Pollen amount estimation system, pollen amount estimation method, and pollen amount estimation program

Abstract

To increase the amount of pollen collected and improve the efficiency of pollen collection by a technique that replaces conventional visual discrimination, specifically by using AI based on machine learning to classify flowering stages from image data of flowers blooming on multiple branches or trees and to estimate the total amount of pollen, thereby improving the efficiency of pollen collection through image analysis.SOLUTION: A pollen amount estimation system includes an image acquisition unit that acquires an image of a flower, an image analysis unit that analyzes the state of the flower from the acquired image, a pollen collection amount data acquisition unit that acquires a predetermined image and pollen collection amount, and a pollen collection amount calculation unit that calculates a pollen collection amount based on the state of the flower analyzed by the image analysis unit and the predetermined image and pollen collection amount data acquired by the pollen collection amount data acquisition unit.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a pollen amount estimation system, a pollen amount estimation method, and a pollen amount estimation program.

Background Art

[0002] Pears require cross-pollination, that is, pollen from other varieties is required for pollination. Although honeybees may be used for this cross-pollination, stable results cannot be obtained because honeybees are inactivated by temperature and weather. For this reason, artificial pollination is necessary, but the procurement of pollen depends on imports and is costly. In addition, in the case of imports, import suspension may occur when important diseases occur. Currently, artificial pollination is the mainstream, but it is necessary to improve the pollen collection efficiency and strengthen the domestic supply-demand system. For example, in the case of pears, the pollen collection efficiency is increased by harvesting flowers all at once at the five-petal stage. In the case of kiwis, the appropriate harvesting period for flower buds is immediately before and after flowering. When the flower buds are immature or fully open, which is before and after the appropriate harvesting period, the pollen yield decreases. In addition, the pollen germination rate of immature flower buds is extremely low. The appropriate harvesting period for flower buds is immediately before and after flowering, when the buds are swollen like a balloon and the petal color is white. Immature flower buds with green remaining on the petals or fully opened flowers have a reduced amount of pure pollen that can be harvested. The pollen germination rate of flower buds at the appropriate harvesting period is around 70%, while that of immature flower buds is extremely low, at 20% or less. Patent Document 1 describes a self-propelled flower bud and medicinal herb harvesting machine and a flower bud and medicinal herb harvesting method using the same. Patent Document 2 describes a method for artificial pollination and an apparatus for performing the same. Document 3 describes a flower bud harvesting machine. In addition, Patent Document 3 describes an agricultural support system including a model generation unit that generates a three-dimensional model of a plant by combining predetermined generation conditions, and describes the operation in the pollination work, but there is no description or suggestion regarding the pollen collection amount and its discrimination. Conventionally, since the appropriate harvesting period of flower buds has been determined visually, experience has been required.

Prior Art Documents

Patent Documents

[0003] [Patent Document 1] Japanese Patent Application No. 2021-088738 [Patent Document 2] Japanese Patent Publication No. 2023-520743 [Patent Document 3] Japanese Unexamined Patent Application Publication No. 2019-041720 [Patent Document 4] Japanese Unexamined Patent Application Publication No. 2023-055100 [Summary of the Invention] [Problems to be Solved by the Invention]

[0004] An object of the present invention is to increase the pollen collection amount and improve the pollen collection efficiency instead of the conventional visual discrimination. Another object of the present invention is to improve the pollen collection efficiency by using image analysis, such as classifying the flowering stage from the image data of flowers blooming on a plurality of branches or trees using AI by machine learning and estimating the total amount of pollen. [Means for Solving the Problems]

[0005] The pollen amount estimation system according to claim 1 of the present invention includes an image acquisition unit that acquires an image of a flower, an image analysis unit that analyzes the state of the flower from the acquired image, a pollen collection amount data acquisition unit that acquires a given image and the pollen collection amount in advance, and a pollen collection amount calculation unit that calculates the pollen collection amount from the state of the flower analyzed by the image analysis unit and the given image and pollen collection amount data acquired by the pollen collection amount data acquisition unit. It is a pollen amount estimation system provided with The pollen amount estimation system according to claim 2 of the present invention wherein the image analysis unit includes a stage classification unit that classifies the flowering stage of the flower into three or more stages from the acquired image. The pollen collection amount data acquisition unit includes a stage-by-stage pollen amount data acquisition unit that acquires the pollen collection amount at each stage, and is the pollen amount estimation system according to claim 1. The pollen amount estimation system according to claim 3 of the present invention is The pollen amount estimation system according to claim 2, wherein the three or more stages are four stages of the immature stage, the balloon stage, the flowering stage, and the post-flowering stage. The pollen amount estimation system according to claim 4 of the present invention is The image analysis unit cuts out a square grid cell from the image data, detects an object based on the feature amount for each grid cell, and estimates the type of the object, and is the pollen amount estimation system according to claim 2. The pollen amount estimation system according to claim 5 of the present invention is The image analysis unit and the pollen collection amount calculation unit are integrally formed as an artificial intelligence unit, and is the pollen amount estimation system according to claim 2. The pollen amount estimation system according to claim 6 of the present invention is that the image analysis unit acquires a color image, and is the pollen amount estimation system according to claim 2. The pollen amount estimation system according to claim 7 of the present invention is that the flower is a flower of a fruit tree, and is the pollen amount estimation system according to claim 6. The pollen amount estimation system according to claim 8 of the present invention is that the flower is an entomophilous flower, and the image is an image using light of a plurality of wavelengths among the wavelengths acquired by at least one insect corresponding to the entomophilous flower, and is the pollen amount estimation system according to claim 2. The pollen amount estimation system according to claim 9 of the present invention is provided with a mobile terminal communication unit that communicates with a mobile terminal, the image acquisition unit acquires an image captured by the mobile terminal, the pollen collection amount calculation unit transmits the pollen collection amount to the mobile terminal, and is the pollen amount estimation system according to any one of claims 1 to 9. The pollen amount estimation system according to claim 10 of the present invention is The pollen amount estimation system according to any one of claims 1 to 9, comprising an imaging device communication unit that communicates with an installed imaging device, and an image acquisition unit that acquires an image captured by the installed imaging device. The pollen amount estimation system according to claim 11 of the present invention comprises a flower bud collection machine communication unit that communicates with a flower bud collection machine that collects flower buds, and is the pollen amount estimation system according to any one of claims 1 to 9, which transmits a signal based on the estimated pollen amount to the flower bud collection machine. The pollen amount estimation method according to claim 12 of the present invention includes an image acquisition step of acquiring an image of a flower, an image analysis step of analyzing the state of the flower from the acquired image, a pollen collection amount data acquisition step of acquiring pre-given image and pollen collection amount data, and a pollen collection amount calculation step that is executed after the image analysis step and the pollen collection amount data acquisition step, and calculates the pollen collection amount from the state of the flower analyzed in the image analysis step and the pre-given image and pollen collection amount acquired in the pollen collection amount data acquisition step. It is a pollen amount estimation method having the above steps. The pollen amount estimation method according to claim 13 of the present invention wherein the image analysis step includes a step classification step of classifying the flowering stage of the flower from the acquired image into three or more stages, and the pollen collection amount data acquisition step includes a step-by-step pollen amount data acquisition step of acquiring the pollen collection amount at each stage, and is the pollen amount estimation method according to claim 12. The pollen amount estimation method according to claim 14 of the present invention The three or more stages are four stages: the immature stage, the balloon stage, the flowering stage, and the post-flowering stage, and it is the pollen amount estimation method according to claim 13. The pollen amount estimation method according to claim 15 of the present invention wherein the image analysis step classifies the flowering stage of each flower using the analysis result in the artificial intelligence unit, and it is the pollen amount estimation method according to claim 13. The pollen amount estimation method according to claim 16 of the present invention is The image analysis step is the pollen amount estimation method according to claim 13, in which, in the artificial intelligence unit, a rectangular grid cell is cut out from the image data, and an object is detected based on the feature amount for each grid cell and the type of the object is estimated. The pollen amount estimation method according to claim 17 of the present invention is The image analysis step uses a color image, which is the pollen amount estimation method according to claim 13. The pollen amount estimation method according to claim 18 of the present invention is The flower is a flower of a fruit tree, which is the pollen amount estimation method according to claim 13. The pollen amount estimation program according to claim 19 of the present invention is An image acquisition step of acquiring an image of a flower, An image analysis step of analyzing the state of the flower from the acquired image, A pollen collection amount data acquisition step of acquiring a pre-given image and the pollen collection amount, A pollen collection amount calculation step that is executed after the image analysis step and the pollen collection amount data acquisition step, and calculates the pollen collection amount from the state of the flower analyzed in the image analysis step and the pre-given image and the pollen collection amount acquired in the pollen collection amount data acquisition step, It is a pollen amount estimation program having

Brief Description of Drawings

[0006]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Mode for Carrying Out the Invention

[0007] In the present invention, the amount of pollen collected is estimated by analyzing an image of a flower using artificial intelligence (AI) or the like. Machine learning may be used for the analysis. Also, the amount of pollen collected as an estimation result may be visualized. In order to enable pollen collection at the optimal time, the total amount of pollen to be collected increases, and it becomes possible to supplement the part that has been covered by imports with home cultivation.

[0008] FIG. 1 shows a configuration example of a pollen amount estimation system 1 in an embodiment of the present invention. In this embodiment, the pollen amount estimation system 1 includes an image acquisition unit 110, an image analysis unit 120, a pollen collection amount data acquisition unit 130, and a pollen collection amount calculation unit 140. The image acquisition unit 110 acquires an image of a flower.

[0009] The image analysis unit 120 analyzes the state of the flower from the image acquired by the image acquisition unit 110. The analysis method may be machine learning described later, or deep learning among machine learning. In the case of deep learning, the pollen collection amount data acquisition unit 130 can use each image and the pollen collection amount corresponding to that image, and it is also possible to integrate the image analysis unit 120 and the pollen collection amount calculation unit 140, or the image analysis unit 120, the collection amount data acquisition unit, and the pollen collection amount calculation unit 140. Also, the analysis method may be image analysis such as pattern matching or feature extraction. The pollen collection amount data acquisition unit 130 acquires a given image and pollen collection amount data in advance. The pollen collection amount calculation unit 140 calculates the pollen collection amount from the state of the flower analyzed by the image analysis unit 120 and the given image and pollen collection amount data acquired by the pollen collection amount data acquisition unit 130.

[0010] FIG. 2 shows a configuration example of the pollen amount estimation system 1 in an embodiment of the present invention. In the present embodiment, the image analysis unit 120 includes a stage classification unit 121 that classifies the flowering stage of the flower from the acquired image into three or more stages. The number of stages may be substantially infinite, that is, continuous stages.

[0011] The pollen collection amount data acquisition unit 130 includes a stage-by-stage pollen amount data acquisition unit 131 that acquires pollen collection amount data at each stage. In the present embodiment, the pollen collection amount data is the amount of pollen per flower, but other pollen collection amount data such as per branch or per unit area may be used. The stage in the stage classification unit 121 and the stage in the stage-by-stage pollen amount data acquisition unit 131 are the same. In the stage classification unit 121 within the image analysis unit 120, the stage of the flower in the target image is classified, and the pollen collection amount is calculated using the stage-by-stage pollen amount obtained by the stage-by-stage pollen amount data acquisition unit 131.

[0012] In an embodiment of the present invention, the three or more stages are four stages of "early" (immature stage), "best" (bladder stage), "late" (flowering stage), and "too late" (post-flowering stage). In this specification, the "stage" that is the flowering stage of the flower may be referred to as "stage". In an embodiment of the present invention, the image analysis unit 120 cuts out a square grid cell from the image data, detects an object based on the feature amount for each grid cell, and estimates the type of the object.

[0013] In this embodiment, as an estimation technique by AI, YOLO (You Only Look Once), which extracts features through multiple layers for each grid cell, is used. YOLO is an object detection algorithm based on machine learning. The advantages of using an algorithm such as YOLO include fast processing due to its simplified structure, high accuracy in distinguishing the background and objects, and high generalization performance. In YOLO, the image data is resized into a square, and based on the features for each divided grid cell, an object is detected, the type of the object is estimated from the content of the grid cell, and the result is output.

[0014] When discriminating images in which the flowering stages of flowers, that is, stages are mixed, the correct answer rate in the state where "late" and "too late" are mixed is about 89%, and the correct answer rate in the state where "early" and "best" are mixed is about 94%, etc., enabling high-precision classification. In this embodiment, YOLO is used, but other methods may also be used.

[0015] FIG. 3 shows a configuration example of the pollen amount estimation system 1 in an embodiment of the present invention. In this embodiment, the image analysis unit 120 and the pollen collection amount calculation unit 140 are integrally formed as the artificial intelligence unit 150.

[0016] In an embodiment of the present invention, the image analysis unit 120 acquires a color image. Here, the color image is not only an image represented by the three primary colors of R (red), G (green), and B (blue) in the general visible region of 380 nm to 780 nm, but also includes the near-ultraviolet region of 300 nm to 380 nm.

[0017] In the above embodiment, the flower can be a flower of a fruit tree. Here, the fruit tree refers to a herbaceous plant and a woody plant that are cultivated for two or more years and whose fruits are for edible use, including pears, apples, plums, and kiwis. In the above example, a pear is given as an example, but it is known that the collection amount varies depending on the flowering stage even in other fruit trees such as apples, plums, and kiwis. For example, by learning images of flower buds classified by flowering stage for each fruit tree and investigating the pollen collection amount at each stage in advance, it can be used to determine the appropriate timing for pollen collection. In the above embodiments, the flower is an entomophilous flower, and the image can be an image using light of a plurality of wavelengths among the wavelengths acquired by at least one insect corresponding to the entomophilous flower. When using a color image, discrimination by the eyes of insects can also be utilized. In particular, when the near-ultraviolet region is included, discrimination by the eyes of insects can be more effectively utilized.

[0018] FIG. 4 shows a configuration example of the pollen amount estimation system 1 according to an embodiment of the present invention. In the present embodiment, the pollen amount estimation system 1 includes a mobile terminal communication unit 160 that communicates with a mobile terminal. In the present embodiment, the mobile terminal communication unit 160 is connected to an image acquisition unit 110 and a pollen collection amount calculation unit 140. The image acquisition unit 110 acquires an image captured by the mobile terminal, and the pollen collection amount calculation unit 140 transmits the pollen collection amount to the mobile terminal.

[0019] FIG. 5 shows a configuration example of the pollen amount estimation system 1 according to an embodiment of the present invention. In the present embodiment, the pollen amount estimation system 1 includes an imaging device communication unit 170 that communicates with an installed imaging device. In this embodiment, the imaging device communication unit 170 is connected to the image acquisition unit 110.

[0020] The image acquisition unit 110 acquires an image captured by the installed imaging device. As shown in the figure, it may further include a mobile terminal communication unit 160, and the pollen collection amount calculation unit 140 may be configured to transmit the pollen collection amount to the mobile terminal.

[0021] FIG. 6 shows a configuration example of the pollen amount estimation system 1 according to an embodiment of the present invention. In this embodiment, the pollen amount estimation system 1 includes a flower bud collector communication unit 180 that communicates with a flower bud collector for collecting flower buds, and transmits a signal based on the estimated pollen amount to the flower bud collector.

[0022] The flower bud collector may be an aerial vehicle such as a drone or a land vehicle. The connection may be wireless or wired. Alternatively, it may be mounted on the flower bud collector. Then, a signal based on the estimated pollen amount is transmitted to the flower bud collector via the flower bud collector connection unit. As a result, signals such as the pollen amount and whether to perform collection can be transmitted to the flower bud collector, and it becomes possible to collect flower buds at the optimal time by the flower bud collector.

[0023] FIG. 7 shows a configuration example of a pollen amount estimation method according to an embodiment of the present invention. In this embodiment, the pollen amount estimation method includes an image acquisition step S110, an image analysis step S120, a pollen collection amount data acquisition step S130, and a pollen collection amount calculation step S140 that is executed after the image analysis step S120 and the pollen collection amount data acquisition step S130.

[0024] In the image acquisition step S110, an image of a flower is acquired. In the image analysis step S120, the state of the flower is analyzed from the acquired image. In the pollen collection amount data acquisition step S130, pre-provided images and pollen collection amount data are acquired.

[0025] The pollen collection amount is calculated from the state of the flower analyzed in the image analysis step S120 and the pre-provided images and pollen collection amount data acquired in the pollen collection amount data acquisition step S130. A configuration in which the image analysis unit 120, the collection amount data acquisition unit, and the pollen collection amount calculation unit 140 are integrated into deep learning is also included in this configuration.

[0026] FIG. 8 shows a configuration example of a pollen amount estimation method according to an embodiment of the present invention. In this embodiment, the image analysis step S120 includes a stage classification step S121 of classifying the flowering stage of the flower into three or more stages from the acquired image.

[0027] The stages may be continuous infinite stages. It may be image analysis such as pattern matching or feature extraction, or analysis using artificial intelligence such as machine learning or deep learning. The pollen collection amount data acquisition step S130 includes a pollen amount data acquisition step S131 for each stage of acquiring the pollen collection amount data at each stage. In this embodiment, the pollen collection amount data is per flower, but it may also be per branch or per unit area. The stages in the stage classification step S121 and the stages in the pollen amount data acquisition step S131 for each stage are the same. In the stage classification step S121 within the image analysis step S120, the stage of the flower in the target image is classified, and the pollen collection amount is calculated using the pollen amounts for each stage obtained in the pollen amount data acquisition step S131 for each stage.

[0028] In one embodiment of the present invention, for the three or more stages in the pollen amount estimation method, they are four stages of "early" (immature stage), "best" (bladder stage), "late" (flowering stage), and "too late" (post-flowering stage). In one embodiment of the present invention, the image analysis step S120 can be configured to classify the flowering stage of each flower using the analysis result in the artificial intelligence unit 150.

[0029] In one embodiment of the present invention, in the image analysis step S120, in the artificial intelligence unit 150, a square grid cell is cut out from the image data, and based on the features for each grid cell, an object is detected and the type of the object is estimated. In one embodiment of the present invention, the image analysis step S120 uses a color image. Here, the color image is not only an image represented by the three primary colors of R (red), G (green), and B (blue) in the general visible region of 380 nm to 780 nm, but also includes the near-ultraviolet region of 300 nm to 380 nm.

[0030] In the above embodiments, the flower for the pollen amount estimation method can be a flower of a fruit tree. Here, the fruit tree refers to herbaceous plants and woody plants that have been cultivated for two or more years and whose fruits are for human consumption, including pears, apples, peaches, kiwis, etc. In the above example, a pear is given as an example. However, it is known that the collection amount varies depending on the flowering stage in other fruit trees such as apples, peaches, and kiwis. For example, by learning the images of flower buds classified by the flowering stage for each fruit tree and investigating the pollen collection amount at each stage in advance, it can be used to determine the appropriate timing for pollen collection. In the above embodiments, the flower is an entomophilous flower, and the image can be an image using light of a plurality of wavelengths among the wavelengths acquired by at least one insect corresponding to the entomophilous flower. When using a color image, discrimination by the eyes of insects can also be utilized. In particular, when the near-ultraviolet region is included, discrimination by the eyes of insects can be more effectively utilized.

[0031] FIG. 9 shows a configuration example of a pollen amount estimation program in an embodiment of the present invention. In this embodiment, the pollen amount estimation program includes an image acquisition step S110, an image analysis step S120, a pollen collection amount data acquisition step S130, and a pollen collection amount calculation step S140.

[0032] In the image acquisition step S110, an image of a flower is acquired and stored in a memory. In the image analysis step S120, the state of the flower is analyzed from the acquired image using an image analysis program stored in a hard disk and stored in a memory.

[0033] In the pollen collection amount data acquisition step S130, pre-provided images and pollen collection amount data are acquired and stored in a memory. In the pollen collection amount calculation step S140 executed after the image analysis step S120 and the pollen collection amount data acquisition step S130, using the pollen collection amount calculation program stored in the hard disk, from the state of the flower analyzed in the image analysis step S120 stored in the memory, the given image obtained in the pollen collection amount data acquisition step S130, and the pollen collection amount data, the pollen collection amount is calculated.

[0034] In one embodiment, the image analysis step S120 can have a stage classification step S121 that classifies the flowering stage of the flower from the acquired image into three or more stages. In one embodiment, the stages may be continuous infinite stages. It may be image analysis such as pattern matching or feature extraction, or analysis using artificial intelligence such as machine learning or deep learning.

[0035] In one embodiment, the pollen collection amount data acquisition step S130 has a stage-by-stage pollen amount data acquisition step S131 that acquires the pollen collection amount data at each stage. In this embodiment, the pollen collection amount data is per flower, but it may also be per branch or per unit area, etc. In one embodiment, the central processing unit CPU, the memory RAM, and the hard disk HDD unit may use the cloud. The stage in the stage classification step S121 and the stage in the stage-by-stage pollen amount data acquisition step S131 are the same. In the stage classification step S121 within the image analysis step S120, the stage of the flower in the target image is classified, and using the stage-by-stage pollen amounts obtained in the stage-by-stage pollen amount data acquisition step S131, the pollen collection amount is calculated.

[0036] The present invention is not limited to the above embodiments, and it goes without saying that the present invention includes various embodiments without departing from the gist of the present invention. For example, it can also be configured to include a mobile terminal communication unit, an imaging device communication unit, and a flower bud collection machine communication unit.

Explanation of Reference Numerals

[0037] 1 Pollen Quantity Estimation System 110 Image Acquisition Unit 120 Image Analysis Unit 121 Stage Classification Unit 130 Pollen Collection Quantity Data Acquisition Unit 131 Stage-by-Stage Pollen Quantity Data Acquisition Unit 140 Pollen Collection Quantity Calculation Unit 150 Artificial Intelligence Unit 160 Mobile Terminal Communication Unit 170 Imaging Device Communication Unit 180 Bud Collection Machine Communication Unit S110 Image Acquisition Step S120 Image Analysis Step S121 Stage Classification Step S122 Main Image Analysis Step S130 Pollen Collection Quantity Data Acquisition Step S131 Stage-by-Stage Pollen Quantity Data Acquisition Step S132 Main Pollen Collection Quantity Data Acquisition Step S140 Pollen Collection Quantity Calculation Step

Claims

1. An image acquisition unit that acquires an image of a flower, An image analysis unit that analyzes the state of the flower from the acquired image, A pollen collection amount data acquisition unit that acquires a pre-given image and the amount of pollen collected, and A pollen collection amount calculation unit that calculates the amount of pollen collected from the state of the flower analyzed by the image analysis unit and the pre-given image and pollen collection amount data acquired by the pollen collection amount data acquisition unit, A pollen amount estimation system comprising the above.

2. The image analysis unit includes a stage classification unit that classifies the flowering stage of the flower from the acquired image into three or more stages, The pollen collection amount data acquisition unit includes a stage-by-stage pollen amount data acquisition unit that acquires the amount of pollen collected at each stage. The pollen amount estimation system according to Claim 1.

3. The three or more stages are four stages: immature stage, balloon stage, flowering stage, and post-flowering stage. The pollen amount estimation system according to Claim 2.

4. The image analysis unit cuts out a square grid cell from the image data, detects an object based on the feature amount for each grid cell, and estimates the type of the object. The pollen amount estimation system according to Claim 2.

5. The image analysis unit and the pollen collection amount calculation unit are integrally formed as an artificial intelligence unit. The pollen amount estimation system according to Claim 2.

6. The image analysis unit acquires a color image. The pollen amount estimation system according to Claim 2.

7. The flower is a flower of a fruit tree. The pollen amount estimation system according to Claim 6.

8. The flower is an entomophilous flower, and the image is an image using light of a plurality of wavelengths among the wavelengths acquired by at least one insect corresponding to the entomophilous flower. The pollen amount estimation system according to Claim 2.

9. Comprising a mobile terminal communication unit that communicates with a mobile terminal, The image acquisition unit acquires an image captured by the mobile terminal, The pollen collection amount calculation unit transmits the pollen collection amount to the mobile terminal. The pollen amount estimation system according to any one of Claims 1 to 9.

10. Comprising an imaging device communication unit that communicates with an installation-type imaging device, and the image acquisition unit acquires an image captured by the installation-type imaging device. The pollen amount estimation system according to any one of Claims 1 to 9.

11. Comprising a flower bud collection machine communication unit that communicates with a flower bud collection machine that collects flower buds, Transmits a signal based on the estimated pollen amount to the flower bud collection machine. The pollen amount estimation system according to any one of Claims 1 to 9.

12. An image acquisition step of acquiring an image of a flower, An image analysis step of analyzing the state of the flower from the acquired image, A pollen collection amount data acquisition step of acquiring a given image and pollen collection amount data in advance, A pollen collection amount calculation step that is executed after the image analysis step and the pollen collection amount data acquisition step, and calculates the pollen collection amount from the state of the flower analyzed in the image analysis step and the given image and pollen collection amount data acquired in the pollen collection amount data acquisition step, A pollen amount estimation method having the above.

13. The image analysis step includes a step classification step of classifying the flowering stage of the flower from the acquired image into three or more stages, The pollen collection amount data acquisition step includes a step-by-step pollen amount data acquisition step of acquiring pollen collection amount data at each stage. The pollen amount estimation method according to claim 12.

14. The three or more stages are four stages: immature stage, balloon stage, flowering stage, and post-flowering stage. The pollen amount estimation method according to claim 13.

15. The image analysis step classifies the flowering stage of each flower using the analysis result in the artificial intelligence unit. The pollen amount estimation method according to claim 13.

16. In the image analysis step, in the artificial intelligence unit, a rectangular grid cell is cut out from the image data, and an object is detected based on the feature amount for each grid cell and the type of the object is estimated. The pollen amount estimation method according to claim 13.

17. The image analysis step uses a color image. The pollen amount estimation method according to claim 13.

18. The flower is a flower of a fruit tree. The pollen amount estimation method according to claim 13.

19. An image acquisition step of acquiring an image of a flower, An image analysis step of analyzing the state of the flower from the acquired image, A pollen collection amount data acquisition step of acquiring a given image and pollen collection amount in advance, A pollen collection amount calculation step that is executed after the image analysis step and the pollen collection amount data acquisition step, and calculates the pollen collection amount from the state of the flower analyzed in the image analysis step and the given image and pollen collection amount acquired in the pollen collection amount data acquisition step, A pollen amount estimation program having the above.

Images