Vision-based greenhouse tomato leaf miner trapping method

Abstract

The invention discloses a greenhouse tomato leaf miner trapping method based on vision, and relates to the technical field of greenhouse pest control, and the greenhouse tomato leaf miner trapping method comprises the following steps: obtaining an image of an insect sticking plate in a tomato leaf miner trapping device and dynamically changing temperature and humidity information; respectively determining the service life of the insect sticking plate and the service life of the lure according to the image information of the insect sticking plate and the dynamically changed temperature and humidity information; the tomato leaf miner trapping device is adjusted according to the determined service life of the insect sticking plate and the determined service life of the trapping core; and after the adjustment is completed, the tomato leaf miner outbreak probability density is determined based on any tomato leaf miner trapping device, and the collection frequency of each tomato leaf miner trapping device is dynamically adjusted. Quantitative calculation of the service life of the sticky card and the service life of the lure is realized through the acquired image information and temperature and humidity information of the sticky card, the problem of pheromone concentration reduction caused by the reduction of the service life of the lure is dynamically compensated, and the image acquisition frequency is adjusted according to the number of insects trapped by the tomato leaf miner trapping device at different positions. And the early warning accuracy of the outbreak of the tomato leaf miner is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of greenhouse pest control, in particular to a greenhouse tomato leaf roller trapping method based on vision. BACKGROUND

[0002] Tomato leaf roller, also known as tomato leaf roller, is a kind of insect in Lepidoptera and Pyralidae, which is a kind of extremely destructive alien invasive pest, causing serious damage to tomato and other solanaceous crops. The specific damage is as follows: leaf damage, larvae hiding in leaf to feed on leaf pulp, forming small tunnels at the initial stage, with strong concealment; the tunnel becomes wider at the later stage, the leaf shrinks and dries, affecting photosynthesis and causing plant growth to be hindered. Fruit damage, larvae feeding on fruit, forming holes or deformities, which are easy to be parasitized by pathogenic bacteria, leading to rot; especially at the connection between calyx and young fruit, larvae feed on fruit, causing fruit to fall off, seriously affecting yield and quality. Tender stem and terminal bud damage, larvae feeding on terminal shoots, axillary buds and tender stems, causing growth points to die, forming clumps or leaf clusters, plant development is abnormal, and even absolute yield is lost. Therefore, monitoring the outbreak period of tomato leaf roller is the key to prevention and control.

[0003] Chinese patent CN118805761A discloses a tomato leaf roller trapping device and method for preventing and controlling tomato leaf roller, which covers the insect attractant by mechanical curtain structure, reducing the waste of volatilization during non-use period (such as daytime). Chinese patent CN115885953A discloses a method for killing tomato leaf roller by ordinary single-sided sticky insect board barrel-shaped tomato leaf roller trapping device, which cooperates the trapping effect of tomato leaf roller sex pheromone lure core, the killing effect of liquid and the adhesion effect of ordinary single-sided sticky insect board, to achieve better trapping effect. Chinese patent CN218999314U discloses a simple device for trapping tomato leaf roller, which suspends centrifuge tubes on the trapping frame while combining the physiological characteristics of tomato leaf roller, installs sex pheromone lure core in the centrifuge tubes, can trap a large number of tomato leaf rollers, and provides a powerful guarantee for long-term green prevention and control of tomato leaf roller.

[0004] The above-mentioned schemes can achieve the prevention and control of tomato leaf roller, but the lure core and sticky insect board for inducing and capturing tomato leaf roller have a certain service life, which affects the analysis and early warning of regional pest occurrence probability density. SUMMARY

[0005] In order to solve the above technical problems, the present application provides the following technical scheme: In a first aspect, the present application provides a greenhouse tomato leaf roller trapping method based on vision, comprising: acquiring images of the sticky insect board in the tomato leaf roller trapping device and dynamic change information of temperature and humidity; According to the image information of the moth board and the dynamically changing temperature and humidity information, the life of the moth board and the life of the lure core are determined respectively; The tomato leaf moth trapping device is adjusted according to the determined life of the moth board and the life of the lure core; After the adjustment is completed, the collection frequency of each tomato leaf moth trapping device is dynamically adjusted based on the tomato leaf moth outbreak probability density determined by any tomato leaf moth trapping device.

[0006] In a possible implementation, the tomato leaf moth trapping device comprises a device shell, the moth board is arranged in the device shell, and the lure core is arranged on the moth board; a temperature and humidity sensor, an image acquisition device, a fan and a solar power supply module are further arranged on the device shell, and the solar power supply module is electrically connected with the temperature and humidity sensor, the image acquisition device and the fan respectively.

[0007] In a possible implementation, the moth board life and the lure core life are determined according to the image information of the moth board and the dynamically changing temperature and humidity information, comprising: The life of the moth board is determined according to the moving trace of the tomato leaf moth on the moth board in the image information of the moth board; The life of the lure core is determined according to the initial life of the lure core and the dynamically changing temperature and humidity information of the use time period.

[0008] In a possible implementation, the life of the moth board is determined according to the moving trace of the tomato leaf moth on the moth board in the image information of the moth board, comprising: The tomato leaf moth on the moth board is identified through a deep learning neural network YOLO v12 model; After the tomato leaf moth is determined, the candidate box is enlarged by a preset multiple times in turn; The moving trace of the tomato leaf moth in the candidate box enlarged in turn is detected; The failure score of the moth board is determined in combination with the number of the identified tomato leaf moths and the minimum candidate box magnification multiple of each moving trace of the tomato leaf moth; If the failure score is greater than a first preset value, the moth board cannot effectively fix the tomato leaf moth.

[0009] In a possible implementation, the failure score of the moth board is determined in combination with the number of the identified tomato leaf moths and the minimum candidate box magnification multiple of each moving trace of the tomato leaf moth, comprising: In which: is the failure score, the number of tomato leafworms on the sticky board, the minimum candidate frame magnification of the i-th detected tomato leafworm.

[0010] In a possible implementation, the determining the life of the lure core according to the dynamic change of the temperature and humidity information of the lure core in the initial life and the use time period comprises: wherein: the current life of the lure core, the initial life of the lure core, the temperature coefficient, the humidity coefficient, the temperature on the i-th day, the humidity on the i-th day, the total number of days used by the lure core.

[0011] In a possible implementation, the adjusting the tomato leafworm trapping device according to the determined life of the sticky board and the life of the lure core comprises: if the failure score is greater than a first preset value, replacing the sticky board; according to the proportion of the determined life of the lure core to the initial life of the lure core, adjusting the release rate and the induction range of the lure core by the fan.

[0012] In a possible implementation, the adjusting the release rate and the induction range of the lure core by the fan according to the proportion of the determined life of the lure core to the initial life of the lure core comprises: if the failure score is greater than a first preset value, replacing the sticky board; In the interval of (0.2 , 0.8 ), keeping the lure core unchanged in the current release rate and the induction range; or, if the failure score is greater than a first preset value, replacing the sticky board; In the interval of (0.6 , 0.8 ], starting the fan for a first preset time length every day to adjust the release rate and the induction range of the lure core; or, if the failure score is greater than a first preset value, replacing the sticky board; In the interval of (0.4 , 0.6 ], starting the fan for a second preset time length every day to adjust the release rate and the induction range of the lure core; or, if the failure score is greater than a first preset value, replacing the sticky board; In the interval of (0.2 0.4 If the fan is activated for a third preset duration each day within a certain range, the release rate and induction range of the decoy core will be adjusted. or, If the above ≤0.2 Then, the fan is activated daily for a fourth preset duration to regulate the release rate and induction range of the decoy core.

[0013] In one possible implementation, after the adjustment is completed, the tomato leafminer outbreak probability density is determined based on any tomato leafminer trap, and the collection frequency of each tomato leafminer trap is dynamically adjusted, including: Determine the probability density of tomato leafminer outbreaks at the first tomato leafminer trapping device; If the probability density of the tomato leafminer outbreak is greater than the preset probability value, the number of times the sticky insect board image is collected for each tomato leafminer trap will be increased. The number of tomato leafminers on the next day is predicted based on the first detection of the number of tomato leafminers on the sticky insect board, the occurrence trend of tomato leafminers, and the probability density of tomato leafminer outbreaks. If the number of tomato leafminer moths exceeds the preset number the next day, the number of times the sticky insect board images are collected for each tomato leafminer moth trap will be increased.

[0014] In one possible implementation, determining the probability density of tomato leafminer outbreaks at the first tomato leafminer trap includes: in: This represents the probability density of an outbreak of tomato leafminer moth. It is a two-dimensional Gaussian kernel function. For dynamic bandwidth, This is the bandwidth reference value. for, To calculate the longitude of a point, Let be the longitude of the i-th sampling point. To calculate the latitude of a point, Let be the latitude of the i-th sampling point. for , for , This is the distance coefficient. Center point ( The distance from (x, y) to (x, y).

[0015] In the embodiment of the present application, the image information and temperature and humidity information of the obtained armyworm plate are used to realize the quantitative calculation of the armyworm plate life and the lure core life, dynamically compensate the problem of pheromone concentration decline caused by the decline of lure core life, and adjust the image acquisition frequency according to the number of trapped insects of the tomato leaf moth trapping device at different positions, thereby improving the accuracy of tomato leaf moth outbreak early warning. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of a greenhouse tomato leaf moth trapping method based on vision provided by the embodiment of the present application is shown in the figure. Figure 2 A schematic diagram of a tomato leaf moth trapping device provided by the embodiment of the present application is shown in the figure. Figure 3 A schematic diagram of an armyworm plate image provided by the embodiment of the present application is shown in the figure. Figure 4 A schematic diagram for determining the moving trace of the tomato leaf moth provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0017] The present scheme will be described below in combination with the accompanying drawings and specific embodiments.

[0018] Referring to Figure 1 , the greenhouse tomato leaf moth trapping method based on vision in the embodiment includes: S101, obtaining the image of the armyworm plate and the dynamically changing temperature and humidity information in the tomato leaf moth trapping device.

[0019] In the embodiment, the tomato leaf moth trapping device and the cloud service are used to constitute the greenhouse tomato leaf moth intelligent trapping system based on vision. Referring to Figure 2 , the tomato leaf moth trapping device includes a device shell, the armyworm plate is arranged in the device shell, and the lure core is arranged on the armyworm plate; the device shell is further provided with a temperature and humidity sensor, an image acquisition device, a fan and a solar power supply module, and the solar power supply module is electrically connected with the temperature and humidity sensor, the image acquisition device and the fan respectively.

[0020] The temperature and humidity sensor is used to obtain the temperature and humidity information of the air in the greenhouse, and the image acquisition device is used to obtain the image of the armyworm plate, as shown in Figure 3 , which is a schematic diagram of the armyworm plate image. The obtained temperature and humidity information is sent to the cloud together with the armyworm plate image for processing.

[0021] When the pests are few, the image acquisition device takes pictures to collect pest images at 11 o'clock in the morning every day, the images are encoded and packaged, and then sent to the cloud through the mobile network. After receiving all the packages, the cloud replies to the confirmation frame, which includes the image code, the cloud time and the image acquisition.

[0022] S102, respectively according to the image information of the sticky insect plate and the dynamic changing temperature and humidity information, determine the life of the sticky insect plate and the life of the lure core.

[0023] After the cloud receives the image information of the sticky insect plate, the pests are classified and recognized by a deep learning neural network YOLO v12, focusing on distinguishing tomato leaf roller and other organisms (whitefly, fly, aphid fly, etc.) that can be captured by the sticky insect plate in the greenhouse.

[0024] In the high temperature and high humidity environment of the greenhouse, the sticky insect plate is prone to failure, and the life is obviously shorter than that in the open environment. The failure of the sticky insect plate will cause the pests to be unable to be effectively fixed, and the tomato leaf roller will roll on the sticky insect plate, and the scales will fall off to produce moving traces. Therefore, the adhesion of the sticky insect plate is quantitatively evaluated by the moving trace quantity of the tomato leaf roller.

[0025] The specific process is as follows: the tomato leaf roller on the sticky insect plate is identified by a deep learning neural network YOLO v12 model. After the tomato leaf roller is determined, the candidate box is enlarged by a preset multiple times in turn; the moving trace of the tomato leaf roller is detected in the candidate box after being enlarged in turn; the failure score of the sticky insect plate is determined in combination with the number of identified tomato leaf rollers and the minimum candidate box magnification of each moving trace of the tomato leaf roller. If the failure score is greater than a first preset value, the sticky insect plate cannot effectively fix the tomato leaf roller.

[0026] As shown in Figure 4 One of the selected tomato leaf rollers with moving traces is generated a candidate box, and is enlarged by 1-3 times in turn. It is found that the trace caused by the scale falling off still exists when the magnification is 2 times, but the area that is more than the candidate box when the magnification is 2 times does not exist scale falling trace when the magnification is 3 times. Therefore, it is determined that the minimum candidate box magnification of the tomato leaf roller is 2.

[0027] The minimum candidate box magnification of the moving trace of the tomato leaf roller determines the failure score of the sticky insect plate, comprising: Among them: The failure score is The number of tomato leaf rollers of the sticky insect plate is The minimum candidate box magnification of the i-th detected tomato leaf roller with moving trace. If the failure score is greater than 100, it indicates that the adhesion of the sticky insect plate has failed to effectively fix the tomato leaf roller.

[0028] For the lure core, the life of the lure core is affected by temperature and humidity, and is calculated according to the following formula: Among them: The current life of the lure core is is the initial life of the lure, is the temperature coefficient, is the humidity coefficient, is the temperature of the i-th day, is the humidity of the i-th day, is the total number of days the lure is used.

[0029] S103, adjusting the tomato leaf moth trapping device according to the determined life of the sticky board and the life of the lure.

[0030] If the failure score is greater than 100, the sticky board is replaced, and a new sticky board is used to capture tomato leaf moths. As the lure is used, its release rate will slow down, and its induction range will become smaller. Therefore, in this embodiment, the release rate and induction range of the lure are regulated by the fan according to the proportion of the determined life of the lure to the initial life of the lure.

[0031] Specifically, if the failure score is greater than 100, the sticky board is replaced, and a new sticky board is used to capture tomato leaf moths. As the lure is used, its release rate will slow down, and its induction range will become smaller. Therefore, in this embodiment, the release rate and induction range of the lure are regulated by the fan according to the proportion of the determined life of the lure to the initial life of the lure. In the interval of [0.6, 0.8], the fan is started for 1 hour per day to regulate the release rate and induction range of the lure. , 0.8 interval, the release rate and induction range of the lure are not changed. If the failure score is in the interval of (0.6 , 0.8 ], the fan is started for 1 hour per day to regulate the release rate and induction range of the lure. If the failure score is in the interval of (0.4 , 0.6 ], the fan is started for 2 hours per day to regulate the release rate and induction range of the lure. If the failure score is in the interval of (0.2 , 0.4 ], the fan is started for 3 hours per day to regulate the release rate and induction range of the lure. If the failure score is ≤0.2 , the fan is started for 4 hours per day to regulate the release rate and induction range of the lure. S104, after the adjustment, the collection frequency of each tomato leaf moth trapping device is dynamically adjusted based on the tomato leaf moth outbreak probability density determined by any tomato leaf moth trapping device. S104, after the adjustment, the collection frequency of each tomato leaf moth trapping device is dynamically adjusted based on the tomato leaf moth outbreak probability density determined by any tomato leaf moth trapping device.

[0032] S104, after the adjustment, the collection frequency of each tomato leaf moth trapping device is dynamically adjusted based on the tomato leaf moth outbreak probability density determined by any tomato leaf moth trapping device.

[0033] ​​Each tomato leafminer trap is located via a mobile app and the location is sent to the server upon setup. For each trap, the cloud first detects the number of tomato leafminers on the sticky trap, calculates the pest occurrence trend and predicts the next day's pest population (z+1) using linear interpolation based on the number of traps caught over the past three days, and then adjusts the prediction based on a ternary linear regression function of air temperature, humidity, and the next day's pest population to obtain the predicted pest population for the next day. .

[0034] The cloud-based system calculates the probability density of tomato leafminer occurrence based on the location of the tomato leafminer trap and the number of tomato leafminers, using the following formula. in: This represents the probability density of an outbreak of tomato leafminer moth. It is a two-dimensional Gaussian kernel function. For dynamic bandwidth, This is the bandwidth reference value. for, To calculate the longitude of a point, Let be the longitude of the i-th sampling point. To calculate the latitude of a point, Let be the latitude of the i-th sampling point. for , for , This is the distance coefficient. Center point ( The distance from (x, y) to (x, y).

[0035] If there were no pests previously present in the eggplant leafminer trap, images of the trapped insects will be collected daily at 11:00 AM by default. If the value is greater than 0.9, an additional image capture will be performed at 5 PM. This is to determine the predicted pest population for the following day. If the value is >30, an additional image capture will be performed at 7:00 AM.

[0036] In the embodiments of the present application, "at least one" refers to one or more, and "multiple" refers to two or two or more. The "and / or" describes the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B, which can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Wherein A, B can be singular or plural. The character " / " generally represents that the associated objects before and after are in an "or" relationship. "At least one of the following" and the like means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b and c can represent: a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, c can be single or multiple.

[0037] The above is only a specific embodiment of the present application, and any skilled person in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. The protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A vision-based method for attracting leafminer moths on greenhouse tomatoes, characterized in that, include: Acquire images of the sticky traps inside the tomato leafminer trap and dynamically changing temperature and humidity information; The lifespan of the sticky insect board and the lifespan of the lure are determined based on the image information of the sticky insect board and the dynamically changing temperature and humidity information, respectively. The tomato leafminer trapping device is adjusted by determining the lifespan of the sticky insect board and the lifespan of the lure. After adjustment, the probability density of tomato leafminer outbreaks is determined based on any tomato leafminer trap, and the collection frequency of each tomato leafminer trap is dynamically adjusted.

2. The vision-based method for attracting greenhouse tomato leafminers according to claim 1, characterized in that, The tomato leafminer trap includes a housing, a sticky insect board inside the housing, and a lure on the sticky insect board. The housing also includes a temperature and humidity sensor, an image acquisition device, a fan, and a solar power module, which are electrically connected to the temperature and humidity sensor, the image acquisition device, and the fan.

3. The vision-based method for attracting greenhouse tomato leafminers according to claim 2, characterized in that, The determination of the lifespan of the sticky insect board and the lifespan of the lure core based on the image information of the sticky insect board and the dynamically changing temperature and humidity information includes: The lifespan of the sticky insect board is determined by analyzing the movement traces of the tomato leafminer on the sticky insect board based on the image information of the sticky insect board. The lifespan of the lure is determined based on the initial lifespan of the lure and the dynamic temperature and humidity information during the usage period.

4. The vision-based method for attracting greenhouse tomato leafminers according to claim 3, characterized in that, The lifespan of the sticky insect board is determined based on the movement traces of the tomato leafminer on the sticky insect board in the image information, including: Tomato leafminer moths on sticky insect boards were identified using the YOLO v12 deep learning neural network model. Once the tomato leafminer is identified, the candidate boxes are enlarged sequentially by a preset magnification. Detect the movement traces of tomato leafminer moths within the candidate frames that have been magnified sequentially. The failure score of the sticky trap is determined by combining the number of tomato leafminers identified and the minimum candidate frame magnification of each tomato leafminer movement trace. If the failure score is greater than the first preset value, the sticky insect board cannot effectively fix the tomato leafminer.

5. The vision-based method for attracting greenhouse tomato leafminers according to claim 4, characterized in that, The determination of the sticky trap's failure score by combining the number of identified tomato leafminers and the minimum candidate box magnification of each tomato leafminer movement trace includes: in: For failure score, The number of tomato leafminers on sticky traps. is the magnification factor of the smallest candidate box for the i-th tomato leafminer that was detected to have movement traces.

6. The vision-based method for attracting greenhouse tomato leafminers according to claim 4 or 5, characterized in that, The determination of the lifespan of the decoy core based on its initial lifespan and the dynamic temperature and humidity information during its usage period includes: in: This refers to the current lifespan of the decoy core. The initial lifespan of the decoy core. For temperature coefficient, Humidity coefficient Let i be the temperature on day i. Let the humidity be on day i. The total number of days the lure core was used.

7. The vision-based method for attracting greenhouse tomato leafminers according to claim 6, characterized in that, The adjustment of the tomato leafminer trapping device by determining the lifespan of the sticky insect board and the lifespan of the lure includes: If the failure score is greater than a first preset value, the sticky insect board shall be replaced. The release rate and induction range of the decoy are controlled by the fan based on the ratio of the determined decoy's lifespan to its initial lifespan.

8. The vision-based method for attracting greenhouse tomato leafminers according to claim 7, characterized in that, The step of regulating the release rate and induction range of the decoy via the fan based on the ratio of the determined decoy's lifetime to its initial lifetime includes: If the above exist[ 0.8 If the range is specified, the decoy core will remain unchanged according to the current release rate and induction range. or, If the above In (0.6) 0.8 If the fan is activated for a preset duration each day within a certain range, the release rate and induction range of the decoy core will be adjusted. or, If the above In (0.4) 0.6 If the fan is activated for a second preset duration each day within a certain range, the release rate and induction range of the decoy core will be adjusted. or, If the above In (0.2) 0.4 If the fan is activated for a third preset duration each day within a certain range, the release rate and induction range of the decoy core will be adjusted. or, If the above ≤0.2 Then, the fan is activated daily for a fourth preset duration to regulate the release rate and induction range of the decoy core.

9. The vision-based method for attracting greenhouse tomato leafminers according to claim 1, characterized in that, After the adjustment is completed, the probability density of tomato leafminer outbreaks is determined based on any tomato leafminer trap, and the collection frequency of each tomato leafminer trap is dynamically adjusted, including: Determine the probability density of tomato leafminer outbreaks at the first tomato leafminer trapping device; If the probability density of the tomato leafminer outbreak is greater than the preset probability value, the number of times the sticky insect board image is collected for each tomato leafminer trap will be increased. The number of tomato leafminers on the next day is predicted based on the first detection of the number of tomato leafminers on the sticky insect board, the occurrence trend of tomato leafminers, and the probability density of tomato leafminer outbreaks. If the number of tomato leafminer moths exceeds the preset number the next day, the number of times the sticky insect board images are collected for each tomato leafminer moth trap will be increased.

10. The vision-based method for attracting greenhouse tomato leafminers according to claim 9, characterized in that, The calculation of the probability density of tomato leafminer outbreak using the first tomato leafminer trapping device includes: in: This represents the probability density of an outbreak of tomato leafminer moth. It is a two-dimensional Gaussian kernel function. For dynamic bandwidth, This is the bandwidth reference value. for, To calculate the longitude of a point, Let be the longitude of the i-th sampling point. To calculate the latitude of a point, Let be the latitude of the i-th sampling point. for , for , This is the distance coefficient. Center point ( The distance from (x, y) to (x, y).

Images