Monitoring system for pollinator activity in front of nest entrances

Abstract

By monitoring the activity of pollinators that leave the nest to visit flowers in front of the nest entrance, we can help prevent the occurrence of poorly fertilized fruit in fruit and vegetable crops. [Solution] The system comprises a camera unit (12) that takes an overhead shot of pollinators (P) from above a background board (19), a video analysis unit (13) that analyzes the captured video of the pollinators to detect their outlines and measure the number of pollinators active in front of the hive entrance, and an output unit (14) that displays the analysis results. The analysis area (Z) processed by the video analysis unit is divided into three zones: a return zone (Z1), a stagnant zone (Z3), and an exit zone (Z3), sequentially from the hive entrance side to the field side. The system measures the number of pollinators exiting the nest when moving from the stagnant zone to the exit zone, and the number of pollinators returning when moving from the stagnant zone to the return zone, respectively, and sets the system to display at least the number of pollinators exiting the nest as the number of pollinator flower-visiting activities.

Description

Technical Field

[0001] The present invention relates to a monitoring system for activities in front of the entrance of a polinator nest, which is useful for greenhouse cultivation of strawberries, melons, tomatoes, etc.

Background Art

[0002] A management device and method for insects for pollen mating in greenhouse cultivation of this type of plant are disclosed in Patent Document 1. This is considered to be the most similar known technology to the present invention in that it photographs insects entering and leaving the entrance of a nest box and measures the number of insects for pollen mating identified from the photographed images per day.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, the invention described in Patent Document 1 has a mechanism in which a monitoring camera is installed near the front of a nest box to photograph the entrance of the nest and its surroundings, and the entry and exit of bees to and from the entrance of the nest are measured by image analysis.

[0005] That is, it is only a device / method for grasping the number of bees active around the entrance of the nest and determining whether the bees are healthy or not, and not a device / method for determining whether the bees fly to the plant field (field) and are active.

[0006] For example, in the case of honeybees, it is known that many individuals fly out of the entrance of the nest, wander in front of the entrance without going to the field, and then return to the nest box. Even if the number of bees in front of the entrance of the nest is measured, it does not accurately grasp the flower-visiting activities of the bees necessary for fruit production.

[0007] Furthermore, the invention described in Patent Document 1 incorporates information on the activity level of bees around the beehive, investigated under various environmental conditions such as immobile and active bee conditions, and determines whether the captured bee activity level is normal or abnormal. However, it does not incorporate information on the relationship between the bees' flower-visiting activity and the occurrence of poorly fertilized fruit in strawberries, etc. It is merely a device / method that only determines whether or not there is an abnormality in the bees' activity. [Means for solving the problem]

[0008] In view of these problems, the present invention aims to monitor the activity of pollinators visiting flowers in order to prevent the occurrence of poorly fertilized fruits in fruit and vegetable crops, and in claim 1, a camera unit is installed in front of the entrance of a beehive, and a background board is placed in front of the entrance of the beehive as the background, and the camera unit takes an overhead shot of pollinators active in front of the entrance from above the background board so as not to show the entrance,

[0009] The video analysis unit reads the video of the pollinator captured by the camera unit, binarizes the difference image between the input image and the background image using a predetermined threshold, detects the outline of the pollinator, and measures the number of pollinators active in front of the nest entrance.

[0010] A monitoring system for pollinator activity in front of the nest entrance, comprising an output unit that displays the analysis results from the video analysis unit,

[0011] The analysis area required for the analysis process in the above video analysis unit is set to a rectangle the size of the background plate itself or smaller, and the analysis area is divided into three regions sequentially from the hive entrance side to the field side: the returning area, the stagnant area, and the exiting area.

[0012] Pollinators moving from the intermediate resting area to the emergence area, or from the return area to the emergence area via the resting area, are determined to have emerged from the nest. Similarly, pollinators moving from the resting area to the return area, or from the emergence area to the return area via the intermediate resting area, are determined to have returned to the nest. The number of emergences alone, or the number of emergences and returnees, are then measured.

[0013] The method is characterized by displaying the measured number of nest emergences as the number of pollinator visits.

[0014] Claim 2 is characterized in that the background plate is a hard, flat synthetic resin plate of white or other bright single color, excluding colors that have a protective color relationship with the pollinator, and is installed in front of the entrance of the beehive at a certain angle, with an upward sloping position.

[0015] Claim 3, in the video analysis program, the virtual first and second boundary lines that divide the analysis area into three regions—a homing area, a resting area, and an exiting area—are all defined as circular arc boundary lines.

[0016] The horizontal axes of these three regions are all set to be the same as the horizontal axis of the rectangular analysis region,

[0017] The vertical axis for the homing area is set to 1 / 6 of the vertical axis of the analysis area, and the vertical axis for the departure area is set to 1 / 2 of the vertical axis of the analysis area.

[0018] Furthermore, claim 4 is characterized in that the video analysis unit is set to create and update a reference background image for background difference processing from the median pixel values ​​of a certain number of frames going back from the frame to be analyzed. [Effects of the Invention]

[0019] According to the configuration of the present invention as described in claim 1, the problem of the "insect management device and management method for pollination" described in Patent Document 1 can be completely solved.

[0020] In other words, in the configuration of the present invention, pollinators active in front of the entrance of the beehive are captured from above by a surveillance camera, viewed from above a background plate. The entrance is not captured, and therefore the entrance is excluded from the analysis area for video analysis. Combined with the fact that the background of the camera image is unified by the background plate, pollinator activity can be detected accurately and without error.

[0021] It has been found that many of the pollinators that fly out of the hive box only wander in front of the hive entrance without going to the field and then return to the hive box. Therefore, even if the number of activities of the pollinators in front of the hive entrance is measured, it does not accurately grasp the flower-visiting activities of the pollinators necessary for the production of fruits and vegetables.

[0022] In this regard, in the configuration of the present invention, in the analysis process of the acquired video, three regions, namely, a homing region, a stagnant region, and a nest-exiting region, are sequentially partitioned and formed from the hive entrance side to the field side within the analysis region. Since the pollinators that move from the stagnant region in the middle part to the nest-exiting region are detected as individuals going to visit flowers and the number of nest-exiting is counted, it is possible to reliably grasp the number of flower-visiting activities of the pollinators necessary for normal pollination such as strawberry cultivation, which helps to prevent the occurrence of poorly pollinated fruits.

[0023] In strawberry cultivation and the like, it is required to manage the appropriate number of activities of pollinators according to the number of blooming flowers in the field (farmland). Based on the highly accurate measurement data of the number of activities (at least the number of nest-exiting going to visit flowers) obtained by the above monitoring system of the present invention, it is possible to confirm the gap with the appropriate number of activities of the pollinator group in the hive box, and it becomes possible to take effective measures against the occurrence of poorly fertilized fruits.

Brief Description of the Drawings

[0024] [Figure 1] It is a schematic inclined view showing the facilities of the monitoring system according to a preferred embodiment of the present invention. [Figure 2] It is also a front view of the facilities of the monitoring system seen from the front of the hive box. [Figure 3] It is a side view seen from the right direction of FIG. 2. [Figure 4] It is a plan view seen from above FIG. 2. [Figure 5] It is an overall functional block diagram of the above monitoring system. [Figure 6] It is a planar schematic diagram for explaining the acquisition of the analysis region by the AR marker and the three regions partitioned in the analysis region. [Figure 7]This is a schematic plan view illustrating the extent of the stagnation zone secured in the intermediate part of the three regions mentioned above. [Figure 8] This is a schematic diagram illustrating a pollinator detection method using background subtraction and binarization. [Figure 9] This is a schematic diagram illustrating the method for acquiring contours and determining their entry and exit points in a pollinator. [Figure 10] This is a schematic diagram illustrating the measurement method for leaving and returning to the nest in pollinators. [Figure 11] (a) and (b) are state transition diagrams using the trajectory of the pollinator. [Figure 12] This graph shows the measurement results of pollinator activity in front of the nest entrance. [Figure 13] This graph shows the relationship between the flowering time of honeybees and bumblebees and the occurrence of poorly fertilized fruit. [Figure 14] (a) is a graph showing the changes in the average daily flowering time and nesting time on the first day, and (b) is a graph showing the changes in the average daily flowering time and nesting time on the second day. [Figure 15] Both (a) and (b) are flowcharts for determining countermeasures using a monitoring system, but (a) shows the case where the number of pollinator activities is below a threshold, and (b) shows the case where the number of pollinator activities is above a threshold. [Modes for carrying out the invention]

[0025] Preferred embodiments of the present invention will be described in detail below with reference to the drawings. Figures 1-4 show schematic hardware of a pollinator pre-entrance activity monitoring system according to the embodiment, where (10) is a beehive, which is installed in a greenhouse, hothouse, or other facility (not shown) for cultivating fruit and vegetable crops such as strawberries, melons, and tomatoes, but may also be installed outside the cultivation facility with its entrance (11) opening towards the inside of the cultivation facility.

[0026] The pollinator activity monitoring system in front of the hive entrance consists of a pollinator (P) (in the illustrated embodiment, a honeybee or a bumblebee) (hereinafter sometimes simply referred to as a bee) as shown in the functional block diagram of Figure 5, an analysis unit (13) for the captured video, and an output unit (14). It is equipped with a general-purpose computer (15), preferably a small microcontroller such as a Raspberry Pi, on which a program for setting shooting conditions, a shooting control program, an analysis program, and an analysis result display control program for executing these functions are installed.

[0027] As illustrated in Figures 1-4, the surveillance camera (16) forming the above-mentioned shooting section (12) is mounted at the upper central position of a support frame (F) consisting of a gate-shaped support column (17) and its rectangular base frame (18), and a background plate (19) that serves as the background for the shooting is attached to the base frame (18).

[0028] In other words, the surveillance camera (16) is positioned at a certain distance (for example, 50 cm) (H) above the center of the background plate (19), allowing it to view the background plate (19) from above. In addition, a computer (15) equipped with a CPU (Central Processing Unit) that serves as the video analysis unit (13) is installed within the base frame (18) of the support frame (F), and a monitor (20) such as an LCD that serves as the output unit (14) is mounted at an intermediate height position on the gate-shaped support column (17) of the same support frame (F).

[0029] Furthermore, the surveillance camera (16) can be any digital camera equipped with an imaging sensor such as a CMOS image sensor or a CCD image sensor, and the video of the bee (P) captured by it can be input (transmitted) to the computer (15).

[0030] The background panel (19) described above is preferably a single color, excluding yellow and orange which would provide camouflage for the bee (P), and also excluding dark colors similar to the black color in which the bee (P) appears. This is done in order to unify the background for the surveillance camera (16) and to allow the surveillance camera (16) to clearly capture the outline (figure) of the bee (P) as it passes (flies) through the space above and below the surveillance camera (16). Among these, white is the most preferable.

[0031] Furthermore, it is desirable to use a flat plate made of melamine resin, acrylic resin, or other hard resin as the background plate (19) so that it is easy to wipe off the feces and other viscous dirt excreted by the bees (P), and to set it up on a slope that is slightly upward at a certain angle (for example, 10 degrees) (α), so that weakened bees (P) excreted from the beehive (10) can easily fall off and their carcasses do not remain on the background plate (19).

[0032] Then, the shooting condition setting program for the shooting unit (12) shown in Figure 5 sets the camera image preview, FPS, resolution, shooting time, and various other shooting conditions, and saves these settings. Similarly, the shooting control program for the shooting unit (12) that performs video shooting and video saving records the video under the previously set conditions and time, and saves the video of the area in front of the hive entrance that was filmed.

[0033] Furthermore, the analysis program that performs video processing in the analysis unit (13), also shown in Figure 5, loads (inputs) the video saved by the aforementioned shooting control program into the analysis program and obtains video information such as the total number of frames, FPS (Frames Per Second), resolution, and video shooting time.

[0034] In this case, when acquiring the analysis area using AR markers in the analysis program, the AR markers appearing in the loaded video are detected, and the analysis area is identified based on these. As shown in Figures 6 and 7, the AR markers (21) are attached to the background plate (19) in three directions other than the hive entrance (11) side (a total of three points: one on the top and one on each side). The largest inscribed rectangle of these markers is defined as the analysis area (Z), and cropping is performed.

[0035] In the early morning during winter, insufficient light may prevent the recognition of AR markers (21). Therefore, this process is performed on the first frame in which all AR markers (21) are recognized, and this is used as the reference background acquisition (frame loading and cropping).

[0036] The acquired analysis region (Z) is set to be smaller than the size (area) of the background plate (19). However, if it is possible to capture only the background plate (19) during shooting, it is not necessary to attach and use the AR marker (21), and the analysis region (Z) can be set to be the same size (area) as the background plate (19). However, in any case, the analysis region (Z) does not include the hive entrance (11) itself, and the settings are made so that the hive entrance (11) is not photographed. This is to eliminate disturbances other than the background plate (19) from the captured image.

[0037] Among the bees (P) that fly out of the hive (10), some leave the hive completely to visit flowers, while others fly around the vicinity of the hive entrance (11) for learning flights and then return to the hive (10). Furthermore, there is a tendency for these returning bees to linger temporarily before returning to the hive (10) to prevent them from crossing paths at the hive entrance (11).

[0038] Therefore, in order to correctly determine whether an animal has left the nest or returned to it, and to prevent false detection or mismeasurement of an individual, the following three regions are defined within the analysis region (Z) in the analysis program. When analyzing with this program, the boundaries of the three regions are drawn on the video to visualize the process, making it easier to check the processing steps.

[0039] In this regard, for the sake of clarity and ease of explanation, the first and second boundary lines (22) and (23) shown in Figures 6 and 7 are displayed within the analysis area (Z) on the background plate (19). However, as mentioned above, the first and second boundary lines (22) and (23) are not actually drawn on the background plate (19), but are purely virtual. These virtual first and second boundary lines (22) and (23) divide the analysis area (Z) into three zones: a return zone (Z1), a stagnant zone (Z3), and an exit zone (Z2), sequentially from the hive entrance side to the field (field) side. By securing a stagnant zone (Z3) in the middle, the system is set up to grasp the activity of pollinators (P) necessary for the pollination of fruit and vegetable crops.

[0040] Specifically, in order to measure only individuals that go to visit flowers as leaving the nest, a resting area (an area where bees temporarily rest in front of the nest entrance and fly) (Z3) is established within the above analysis area (Z), and the determination of leaving the nest is set to be based on the outer area far from the nest entrance (11) (the leaving area) (Z2). On the other hand, the determination of returning to the nest is based on the inner area close to the nest entrance (11) (the returning area) (Z1), and bees (P) that enter the returning area (Z1) from the above resting area (Z3) are considered.

[0041] In this case, the entrance (11) through which the bees (P) enter and exit has an opening width (W) of only about 2-3 cm relative to the analysis area (Z), and since flight in the stagnant area (Z3) tends to occur at a certain distance centered on the entrance (11), the three areas (Z1), (Z2), and (Z3) are partitioned by two arcs (boundary lines / dividing lines) (22) and (23) centered on the entrance (11).

[0042] Furthermore, the first arc-shaped boundary line (22) that separates the inner homing area (Z1) close to the hive entrance (11) from the intermediate stagnant area (Z3) has its horizontal axis (X) set to the same value as that of the analysis area (Z), and its vertical axis (Y / 6) set to 1 / 6 the length of the vertical axis (Y) of the analysis area (Z). This distance allows for the detection of the flight speed of the bees (P) under the condition of 30 FPS while eliminating the bees (P) that are stagnating in front of the hive entrance.

[0043] Furthermore, the arc-shaped second boundary line (23) that separates the outer emergence area (Z2), farther from the hive entrance (11), from the stagnant area (Z3) has its horizontal axis (X) set to the same value as that of the analysis area (Z), and its vertical axis (Y / 2) set to half the length of the vertical axis (Y) of the analysis area (Z). This is the distance at which the flight speed of the bees (P) at the time of emergence can be captured under the same condition of 30 FPS.

[0044] However, the size (area) of the three regions (Z1), (Z2), and (Z3) separated by the arc-shaped first and second boundary lines (22) and (23) can be changed according to the size (area) of the analysis region (Z). Furthermore, the planar contour shape of the three separated regions, especially the stagnant region (Z3), is not limited to an arc shape or crescent shape separated by two arc-shaped boundary lines (22) and (23) with different radii of curvature, but may also be an isosceles triangle, an isosceles trapezoid, or other symmetrical planar contour shape centered on the hive entrance (11), separated by straight first and second boundary lines (22) and (23).

[0045] The video recording conditions for the bee (P) using the surveillance camera (16) are not limited to the 30 FPS condition described above. In the tracking process used for measuring nest departure and return described later, it is necessary to capture the flying bee (P) within the analysis area (Z) for at least two frames. Any FPS setting that satisfies this requirement is acceptable, and 20 FPS or less is acceptable, but 30 FPS or higher is preferable.

[0046] Next, the analysis unit (13) in the above monitoring system uses the background subtraction method for its analysis process. By taking the difference between the acquired reference background image (24) and the analysis image (25) of the frame to be analyzed, a difference image (26) is obtained, which extracts elements that differ from the reference background image (24), as shown in Figure 8. In this difference image (26), the areas that differ from the reference background image (24) are emphasized more intensely the greater the difference.

[0047] Therefore, in order to remove noise (tiny changes in brightness) from the difference image (26), binarization is performed to extract areas that are emphasized with an intensity of 40 or higher. The outlines (shapes) of the emphasized areas obtained by this binarization process are detected, and those with small outline sizes (areas) are considered to be unnecessary elements such as dust and are excluded (exclusion is performed using an area threshold), thereby obtaining a binarized image (27).

[0048] Based on the background subtraction method and binarization process described above, once the outline (shape) of the pollinator (P) is detected, as shown in Figure 9, the movement path of the pollinator (P) is estimated from the closest position between the outline recorded in the previous frame and the outline detected in the current frame, and its trajectory (movement path) is tracked.

[0049] The tracking process, though its details are omitted from the illustration, detects the outline (shape) of the bee (P) and identifies the centroid coordinates of the smallest bounding rectangle of that outline as the center of the bee (P).

[0050] If the outline of the previous frame and the outline of the current frame are far apart, the system assumes that it is the same bee (P) moving if the distance is within a threshold (e.g., threshold: 300px), and assumes that it is a different, new bee if the distance is greater than the threshold.

[0051] Then, as shown in Figure 10, when the migration path moves from the stationary area (Z3) in the analysis region (Z) on the background plate (19) through the second boundary line (23) to the outer nesting area (Z2), that individual is measured as having left the nest. Similarly, when it moves from the stationary area (Z3) through the first boundary line (22) to the inner homing area (Z1), it is measured as having returned to the nest.

[0052] In other words, the trajectory log of each individual is taken for two frames before and after, and the departure and return to the nest are recorded from this information. When the individual passes the boundary line (23) between the two areas (Z2) and (Z3), as shown by the movement arrows in Figures 10 and 11, it is measured as a departure. This method has the advantage of not increasing the load on the analysis process and preventing a decrease in processing speed. However, as shown by the dashed arrow in Figure 11(a) and the arc arrow in Figure 11(b), there is a risk of incorrectly measuring too many departures due to the presence of pollinators (P) that flutter along the second boundary line (23) between the departure area (Z2) and the stagnant area (Z3).

[0053] Therefore, it is preferable to record the pollinator's (P) trajectory for all frames and measure that the bird has left the nest when it passes through the two first and second boundary lines (22) and (23), as shown by the movement line (straight line) arrows in Figure 11(b), from the homing area (Z1) through the resting area (Z3) to the leaving area (Z2). This makes the analysis process more computationally intensive, but it allows for highly accurate measurement of the bird leaving the nest.

[0054] Furthermore, although the analysis processing becomes more complex when measuring homing, it is preferable to measure homing when the pollinator (P) moves in the opposite direction to the movement line (straight line) arrow in Figure 11(b), passing through the two first and second boundary lines (22) and (23) from the emergence area (Z2) through the intermediate resting area (Z3) to the homing area (Z1). Since the movement of the pollinator (P) is slower when returning to the nest than when leaving, a higher measurement accuracy can be obtained.

[0055] However, in the illustrated embodiment, the stationary area (Z3) of the analysis region (Z) is divided into an arc shape or crescent shape by two circular boundary lines (22) and (23), and since a narrow planar contour shape is adopted at both the left and right ends, it is extremely difficult to detect the position of pollinators (P) that instantaneously pass through the inside of the stationary area (Z3) at these narrow left and right ends. Therefore, depending on the division shape of the stationary area (Z3), it is possible to rarely determine that a pollinator (P) moving from the homing area (Z1) to the out-of-nest area (Z2) is leaving the nest, and that a pollinator (P) moving from the out-of-nest area (Z2) to the homing area (Z1) is returning to the nest, and then measure the number of individuals.

[0056] As part of the analysis process in the analysis unit (13), once the detection of pollinator (P) contours and the detection of the number of individuals leaving and returning to the nest using the background subtraction method described above has been completed for 10 frames, the reference background is updated every 10 frames, as shown in Figure 5. For this update, images from 10 frames prior, 20 frames prior... up to 50 frames prior are extracted at equal intervals, and a background image (median image) is created using the median of this group of images, which is then used as the reference background image (24) for updating. The number of frames traced back from the frame being analyzed can also be changed depending on the FPS of the surveillance camera (16).

[0057] For example, if a dead bee, debris, or other unchanging foreign object appears on the background plate (19) during shooting, the system treats it as part of the background and updates the reference background image (24) to prevent the system from continuously detecting that object. Since the reference background image (24) used for updating is created from the median pixels of multiple frames preceding the frame being analyzed, shadows and dirt on the single-color background plate (19), including white, do not lead to a decrease in analysis accuracy.

[0058] For data storage in the analysis unit (13), the above analysis results are accumulated and saved as CSV files or image data. The number of active pollinators (P) and the number of individuals leaving and returning to the nest can be obtained on a frame-by-frame basis, and these can be aggregated and converted into units of seconds, minutes, hours, days, etc., for recording. In addition, if there is an excessive detection in a single frame (for example, threshold: 12 or more individuals), it is considered possible that an abnormal recording is captured in front of the nest entrance, so the image of that frame is saved.

[0059] Furthermore, the output unit (14) of the monitoring system outputs the obtained analysis results to be displayed on a monitor (20) such as an LCD. It is also possible to display time-series data as a graph on an LCD or web display other than an LCD monitor. Figure 12 shows an example of the analysis results, displaying the cumulative number of activities per day (total number of departures and returns) within the analysis area in front of the hive entrance. The display and transfer of the analysis results data are handled by the result display control program. After the program finishes, it waits until the start time of shooting the next day, and then video shooting is performed again based on the shooting control program in the shooting unit (12), after which the analysis flow in Figure 5 is repeated.

[0060] According to the experimental data in Figure 12, the number of bees detected active in front of the hive entrance was highest under sunny conditions with long daylight hours, and the number of bees detected tended to be lower under other conditions with shorter daylight hours. It was also confirmed that bee activity tends to start around 7:00 AM and continue until around 6:00 PM.

[0061] Furthermore, regarding honeybees and bumblebees, which are pollinators in strawberry cultivation, we investigated the relationship between the time spent visiting flowers and the occurrence of poorly fertilized fruit. As can be seen from the total time spent visiting flowers per flower shown in the graph in Figure 13, when the time spent visiting flowers by bees is short, poorly fertilized fruit occurs due to insufficient pollination. On the other hand, when the time spent visiting flowers is long, the occurrence of poorly fertilized fruit is low, but excessive visiting can damage the pistil and may lead to poorly fertilized fruit.

[0062] Furthermore, we investigated the changes in the average daily flower-visiting time and the average daily time spent leaving the nest for two periods, the first and second days, which had different daylight hours. As suggested by the graphs in Figures 14(a) and 14(b), there was a close relationship between the flower-visiting time and the nest-leaving time, with a near-perfect correspondence. We observed a tendency for the number of bees active in front of the nest entrance, and especially the number of bees leaving the nest, to increase as the average daily flower-visiting time increased.

[0063] Therefore, by simply monitoring the number of active bees (P) within the analysis area (Z) in front of the hive entrance, and at least the number of bees (P) that move from the stagnant area (Z3) in the analysis area (Z) to the exit area (Z2), it becomes possible to confirm whether or not those bees (P) are visiting flowers (number of flower-visiting activities).

[0064] In other words, in the monitoring system with the above configuration according to the illustrated embodiment, the entrance (hive entrance) (11) of the beehive (10), where bees (P) tend to swarm and where individuals may cross paths, is prone to false detections and duplicate measurements. Therefore, the hive entrance (11) is excluded from the analysis area (Z) in the video analysis, and the monitoring camera (16) is set to take an overhead shot from above the background plate (19) so as not to photograph the hive entrance (11).

[0065] In this case, the background plate (19) is set as a white flat plate and is installed in front of the hive entrance (11) at a certain angle (α) upward sloping forward. This makes it easy to wipe away dirt from the background plate (19) and prevents the accumulation of dead bees (P) and other debris. Even if dirt, dead bees, other unchanging foreign objects, or shadows appear on the white background plate (19), a reference background image for background subtraction processing is generated and updated from the median pixel values ​​of several frames preceding the frame being analyzed. This prevents a decrease in analysis accuracy caused by the aforementioned foreign objects and shadows.

[0066] Furthermore, the analysis region (Z) is set to a size (area) smaller than the background plate (19) by using AR markers (21) attached to the three directions other than the hive entrance (11) side, as the largest inscribed rectangle. In addition, in order to visualize the video analysis process in an easy-to-understand manner, the analysis region (Z) is divided into three areas in the analysis program by virtual first and second arc boundary lines (22) (23): the homing area (Z1), the stagnant area (Z3), and the departure area (Z2).

[0067] Then, using a stationary area (Z3) secured in the middle of these three regions as a reference, individual bees (P) moving from the stationary area (Z3) to the emergence area (Z2) are detected as those flying off to visit flowers, and the number of those emerging is measured. On the other hand, individuals moving from the stationary area (Z3) to the return area (Z1) are detected as, so to speak, those returning from visiting flowers, and the number of those returning is measured. Individual bees (P) remaining stationary in the stationary area (Z3) are not detected, and their number is not measured.

[0068] In this way, by reliably detecting only individual bees (P) that are visiting flowers and obtaining high measurement accuracy, it is possible to monitor the number of bees active in front of the hive entrance, either by the number of bees leaving the hive or the sum of the number of bees leaving and returning, and to check for gaps with the appropriate number of pollinator activity. This can be useful in preventing the occurrence of poorly fertilized fruit in strawberry cultivation and for taking countermeasures. Figures 15(a) and 15(b) illustrate the countermeasures and decision-making flow. [Explanation of Symbols]

[0069] (10) .....Nest box (11)...Sumon (entrance / exit) (12)...Photography Department (13)...Analysis department (14) Output section (15) Computer (16) Surveillance cameras (19)...Background board (20) Monitor (21) AR Marker (22) ····First boundary line (23)...Second boundary line (24) ····Reference background image (25)...Analysis image (26) Difference image (27) Binarized image (F)·····Support frame (H) Camera installation height (P)·····Pollinator (bee) (W)·····Width of the hive entrance (X)·····Horizontal axis (Y)·····Vertical axis (Z)...Analysis area (Z1)...Homing area (Z2)...Nesting area (Z3)...Stagnation area (α)·····Angle of the background board

Claims

1. A camera crew will use a background board placed in front of the hive entrance to film pollinators active in front of the entrance, from above the background board, ensuring the entrance is not visible in the shot. The video analysis unit reads the video of the pollinator captured by the camera unit, binarizes the difference image between the input image and the background image using a predetermined threshold, detects the outline of the pollinator, and measures the number of pollinators active in front of the nest entrance. A monitoring system for pollinator activity in front of the nest entrance, comprising an output unit that displays the analysis results from the video analysis unit, The analysis area required for the analysis process in the above video analysis unit is set to a rectangle the size of the background plate itself or smaller, and the analysis area is divided into three regions sequentially from the hive entrance side to the field side: the return area, the resting area, and the departure area. Pollinators moving from the resting area secured in the intermediate section to the emergence area, or from the return area to the emergence area via the resting area, are determined to have emerged from the nest. Similarly, pollinators moving from the resting area to the return area, or from the emergence area to the return area via the resting area, are determined to have returned to the nest. The number of emergences alone, or the number of emergences and returnees, are then measured. A pollinator activity monitoring system in front of the nest entrance, characterized by displaying the measured number of at least the number of nest departures as the number of pollinator flower-visiting activities.

2. The pollinator activity monitoring system in front of the entrance of a hive according to claim 1, characterized in that the background plate is a hard, synthetic resin flat plate of white or other bright single color, excluding colors that provide camouflage with the pollinator, and is installed in front of the entrance of the hive at a certain angle of upward sloping.

3. In the video analysis program, the analysis area is divided into three regions: the homing area, the resting area, and the leaving area. The first and second hypothetical boundary lines are both defined as circular arcs. The horizontal axes of these three regions are all set to be the same as the horizontal axis of the rectangular analysis region, The pollinator activity monitoring system in front of the nest entrance according to claim 1, characterized in that the vertical axis of the homing area is set to 1 / 6 of the vertical axis of the analysis area, and the vertical axis of the departure area is set to 1 / 2 of the vertical axis of the analysis area.

4. The pollinator activity monitoring system in front of the nest entrance according to claim 1, characterized in that the video analysis unit is configured to create and update a reference background image for background subtraction processing from the median pixel values ​​of a certain number of frames going back from the frame to be analyzed.

Images