Method for estimating the fruit weight of strawberries at harvest, program for estimating the fruit weight of strawberries at harvest, and method for obtaining the number of achenes and program for obtaining the number of achenes.

Abstract

This method efficiently estimates the harvest weight of each fruit in a cluster of strawberries on a plant. [Solution] The computer obtains the number of achenes of any fruit on a strawberry plant, information on the strawberry variety, information identifying the fruit cluster to which the fruit belongs, and fruit rank information indicating which fruit the fruit is in that fruit cluster. From among the first information that shows the relationship between the number of achenes of a fruit determined for each variety and an index corresponding to the fruit's harvest weight, the computer identifies the first information corresponding to the obtained variety information. Based on the identified first information and the obtained number of achenes of the fruit, the computer estimates the harvest weight of the fruit. Based on the second information that shows the relationship between the index corresponding to the harvest weights of the Nth and Mth fruits and the index corresponding to the harvest weight of the fruit, the computer estimates the harvest weight of the fruits included in the fruit cluster to which the fruit belongs, and outputs the estimated harvest weight.

Description

Technical Field

[0001] The present invention relates to a method for estimating the fruit weight at the time of harvesting strawberries, a program for estimating the fruit weight at the time of harvesting strawberries, a method for obtaining the number of achenes, and a program for obtaining the number of achenes.

Background Art

[0002] In farmers and agricultural corporations engaged in crop production, in order to conduct efficient management, it is important to predict the yield before harvesting and formulate a harvesting work (labor management) plan and a shipping plan.

[0003] Conventionally, when estimating the yield of crops, it has generally been carried out based on past yield data and experience after producers, etc., grasped the cultivation situation. On the other hand, in recent years, when estimating the yield of fruit vegetables cultivated in facilities, a method has been developed that uses the biological information of crops regularly acquired and the information on the environment in which the crops grow (see, for example, Patent Document 1, Non-Patent Document 1, etc.). In addition, as methods for estimating the yield of strawberries and the like, the methods described in Patent Documents 2 to 4 and the like are known.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Patent Document 3

Patent Document 4

Non-Patent Documents

[0005]

Non-Patent Document 1

[0006] Research is underway on the relationship between the number of achenes in strawberry fruit and fruit weight (see, for example, Non-Patent Documents 2 and 3). Non-Patent Document 2 shows that the earlier the fruiting order, the more achenes there are, and the more achenes there are, the greater the fruit weight.

[0007] However, while Non-Patent Document 2 describes trends regarding the relationship between fruit setting order, number of achenes, and fruit weight, it merely indicates trends.

[0008] Therefore, the present invention aims to provide a method and program for estimating the harvest weight of strawberries, which can efficiently estimate the harvest weight of each fruit contained in a fruit cluster on a strawberry plant. Furthermore, the present invention aims to provide a method and program for obtaining the number of achenes, which can accurately estimate the number of achenes. [Means for solving the problem]

[0009] The present invention provides a method for estimating the harvest weight of strawberries, in which a computer performs the following processes: obtaining the number of achenes of an arbitrary fruit on a strawberry plant; obtaining information on the strawberry variety; information identifying the fruit cluster to which the arbitrary fruit belongs; identifying the first information corresponding to the obtained variety information from among the first information showing the relationship between the number of achenes of a fruit determined for each variety and an index corresponding to the harvest weight of that fruit; estimating the harvest weight of the arbitrary fruit based on the identified first information and the obtained number of achenes of the arbitrary fruit; estimating the harvest weight of the fruits included in the fruit cluster to which the arbitrary fruit belongs based on the second information showing the relationship between the index corresponding to the harvest weight of the Nth fruit (where N is a natural number greater than or equal to 1) and the index corresponding to the harvest weight of the Mth fruit (where M is a natural number other than N greater than or equal to 1) and the index corresponding to the harvest weight of the arbitrary fruit; and outputting the estimated harvest weight. [Effects of the Invention]

[0010] The strawberry harvest weight estimation method and strawberry harvest weight estimation program of the present invention have the effect of efficiently estimating the harvest weight of each fruit contained in a fruit cluster on a strawberry plant. Furthermore, the achene count acquisition method and achene count acquisition program of the present invention have the effect of accurately estimating the number of achenes. [Brief explanation of the drawing]

[0011] [Figure 1] Figure 1 is a schematic diagram showing the configuration of a strawberry harvest weight estimation system according to one embodiment. [Figure 2] Figure 2 shows an example of the hardware configuration of an information processing device. [Figure 3] Figure 3 is a functional block diagram of the information processing device. [Figure 4] Figure 4(a) shows an example of an image of a strawberry plant, and Figure 4(b) is a diagram illustrating the achene of a strawberry. [Figure 5]FIG. 5 is a diagram for explaining a method of identifying the first fruit cluster, the second fruit cluster, etc. from images. [Figure 6] FIG. 6 is a diagram for explaining a method of identifying the ranking information of a fruit (the Nth fruit) that has started to grow. [Figure 7] FIGS. 7(a) and 7(b) are diagrams showing the results of an investigation on the relationship between the total number of achenes per strawberry fruit and the fruit weight at harvest (potential) for each of the strawberry varieties "Koi Midori" and "Yotsuboshi". [Figure 8] FIG. 8(a) is a graph showing the relationship between the average temperature from the flowering date to the harvest date and the fruit weight increase rate a, and FIG. 8(b) is a graph showing the relationship between the accumulated solar radiation from the flowering date to the harvest date and the fruit weight increase rate b. [Figure 9] FIG. 9 is a graph showing the relationship between the fruit weight at harvest (potential) of the first fruit in an arbitrary fruit cluster and the fruit weight at harvest (potential) of other fruits in the same fruit cluster. [Figure 10] FIG. 10 is a flowchart showing the processing of the information processing device. [Figure 11] FIG. 11 is a diagram showing an example of the screen output by the output unit.

Embodiments for Carrying Out the Invention

[0012] Hereinafter, a strawberry fruit weight estimation system according to an embodiment will be described in detail.

[0013] FIG. 1 schematically shows the configuration of a strawberry fruit weight estimation system 100 according to an embodiment. In this embodiment, the strawberry fruit weight estimation system 100 includes a camera 10, an information acquisition device 12, and an information processing device 70. The camera 10, the information acquisition device 12, and the information processing device 70 are connected to a network 80 such as the Internet, a mobile phone communication network, a WAN (Wide Area Network), or a LAN (Local Area Network), and information can be exchanged between the devices.

[0014] (Camera 10) Camera 10 is a camera used to photograph strawberry plants being cultivated in a field, and can be a CCD (Charge Coupled Device) camera, a CMOS (Complementary Metal Oxide Semiconductor) camera, an infrared camera, or other type of camera. Camera 10 may be a camera that can be carried by an operator, or it may be a fixed-point camera installed in large numbers in the field. Furthermore, Camera 10 may be a camera mounted on a mobile device that can move along rails laid in the field, or it may be a camera mounted on a flying device such as a drone.

[0015] (Information acquisition device 12) The information acquisition device 12 is a device that acquires information necessary to identify the strawberry plant (the plant being photographed) that is the subject of the image captured by the camera 10. For example, when the information processing device 70 identifies the plant being photographed based on the position and shooting direction of the camera 10, the information acquisition device 12 acquires position information and shooting direction information from the GNSS (Global Navigation Satellite System) sensor, angular velocity sensor, etc., installed on the camera 10. If the strawberry field is indoors or otherwise unable to be positioned using the GNSS sensor, the camera 10's position information may be acquired using other means (such as beacons). Also, for example, when the information processing device 70 identifies the plant being photographed based on information that can identify the plant contained in the image captured by the camera 10 (for example, text information on a sign installed near the plant or information on a 2D code), the information acquisition device 12 acquires information that can identify the plant from the image. Also, for example, when the information processing device 70 identifies the plant being photographed based on the photographer's input information (such as identification information of the plant being photographed), the information acquisition device 12 acquires the photographer's input information. The information acquisition device 12 may be integrated with the camera 10 or with the information processing device 70.

[0016] (Information processing device 70) The information processing device 70 can be a smartphone, a PC (Personal Computer), or a server, and it receives images captured by the camera 10 and information acquired by the information acquisition device 12. The information processing device 70 then uses the received images and information to estimate the harvest weight of each fruit in a specific fruit cluster on the photographed plant.

[0017] Figure 2 shows the hardware configuration of the information processing device 70. As shown in Figure 2, the information processing device 70 includes a CPU (Central Processing Unit) 190, ROM (Read Only Memory) 192, RAM (Random Access Memory) 194, a storage unit (such as an HDD (Hard Disk Drive) or SSD (Solid State Drive)) 196, a network interface 197, a display unit 193, an input unit 195, and a portable storage medium drive 199. The display unit 193 includes a liquid crystal display or an organic EL display, and the input unit 195 includes a keyboard, mouse, touch panel, etc. Each of these components of the information processing device 70 is connected to a bus 198. In the information processing device 70, the CPU 190 executes programs stored in the ROM 192 or the storage unit 196, or programs read from the portable storage medium 191 by the portable storage medium drive 199, thereby realizing the functions of each component shown in Figure 3. The functions of each part in Figure 3 may be implemented by integrated circuits such as ASICs (Application Specific Integrated Circuits) or FPGAs (Field Programmable Gate Arrays).

[0018] In the information processing device 70, the CPU 190 executes a program to realize the functions of an image acquisition unit 40, a plant information acquisition unit 41, a fruit cluster identification unit 42, a fruit identification unit 44, an achene number estimation unit 46, a flowering date estimation unit 48, an environmental information acquisition unit 50, a fruit weight estimation unit 52, and an output unit 54, as shown in Figure 3.

[0019] The image acquisition unit 40 acquires images (images of strawberry plants) captured by the camera 10. As an example, the image acquisition unit 40 acquires images of strawberry plants as shown in Figure 4(a).

[0020] The plant information acquisition unit 41 acquires information necessary to identify the plant photographed in the image (the plant being photographed) from the information acquisition device 12, and identifies the identification information of the plant being photographed and the variety information corresponding to the identification information. The plant information acquisition unit 41 refers to a database (not shown) that manages the identification information, variety information, and other information (such as plant location information) of each plant cultivated in the field in a linked manner.

[0021] The fruit cluster identification unit 42 identifies the fruit clusters (flower clusters) included in the image acquired by the image acquisition unit 40, and also identifies which fruit cluster (flower cluster) each identified is. In the following, a fruit cluster (flower cluster) will be simply referred to as "fruit cluster." The P-th fruit cluster will be called the "P-th fruit cluster" (where P is a natural number).

[0022] Here, the method by which the fruit cluster identification unit 42 identifies the first fruit cluster, the second fruit cluster, etc. from the image will be explained based on Figure 5. In this embodiment, the fruit cluster identification unit 42 identifies the first fruit cluster, the second fruit cluster, etc. using machine learning techniques.

[0023] As shown in Figure 5, when using machine learning methods, (1) a learning phase and (2) an estimation phase are performed.

[0024] (1) In the learning phase, a large number of raw images (many images of strawberry plants) are collected, and in the annotation process, a person specifies the correct answers (location and range of the first fruit cluster, second fruit cluster, etc.) to generate training data. Furthermore, by performing machine learning using the training data, a trained model is generated to estimate the location and range of the first fruit cluster, second fruit cluster, etc. from the images.

[0025] On the other hand, in the estimation phase (2), when the image acquired by the image acquisition unit 40 is input to the fruit cluster identification unit 42, the fruit cluster identification unit 42 uses the trained model generated in the learning phase to identify the position and range of the first fruit cluster, the second fruit cluster, ... in the image.

[0026] Returning to Figure 3, the fruit identification unit 44 identifies fruits that have begun to enlarge (green ripe stage) in the images acquired by the image acquisition unit 40, and also identifies the fruit cluster containing those fruits. The fruit identification unit 44 also identifies rank information (the Nth fruit) indicating which fruit the identified fruit is in the identified fruit cluster.

[0027] Specifically, the fruit identification unit 44 identifies the fruit that has begun to enlarge (the Nth fruit) using a machine learning model (a pre-trained model) that extracts fruits that have begun to enlarge from images. This machine learning model can be generated by machine learning a large number of images of fruits that have begun to enlarge that have been collected in advance.

[0028] Furthermore, the fruit identification unit 44 uses a machine learning method as shown in Figure 6 to identify the rank information of the fruit that has begun to enlarge (the Nth fruit). In the machine learning learning phase shown in Figure 6, a large number of raw images (many images of strawberry clusters) are collected, and in the annotation process, a person specifies the correct answers (1st fruit, 2nd fruit, ...) to generate training data. In addition, by performing machine learning using the training data, a trained model is generated to estimate the rank information of the fruits (1st fruit, 2nd fruit, ...) from the images. On the other hand, in the estimation phase, when the image acquired by the image acquisition unit 40 is input to the fruit identification unit 44, the fruit identification unit 44 uses the trained model generated in the learning phase to identify the rank information of each fruit, and also identifies the rank information (N) of the fruit that has begun to enlarge.

[0029] Returning to Figure 3, the achene number estimation unit 46 estimates the total number of achenes in the fruit (the Nth fruit) that has begun to enlarge, as identified by the fruit identification unit 44. Specifically, the achene number estimation unit 46 identifies the location of the achenes in the Nth fruit using a machine learning model (a trained model) that extracts the location of achenes from an image of the Nth fruit. The achene number estimation unit 46 also counts the number of achenes in the Nth fruit. The machine learning model that extracts the location of achenes from an image of the Nth fruit can be generated by machine learning a large number of achene images collected in advance. Here, as shown in Figure 4(b), since the image is two-dimensional, only about half the number of achenes in the fruit can be counted from the image. Therefore, the achene number estimation unit 46 estimates the total number of achenes in the Nth fruit by multiplying the counted number of achenes by a predetermined number n (where n is, for example, 2). The predetermined number n is the ratio of the number of achenes obtained from the image to the total number of achenes, and shall be determined by experiments or simulations conducted in advance.

[0030] The flowering date estimation unit 48 estimates the flowering date of each fruit and flower included in the image. Specifically, the flowering date estimation unit 48 identifies the number of days since flowering for all fruits and flowers included in the image by inputting the image acquired by the image acquisition unit 40 into a trained model generated using a large amount of training data that associates raw images of fruits and flowers with the number of days since flowering of the fruit or flower in question. Furthermore, the flowering date estimation unit 48 estimates the flowering date of each fruit and each flower based on the identified number of days since flowering and the date the image was taken.

[0031] The environmental information acquisition unit 50 acquires measured and predicted values ​​(post-flowering environmental information) of environmental information (temperature and solar radiation) for each fruit and flower from the flowering date to the harvest date. The environmental information acquisition unit 50 shall acquire measured and predicted values ​​from, for example, the servers of the Japan Meteorological Agency or the National Agriculture and Food Research Organization. Normal values ​​may be used as predicted values. The harvest date shall be the day on which the cumulative temperature from the flowering date reaches a predetermined value.

[0032] The fruit weight estimation unit 52 estimates the harvest weight of the fruit identified by the fruit identification unit 44 based on the total number of achenes of the fruit that has begun to enlarge (the Nth fruit) estimated by the achene number estimation unit 46, and the environmental information acquired by the environmental information acquisition unit 50 from the flowering date to the harvest date. The fruit weight estimation unit 52 also estimates the harvest weight of fruits other than the Nth fruit (the Mth fruit). The fruit weight estimation unit 52 performs the above processing using the information shown in Figures 7(a) to 9.

[0033] (Regarding Figures 7(a) and 7(b)) In the case of strawberries, the total number of achenes in each fruit is determined at least before flowering (during the pistil differentiation stage), and it is known that there is a correlation between the total number of achenes and the fruit weight at harvest. Figures 7(a) and 7(b) show the results of an investigation into the relationship between the total number of achenes per strawberry fruit and the fruit weight at harvest (potential) for the strawberry varieties "Koiminori" and "Yotsuboshi," respectively. In Figures 7(a) and 7(b), the horizontal axis represents the total number of achenes per strawberry fruit, and the vertical axis represents the fruit weight at harvest (potential). In Figure 7(a), the approximate straight line for each point (●) is shown as a dashed line, and in Figure 7(b), the approximate straight line for each point (〇) is shown as a solid line. Here, "fruit weight at harvest (potential)" is calculated using the average temperature and cumulative solar radiation from flowering to harvest as a baseline (for example, 15℃, 150MJ / m²). 2 This refers to the fruit weight at harvest time, and can be said to be an indicator corresponding to the fruit weight at harvest time.

[0034] The equation of the approximate line in Figure 7(a) is y = 0.1107x. The coefficient of determination of this approximate line is R 2 The coefficient of determination R is 0.9914, indicating a high correlation between the total number of achenes per strawberry fruit and the fruit weight (potential) at harvest for the "Koiminori" variety. Furthermore, the equation of the approximation line in Figure 7(b) is y = 0.0882x. The coefficient of determination of this approximation line is R. 2The value is 0.9884, indicating a high correlation between the total number of achenes per strawberry fruit and the fruit weight (potential) at harvest for the "Yotsuboshi" variety. Furthermore, Figures 7(a) and 7(b) show that the equation of the approximation line differs depending on the variety. In this embodiment, the relationship (first information) shown in Figures 7(a) and 7(b) is prepared in advance for each variety.

[0035] (Regarding Figures 8(a) and 8(b)) Figure 8(a) is a graph showing the relationship between the average temperature from flowering to harvest and the fruit weight increase rate a. The fruit weight increase rate a is a value that indicates how much fruit weight decreases when the average temperature is higher than the standard value (15°C) and how much fruit weight increases when the average temperature is lower than the standard value (15°C). In this embodiment, the information in Figure 8(a) is obtained from cultivation tests, simulations, etc.

[0036] Figure 8(b) is a graph showing the relationship between cumulative solar radiation from flowering day to harvest day and the fruit weight increase rate b. The fruit weight increase rate b is calculated based on the cumulative solar radiation being at a baseline value (150 MJ / m²). 2 How much does fruit weight increase when the cumulative solar radiation exceeds the standard value (150 MJ / m²)? 2 This value indicates how much the fruit weight decreases when it is less than the specified value. In this embodiment, the information in Figure 8(b) is obtained from cultivation tests, simulations, etc.

[0037] (Regarding Figure 9) Figure 9 is a graph showing the relationship between the harvest weight (potential) of the first fruit in any fruit cluster and the harvest weight (potential) of the other fruits in the same cluster. Note that Figure 9 only shows the relationships between the first and second fruits, the first and fourth fruits, and the first and sixth fruits; the relationships between the other fruits (third, fifth, etc.) and the first fruit are omitted. As shown in Figure 9, there is a proportional relationship between the harvest weight (potential) of the first fruit in any fruit cluster and the harvest weight (potential) of the other fruits in the same cluster. Furthermore, there is a tendency for the harvest weight (potential) to decrease as the fruit's rank increases. Furthermore, using the graph in Figure 9, it is possible to determine the harvest weight (potential) of the Nth fruit (N is a natural number) and the harvest weight (potential) of the Mth fruit (M is a natural number other than N) within any given fruit cluster. Therefore, the graph in Figure 9 can also be considered a second piece of information showing the relationship between the harvest weight (potential) of the Nth and Mth fruits.

[0038] The fruit weight estimation unit 52 identifies the first information corresponding to the variety identified by the plant information acquisition unit 41 from the first information prepared for each variety (Figures 7(a) and 7(b)), and estimates the harvest fruit weight (potential) of the Nth fruit from the total number of achenes of the fruit that has started to enlarge (the Nth fruit) estimated by the achene number estimation unit 46. For example, if the variety is "Koiminori" and the total number of achenes of the fruit that has started to enlarge (the Nth fruit) is 250, the fruit weight estimation unit 52 uses Figure 7(a) (first information) to estimate the harvest fruit weight (potential) as 0.1107 × 250 = 27.675 g.

[0039] Furthermore, the fruit weight estimation unit 52 first determines the harvest weight (potential) of the fruit that has begun to enlarge (the Nth fruit), then uses the graph in Figure 8(a) to identify the fruit weight increase rate a corresponding to the average temperature from the flowering date to the harvest date of the Nth fruit, and uses the graph in Figure 8(b) to identify the fruit weight increase rate b corresponding to the cumulative solar radiation from the flowering date to the harvest date of the Nth fruit. Then, using the harvest weight (potential) of the Nth fruit and the fruit weight increase rates a and b, the fruit weight estimation unit 52 estimates the harvest weight of the Nth fruit from the following equation (1). Fruit weight at harvest = Fruit weight at harvest (potential) × a × b …(1)

[0040] Furthermore, the fruit weight estimation unit 52 uses the graph in Figure 9 to determine the harvest weight (potential) of the Mth fruit from the harvest weight (potential) of the Nth fruit. Then, the fruit weight estimation unit 52 uses the harvest weight (potential) of the Mth fruit, the fruit weight increase rate a corresponding to the average temperature from the flowering date to the harvest date of the Mth fruit, and the fruit weight increase rate b corresponding to the cumulative solar radiation from the flowering date to the harvest date of the Mth fruit to estimate the harvest weight of the Mth fruit from equation (1) above.

[0041] Returning to Figure 3, the output unit 54 outputs the harvest weight of each fruit estimated by the fruit weight estimation unit 52 (for example, by displaying it on the display unit 193 or transmitting it to another terminal, etc.).

[0042] (Regarding flowcharts) Next, the processing of the information processing device 70 will be explained in detail according to the flowchart in Figure 10.

[0043] When the process shown in Figure 10 begins, first, in step S10, the image acquisition unit 40 waits until an image of a strawberry plant is transmitted. Once an image of a strawberry plant is transmitted, the image acquisition unit 40 moves to step S12 and acquires the transmitted image. The plant information acquisition unit 41 also acquires information necessary to identify the plant that is the subject of the image. Based on the acquired information, the plant information acquisition unit 41 identifies the identification information and variety information of the plant that is the subject of the image.

[0044] Next, in step S14, the fruit cluster identification unit 42 identifies the fruit clusters included in the image acquired by the image acquisition unit 40, and also identifies the number of each fruit cluster in the plant (fruit cluster rank). For example, if an image like the one shown in Figure 4(a) is acquired in step S12, the fruit cluster identification unit 42 uses machine learning as shown in Figure 5 to identify the position and range of each fruit cluster in the image, and also identifies the rank of each fruit cluster (1st fruit cluster, 2nd fruit cluster, ...).

[0045] Next, in step S16, the fruit cluster identification unit 42 identifies fruits that have begun to enlarge (green ripe) from among the identified first fruit cluster, second fruit cluster, ..., and identifies the fruit cluster containing those fruits. Here, as an example, we will assume that the "first fruit cluster" is identified as the fruit cluster containing fruits that have begun to enlarge.

[0046] Next, in step S18, the fruit identification unit 44 identifies the rank information of the fruit that has begun to enlarge. Specifically, the fruit identification unit 44 identifies the rank information (first fruit, second fruit, ...) of each fruit included in the fruit cluster (first fruit cluster) identified in step S16 using machine learning as shown in Figure 6, and identifies which fruit it is that has begun to enlarge. Here, as an example, let's assume that the fruit that has begun to enlarge is the "first fruit".

[0047] Next, in step S20, the achene number estimation unit 46 counts the number of achenes of the fruit that has begun to enlarge (the first fruit of the first fruit cluster) in the image, and estimates the total number of achenes per fruit based on the counted number of achenes. If a predetermined number (the ratio of the total number of achenes of the fruit to the number of achenes of the fruit obtained from the image) n is set in advance, and the number of achenes of the first fruit of the first fruit cluster counted from the image is F, the achene number estimation unit 46 estimates the total number of achenes G of the first fruit of the first fruit cluster using the formula G = n × F.

[0048] Next, in step S22, the flowering date estimation unit 48 estimates the flowering dates of all fruits and flowers included in the identified fruit cluster. Specifically, the flowering date estimation unit 48 estimates the flowering dates of each fruit and flower included in the first fruit cluster by inputting the images acquired in step S12 into the aforementioned machine learning model (trained model).

[0049] Next, in step S24, the fruit weight estimation unit 52 estimates the harvest fruit weight (potential) from the total number of achenes of the fruit that has begun to enlarge (the first fruit of the first fruit cluster), and estimates the harvest fruit weight of the first fruit of the first fruit cluster using the environmental information acquisition unit 50 acquired from the flowering date to the harvest date. For example, if the variety information acquired by the plant information acquisition unit 41 is "Koiminori", the fruit weight estimation unit 52 uses the graph and formula in Figure 7(a) to determine the harvest fruit weight (potential) of the first fruit of the first fruit cluster from the total number of achenes G of the first fruit of the first fruit cluster estimated in step S20. For example, if the total number of achenes G is 220, the harvest fruit weight (potential) can be determined to be 0.1107 × 220 = 24.354 g.

[0050] Furthermore, the fruit weight estimation unit 52 acquires the average temperature and cumulative solar radiation as environmental information from the flowering date to the harvest date of the first fruit of the first fruit cluster via the environmental information acquisition unit 50. Using the acquired environmental information, the graphs and equations in Figures 8(a) and 8(b), and equation (1) above, the unit estimates the harvest weight of the first fruit of the first fruit cluster. For example, if the average temperature is 14°C, then from Figure 8(a), the fruit weight increase rate a is 1.2186, and the cumulative solar radiation is 150 (MJ / m²). 2 If this is the case, then from Figure 8(b), the fruit weight increase rate b is 1.005.

[0051] In this case, the harvest weight of the first fruit in the first fruit cluster is, from equation (1) above, Fruit weight at harvest = 24.354 × 1.2186 × 1.005 = 29.826g It can be estimated that...

[0052] Next, in step S26, the fruit weight estimation unit 52 estimates the harvest weight (potential) of the other fruits in the fruit cluster and estimates the harvest weight using the environmental information after flowering. In this case, the harvest weight (potential) of the fruits other than the first fruit in the first fruit cluster can be determined from the graph and formula in Figure 9. In the above example, the harvest weight (potential) of the second fruit in the first fruit cluster is: Fruit weight at harvest (potential) = 0.8648 × 24.354 = 21.061g This is the result. The same can be determined for the other fruits.

[0053] The fruit weight estimation unit 52 then calculates the harvest weight of each fruit from equation (1) based on the harvest weight (potential) of each fruit and the fruit weight increase rates a and b corresponding to the environmental information from the flowering day to the harvest day of each fruit.

[0054] Next, in step S28, the fruit cluster identification unit 42 determines whether or not it has identified all fruit clusters containing fruits that have begun to enlarge. If the determination in step S28 is negative, the process returns to step S18. On the other hand, if the determination in step S28 is positive, the process proceeds to step S30.

[0055] When the process moves to step S30, the output unit 54 outputs information on the harvest weight of the photographed plants. For example, the output unit 54 uses the information obtained through the processing so far to generate a screen like the one shown in Figure 11 and outputs it to the display unit 193, etc. The screen in Figure 11 includes information on plant identification, variety information, harvest weight of each fruit in each fruit cluster, the number of achenes used to estimate the harvest weight (total number of achenes per strawberry fruit), the number of fruits per fruit cluster, and fruit cluster yield (sum of harvest weights of each fruit in the fruit cluster). A person (farmer, etc.) who views the screen in Figure 11 can check the yield of the photographed plants. If the yield of each plant within a predetermined range (for example, within a field) can be estimated based on images of all plants within that range, the screen in Figure 11 may display the yield of each plant as well as the total yield of all plants within the predetermined range.

[0056] As described in detail above, according to this embodiment, the information processing device 70 estimates the total number of achenes per fruit of any fruit from an image of a strawberry plant (Figure 4(a)) (S20), identifies information about the strawberry variety (S12), identifies information about the fruit cluster to which the arbitrary fruit belongs, and identifies fruit ranking information indicating which fruit the arbitrary fruit is in that fruit cluster (S16, S18). Furthermore, the information processing device 70 identifies the first information corresponding to the acquired variety from among the first information (Figure 7(a), Figure 7(b)) that shows the relationship between the total number of achenes per fruit of a fruit determined for each variety and the fruit weight (potential) at harvest of the fruit, and estimates the fruit weight at harvest of the arbitrary fruit based on the identified first information and the estimated total number of achenes per fruit of the arbitrary fruit (S24). Furthermore, the information processing device 70 estimates the harvest weight of each fruit included in the fruit cluster to which a given fruit belongs, based on second information (Figure 9) showing the relationship between the harvest weight (potential) of the Nth fruit (N is a natural number greater than or equal to 1) and the harvest weight (potential) of the Mth fruit (M is a natural number other than N greater than or equal to 1), and the harvest weight (potential) of the given fruit (S26). The information processing device 70 then outputs the estimated harvest weight of each fruit (S30, Figure 11). Thus, in this embodiment, by estimating the total number of achenes per fruit of a given fruit, the harvest weight of each fruit in the fruit cluster containing that given fruit can be efficiently estimated. In this case, since it is not necessary to use plant information other than the total number of achenes per strawberry fruit, the harvest weight can be easily estimated. Furthermore, by using the total number of achenes per fruit of any given fruit (in this embodiment, a fruit that has begun to enlarge), the harvest weight of other fruits can be estimated. This eliminates the need to estimate the total number of achenes per fruit for each individual fruit, making it more efficient.

[0057] Furthermore, in this embodiment, the information processing device 70 uses environmental information (measured values ​​and / or predicted values) from the flowering date to the harvest date when estimating the fruit weight at harvest. This allows for accurate estimation of the fruit weight at harvest.

[0058] Furthermore, in this embodiment, when estimating the total number of achenes per fruit, the information processing device 70 identifies the location of the achenes from the image and estimates the total number of achenes per fruit based on the number of achenes whose locations have been identified. This eliminates the need to manually count the total number of achenes per fruit, making it possible to easily estimate the total number of achenes per fruit.

[0059] Furthermore, in this embodiment, the information processing device 70 uses machine learning to identify fruits that have begun to enlarge and the fruit clusters to which those fruits belong from the images, and estimates the ranking information of the fruit clusters and the ranking information of the fruits. This eliminates the need for a human to identify the fruits or fruit clusters to be processed, thus reducing the effort required for input and other tasks.

[0060] Furthermore, if different trends are observed depending on the variety, separate graphs may be prepared for each variety in Figures 8(a), 8(b), and 9. In addition, while the above embodiment uses average temperature and cumulative solar radiation from flowering to harvest as environmental information, other environmental information (such as CO2 concentration) may also be used.

[0061] In the above embodiment, the case in which environmental information is used when estimating the fruit weight at harvest was described, but this is not the only case. In other words, the fruit weight at harvest (potential) may be used as the estimated value of the fruit weight at harvest.

[0062] In the above embodiment, the information processing device 70 automatically identifies fruits that have begun to enlarge and fruit clusters containing those fruits from the image, and automatically identifies the ranking information of the fruits and fruit clusters. However, the invention is not limited to this. That is, a person may identify fruits that have begun to enlarge and fruit clusters containing those fruits from the image. Alternatively, a person may input the ranking of each fruit and each fruit cluster.

[0063] In the above embodiment, the case in which the total number of achenes per fruit that has begun to enlarge is estimated from an image was described. However, the method is not limited to this, and a person may count and input the total number of achenes per fruit that has begun to enlarge. If a person inputs plant information (such as the number of fruits in the fruit cluster and the total number of achenes per fruit that has begun to enlarge), it becomes unnecessary to send images to the information processing device 70, and thus the camera 10 and information acquisition device 12 can be omitted.

[0064] The embodiments described above are preferred examples of the present invention. However, the invention is not limited thereto, and various modifications are possible without departing from the spirit of the invention. [Explanation of Symbols]

[0065] 10 Cameras 12 Information acquisition device 40 Image acquisition unit 41 Stock Information Acquisition Department 42 Fruit bunch identification department 44 Fruit-Specific Section 46. ​​Achene number estimation unit 48 Flowering date estimation part 50 Environmental Information Acquisition Department 52 Fruit weight estimation section 54 Output section 70 Information Processing Devices 90 CPU (Computer) 100 Strawberry Harvest Weight Estimation System

Claims

1. Obtain the number of achenes in any fruit on a strawberry plant. The following information is obtained: information on the variety of the strawberry, information identifying the fruit cluster to which the arbitrary fruit belongs, and fruit ranking information indicating the position of the arbitrary fruit in that fruit cluster. From the first information that shows the relationship between the number of achenes in a fruit determined for each variety and an index corresponding to the fruit weight at harvest, the first information corresponding to the acquired variety information is identified. Based on the identified first information and the number of achenes of the arbitrary fruit obtained, the harvest weight of the arbitrary fruit is estimated. Based on a second piece of information showing the relationship between an index corresponding to the harvest weight of the Nth fruit (where N is a natural number greater than or equal to 1) and an index corresponding to the harvest weight of the Mth fruit (where M is a natural number other than N greater than or equal to 1), and an index corresponding to the harvest weight of the arbitrary fruit, the harvest weight of the fruits included in the fruit cluster to which the arbitrary fruit belongs is estimated. Output the estimated fruit weight at harvest. A method for estimating the fruit weight of strawberries at harvest time, characterized in that the processing is performed by a computer.

2. The method for estimating the harvest weight of a strawberry according to claim 1, characterized in that, when estimating the harvest weight of the aforementioned arbitrary fruit, post-flowering environmental information, which is measured and / or predicted values ​​regarding the post-flowering environment of the aforementioned arbitrary fruit, is used.

3. The method for estimating the harvest weight of strawberries according to claim 2, characterized in that, when estimating the harvest weight of fruits included in the fruit cluster to which the aforementioned arbitrary fruit belongs, post-flowering environmental information, which is measured and / or predicted values ​​regarding the post-flowering environment of each fruit included in the fruit cluster to which the aforementioned arbitrary fruit belongs, is used.

4. The method for estimating the harvest weight of a strawberry according to claim 1, wherein in the process of obtaining the number of achenes, the position of an achene in an arbitrary fruit is identified from an image of the strawberry plant, and the number of achenes in the arbitrary fruit is obtained based on the number of achenes whose positions have been identified.

5. The method for estimating the harvest weight of strawberries according to claim 1, characterized in that, from an image of the aforementioned strawberry plant, information identifying the fruit cluster to which the aforementioned fruit belongs, and ranking information of the aforementioned fruit are obtained by machine learning.

6. The method for estimating the harvest weight of strawberries according to claim 1, characterized in that the aforementioned arbitrary fruit is a fruit in a fruit cluster that has begun to enlarge.

7. Obtain the number of achenes in any fruit on a strawberry plant. The following information is obtained: information on the variety of the strawberry, information identifying the fruit cluster to which the arbitrary fruit belongs, and fruit ranking information indicating the position of the arbitrary fruit in that fruit cluster. From the first information that shows the relationship between the number of achenes in a fruit determined for each variety and an index corresponding to the fruit weight at harvest, the first information corresponding to the acquired variety information is identified. Based on the identified first information and the number of achenes of the arbitrary fruit obtained, the harvest weight of the arbitrary fruit is estimated. Based on a second piece of information showing the relationship between an index corresponding to the harvest weight of the Nth fruit (where N is a natural number greater than or equal to 1) and an index corresponding to the harvest weight of the Mth fruit (where M is a natural number other than N greater than or equal to 1), and an index corresponding to the harvest weight of the arbitrary fruit, the harvest weight of the fruits included in the fruit cluster to which the arbitrary fruit belongs is estimated. Output the estimated fruit weight at harvest. A program for estimating the harvest weight of strawberries, characterized by having a computer perform the processing.

8. A method for obtaining the number of achenes in a strawberry fruit, Based on the image of the target fruit obtained by the imaging device, the positions of the achenes contained in the target fruit are extracted. Based on the extracted positions of the achenes, the number of countable achenes from the image is counted. The total number of achenes in the target fruit is estimated by multiplying the counted number of achenes by a correction coefficient based on the ratio of the number of achenes that can be counted from an image of the first fruit to the total number of achenes in the first fruit. A method for obtaining the number of achenes, characterized in that the processing is performed by a computer.

9. A program for obtaining the number of achenes in a strawberry fruit, Based on the image of the target fruit obtained by the imaging device, the positions of the achenes contained in the target fruit are extracted. Based on the extracted positions of the achenes, the number of countable achenes from the image is counted. The total number of achenes in the target fruit is estimated by multiplying the counted number of achenes by a correction coefficient based on the ratio of the number of achenes that can be counted from an image of the first fruit to the total number of achenes in the first fruit. A program for obtaining the number of achene fruits, characterized by having a computer perform the processing.

Images