the method for predicting the optimum harvest time for the same corn. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0009] The method for predicting the optimum harvest time of corn according to the first embodiment of the present invention includes a data acquisition step in which the flowering stage estimation model is a color change model at the center of the tassel, and a computer executes the following steps: a data acquisition step in which the computer acquires growing corn image data obtained by imaging the growing corn; a determination step in which the computer uses the flowering stage estimation model to determine at least three stages, i.e., not yet flowering, early flowering, and late flowering, for each growing tassel imaged in the growing corn image data acquired in the data acquisition step; a silking extraction date prediction step in which the computer calculates a predicted silking date using the number of not yet flowering tassels of growing tassels determined as not yet flowering, the number of early flowering tassels of growing tassels determined as early flowering, and the number of late flowering tassels of growing tassels determined as late flowering; and a suitable harvest time prediction step in which the computer predicts the optimum harvest time using the accumulated temperature with the predicted silking date calculated in the silking extraction date prediction step as the starting date. According to this embodiment, the computer can calculate the predicted silking date by using the color change model at the center of the tassel, and therefore can predict the optimum harvest time.
[0010] The second embodiment of the present invention uses the fact that the center of the tassel is yellow in the early flowering stage and brown in the late flowering stage in the method for predicting the optimum harvest time of corn according to the first embodiment. According to this embodiment, the flowering stage can be determined with high accuracy by focusing on the color change in the center of the tassel.
[0011] A third embodiment of the present invention uses a flowering state model based on the ratio of flowered and unflowered parts for one tassel together with a color change model as a flowering stage estimation model in the method of predicting the optimum harvest time for corn according to the first embodiment, and in the determination step, the flowering stage is determined to be at four or more stages, and in the silking date prediction step, the predicted silking date is calculated using the number of tassels determined to be at four or more stages. According to this embodiment, by using a flowering state model together with a color change model as a flowering stage estimation model, it is possible to determine to be at four or more stages, and to calculate a predicted silking date with even greater accuracy.
[0012] A fourth embodiment of the present invention is a method for predicting the optimum harvest time of corn according to the first embodiment, in which the optimum harvest time predicting step calculates the accumulated temperature using the predicted air temperature after the predicted silking date. According to this embodiment, the optimum harvest time can be predicted at an early stage by calculating the accumulated temperature using the predicted air temperature. Furthermore, if the predicted silking date has passed at the time of predicting the optimum harvest time, the prediction error of the air temperature in the accumulated temperature can be reduced by using the actual air temperature up to the time of predicting the optimum harvest time, and the optimum harvest time can be predicted more accurately.
[0013] A fifth embodiment of the present invention is a method for predicting the optimum harvest time of corn according to the first embodiment, in which the optimum harvest time predicting step uses an accumulated temperature according to the variety of corn being grown. According to this embodiment, by using an accumulated temperature according to the variety of corn, it is possible to predict the optimum harvest time with even greater accuracy.
[0014] A sixth embodiment of the present invention is a method for predicting an optimum corn harvest time according to any one of the first to fifth embodiments, in which the growing corn image data is obtained by capturing an image of the growing corn from above. According to this embodiment, capturing an image of the growing corn from above makes it easier to capture an image of the center of the tassel. EXAMPLES
[0015] A method for predicting the optimum harvest time of corn according to one embodiment of the present invention will be described below. The method for predicting the optimum corn harvest time according to this embodiment predicts the optimum corn harvest time using a flowering stage estimation model trained by machine learning using training data in which training image data of corn tassels has been added with information on the flowering stage of the tassels, and a computer executes a data acquisition step, a judgment step, a silk extraction date prediction step, a optimum harvest time prediction step, and an output step. In this embodiment, sweet corn is used.
[0016] FIG. 1 is a flow diagram illustrating the data acquisition steps and the judgment steps in the method for predicting the optimum harvest time of corn according to this embodiment. In the data acquisition step 10, growing corn image data is acquired by capturing an image of the growing corn. The growing corn image data is obtained by capturing an image of the growing corn from above. By capturing an image of the growing corn from above in this manner, it is easy to capture an image of the center of the tassel. A drone is suitable for taking images from the sky, and by taking uniform aerial photographs of the field using a drone, it is possible to obtain images of the growing corn throughout the field. The optimal flight speed for a drone is 1-4 m / s, and the optimal shooting altitude is 3-7 m.
[0017] In the determination step 20, each growing tassel captured in the growing corn image data acquired in the data acquisition step 10 is determined to be in at least three stages: unflowered, early flowering, and late flowering, using a flowering stage estimation model. By using AI to detect developing tassels at multiple stages according to the growth conditions, judgments can be made accurately, quickly and easily. The AI flowering stage estimation model in this embodiment is a model of color change in the center of the tassel. The color change model, called the "Coloring-model," was created based on classification criteria that focused on the color and shape of the tassels. Among the growing tassels in the aerial photographs, those that have not yet flowered are defined as "non-flowering (NF)," those that have flowered and have a yellow center are defined as "early stage of flowering (EF)," and those that have flowered and have a brown center are defined as "late stage of flowering (LF)." As growth progresses, the number of unflowered tassels detected during growth decreases while the number of pre-flowering tassels increases, and as growth continues, the number of pre-flowering tassels decreases while the number of post-flowering tassels increases.
[0018] FIG. 2 is a flow diagram illustrating the silking date predicting step in the method for predicting the optimum corn harvest time according to this embodiment. In the silk extraction date prediction step 30, the predicted silk extraction date is calculated using the number of unflowered tassels of growing tassels determined to be unflowered, the number of early flowering tassels of growing tassels determined to be in the early flowering stage, and the number of late flowering tassels of growing tassels determined to be in the late flowering stage. The silk extraction date prediction step 30 comprises a step 31 of calculating the number of tassels at each stage, and a prediction step 32 . In step 31 of calculating the number of tassels at each stage, the number of unflowered tassels of growing tassels determined to be unflowered, the number of early-flowering tassels of growing tassels determined to be in the early-flowering stage, and the number of late-flowering tassels of growing tassels determined to be in the late-flowering stage are calculated.
[0019] In FIG. 2, the calculation results in step 31 for calculating the number of tassels at each stage show that, for example, the number of unflowered tassels is 157, the number of early flowering tassels is 940, and the number of late flowering tassels is 304. In the prediction step 32, the silk extraction date is predicted using the number of unflowered tassels, the number of early flowering tassels, and the number of late flowering tassels calculated in the tassel number calculation step 31 at each stage, as well as the image capture date. For example, a linear regression model or a multiple regression model can be used to predict the silk extraction date. NF is an unflowered tassel, and LF is a late-flowered tassel. The calculation result in the prediction step 32 is output as, for example, the predicted silk extraction date of July 23rd.
[0020] FIG. 3 is a flow chart illustrating a step of predicting an optimum harvest time and an output step in the method of predicting an optimum harvest time for corn according to this embodiment. In the optimum harvest time predicting step 40, the optimum harvest time is predicted using the integrated temperature, starting from the predicted silk extraction date calculated in the silk extraction date predicting step 30. The optimum harvest time predicting step 40 includes a step 41 of calculating an accumulated temperature starting from the predicted silk extraction date, a variety discrimination step 42, and a predicted optimum harvest time calculating step 43, 44 using the accumulated temperature according to the variety. In Fig. 3, an example is shown in which variety A in the predicted optimum harvest time calculating step 43 is Memi Star, and variety B in the predicted optimum harvest time calculating step 44 is Honey Bantam (registered trademark) 20. Other varieties of sweet corn include Gold Rush (registered trademark), Pure White (registered trademark), and Peter Corn. In step 41, which calculates the accumulated temperature with the predicted silk extraction date as the starting date, predicted weather data is used. The predicted weather data can be, for example, data provided by the Japan Meteorological Agency, the Agro-Meteorological Grid Square Data (NARO) provided by the National Agriculture and Food Research Organization (NARO), or data provided by other organizations. When using the predicted weather data, the location name and latitude and longitude information that identify the field are used. In the optimum harvest time prediction step 40, the optimum harvest time can be predicted at an early stage by calculating the accumulated temperature using the predicted air temperature after the predicted silk extraction date. Also, if the predicted silk extraction date has passed at the time of predicting the optimum harvest time, the prediction error of the air temperature in the accumulated temperature can be reduced by using the actual air temperature up to the time of predicting the optimum harvest time, and the optimum harvest time can be predicted more accurately. Furthermore, in the optimum harvest time predicting step 40, by using an accumulated temperature according to the variety of corn being grown, it is possible to predict the optimum harvest time with even greater accuracy. In the output step 50, the optimum harvest time predicted in the optimum harvest time prediction step 40 is output. As for the optimum harvest time, it is preferable to output the start date and end date of the predicted optimum harvest time.
[0021] FIG. 4 is a photograph illustrating the predictions made by the color change model of this embodiment. FIG. 4(a) shows a tassel and a female ear, and FIG. 4(b) shows the change in the tassel over a given period of time. As shown in Figure 4(a), the tassel is located at the top of the plant body, making it suitable for aerial photography. Figure 4(b) shows images of tassels during growth from July 16 to July 28. As growth progresses, the number of flowering tassels increases, and the center of the tassel changes from yellow in the early flowering stage to brown in the later flowering stage. In this way, by focusing on the color change in the center of the tassel, the flowering stage can be determined with high accuracy. Then, by classifying the tassels into three stages using a color change model in the center of the tassels, the predicted silk extraction date can be calculated using a linear regression model, and the optimum harvest time can be predicted.
[0022] FIG. 5 shows a demonstration example of the method for predicting the optimum harvest time of corn according to this embodiment. FIG. 5 shows an overlap between the optimum corn harvest time predicted by the method for predicting the optimum corn harvest time according to this embodiment and the actual harvest time, and the overlap indicates that the prediction is correct. Aerial photography was conducted using a drone (Mavic Air2 or Mavic 2 Pro, DJI) at two sweet corn fields (variety: Emi Star or Honey Bantam (registered trademark) 20) in verification sites A and B. From the tassel extraction period to the silk extraction period from early July to early August, sweet corn tassels were photographed by pointing the RGB camera attached to the drone directly downward at an altitude of approximately 7 m. The captured images were 3840 x 2160 pixels, and the center of the images was then cropped to 1920 x 1080 pixels. After cropping, each image contained approximately 10 to 20 tassels. Images in which the tassels were unclear due to overexposure or wind influence were deleted. As shown in Figure 5, the overlap between the two verification sites, A and B, exceeded 84%.
[0023] FIG. 6 is a photograph for explaining a flowering state model serving as a flowering stage estimation model in the method for predicting the optimum harvest time of corn according to this embodiment. The flowering status model is a classification standard called "Flowering-mode" that focuses on the flowering status of tassels. Among the growing tassels in the aerial photographs, those that do not show any flowering are defined as "non-flowering (NF)," those that show flowering but also have some unflowered parts are defined as "partially flowering (PF)," and those that show flowering in all parts of the growing tassel are defined as "fully flowering (FF)." Figure 6(a) shows the unflowered state, Figure 6(b) shows the partially flowered state, and Figure 6(c) shows the fully flowered state.
[0024] As shown in FIG. 6, a flowering status model based on the ratio of flowering parts to unflowered parts for one tassel can also be used. In particular, by using such a flowering status model together with a color change model, the judgment step 20 can judge the flowering status into four or more stages, and in the silking date prediction step 30, the predicted silking date can be calculated using the number of tassels judged to be four or more stages. In this way, by using the flowering status model together with the color change model as a flowering stage estimation model, it is possible to determine four or more stages and calculate the predicted silk extraction date with even greater accuracy. [Industrial Applicability]
[0025] The present invention is particularly suited to sweet corn, but is applicable to popcorn, dent corn, flint corn, waxy corn, soft corn, pod corn, and other types of corn. [Explanation of symbols]
[0026] 10 Data Acquisition Steps 20 Judgment Step 30 Silk Extraction Date Prediction Steps 31 Steps for calculating the number of tassels at each stage 32 Prediction Steps 40 Steps for predicting optimal harvest time 41 Step to calculate accumulated temperature 42 Variety Identification Steps 43, 44 Calculation step for predicted harvest time 50 Output Steps
Claims
1. A method for predicting an optimum harvest time of corn, comprising: using a flowering stage estimation model machine-learned using training data obtained by adding flowering stage information of corn tassels to learning image data obtained by capturing an image of the tassels, the method comprising: the flowering stage estimation model is a color change model at the center of the tassel, The computer A data acquisition step of acquiring growing corn image data by capturing an image of the growing corn; a determination step of determining, using the flowering stage estimation model, at least three stages of not yet flowered, early flowering, and late flowering for each growing tassel captured in the growing corn image data acquired in the data acquisition step; a silking extraction date predicting step of calculating a predicted silking extraction date using the number of unflowered tassels of the growing tassels determined as unflowered, the number of early flowering tassels of the growing tassels determined as early flowering, and the number of late flowering tassels of the growing tassels determined as late flowering; a harvesting optimum time prediction step of predicting the harvesting optimum time using an integrated temperature with the predicted silk extraction date calculated in the silk extraction date prediction step as a starting date; Run A method for predicting the optimum harvest time of corn, comprising:
2. In the pre-flowering stage, the central portion of the tassel is yellow, and in the post-flowering stage, the central portion of the tassel is brown. The method for predicting the optimum harvest time of corn according to claim 1 .
3. As the flowering stage estimation model, a flowering status model based on a ratio of a flowering part to an unflowered part for one of the tassels is used together with the color change model, In the judgment step, the judgment is made in four or more stages, In the silking date prediction step, the predicted silking date is calculated using the number of tassels determined to be in four or more stages. The method for predicting the optimum harvest time of corn according to claim 1 .
4. In the step of predicting the optimum harvest time, The integrated temperature is calculated using the predicted air temperature after the predicted silk extraction date. The method for predicting the optimum harvest time of corn according to claim 1 .
5. In the step of predicting the optimum harvest time, The accumulated temperature is determined according to the variety of the corn being grown. The method for predicting the optimum harvest time of corn according to claim 1 .
6. The growing corn image data is an image of the growing corn taken from above. The method for predicting the optimum harvest time of corn according to any one of claims 1 to 5.
Citation Information
Patent Citations
Crop type classification in images
JP2022084655A
Continual crop development profiling using dynamical extended range weather forecasting with routine remotely-sensed validation imagery
WO2016028612A1
Information processing device, harvest time-predicting program and harvest time-predicting method
JP2013042668A
Harvesting season estimating device of crop, and harvesting season estimating program of crop
JP2016019512A
Maize yield prediction method
JP2022064327A
