Intelligent drug delivery method and system for fruit tree leaves

Abstract

The invention discloses an intelligent drug delivery method and system for fruit tree leaves. The method comprises the following steps: receiving a fruit tree leaf picture collected by a high-definition camera arranged in a tree-side monitoring drug administration device; inputting the fruit tree leaf picture into a preset OfficientNet-B1 convolutional neural network model to obtain an identification result matched with the fruit tree leaf picture; an identification result is sent to an upper computer; receiving a drug administration instruction which is sent by the upper computer and corresponds to the identification result; a drug administration module is controlled to spray drugs according to the drug administration instruction, the drug administration module is arranged in the tree-side monitoring drug administration device, and the drug administration module at least comprises a drug storage box, an electric control adjustable nozzle, a piezoelectric micro pressure pump and a control circuit; and the control circuit controls the direction of the electrically-controlled adjustable nozzle and controls the piezoelectric micro pressure pump to push the medicament in the medicament bottle out to the electrically-controlled adjustable nozzle to be sprayed out. The technical problems that in the prior art, the fruit tree leaf disease monitoring and drug administration efficiency is low, and the drug administration intelligence degree is poor are solved.

Description

Technical Field

[0001] This invention relates to the agricultural field, and more specifically, to an intelligent method and system for administering fruit tree leaves. Background Technology

[0002] In fruit tree cultivation, the leaves serve as a direct barometer of the tree's health. Changes in leaf color, shape, texture, and attached substances directly reflect the tree's nutritional level, pest and disease infestation, and environmental stresses (such as light, water, and temperature). Leaf diseases are a key factor affecting fruit yield and quality. Common diseases such as powdery mildew, rust, and leaf spot, if not controlled in time, can spread rapidly and lead to significant yield reductions. Currently, the control of leaf diseases in fruit trees mainly relies on manual inspection and experience-based medication: growers conduct regular field inspections, visually observing changes in leaf morphology and color to determine disease status and type, and then select pesticides, determine spraying dosage and method based on experience, using manual sprayers or large machinery to complete the application. With the development of large-scale and intensive orchards, manual inspection suffers from low efficiency and limited coverage, especially in large orchards, making early disease detection difficult. Furthermore, manual judgment is easily influenced by experience and subjective factors, leading to misdiagnosis or missed diagnosis, resulting in the selection of incorrect pesticides or inappropriate dosages. Meanwhile, traditional drug administration methods mostly involve spraying the entire area, which cannot accurately apply the drug to diseased leaves. This not only wastes the drug and increases planting costs, but also easily leads to excessive pesticide residues and environmental pollution. Furthermore, the lack of efficient linkage with the drug administration system makes it difficult to achieve intelligent prevention and control that integrates identification and drug administration.

[0003] In summary, existing technologies suffer from low efficiency in monitoring and administering medication for fruit tree leaf diseases, and poor level of intelligent medication administration.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This invention provides an intelligent drug delivery method and system for fruit tree leaves, which at least solves the technical problems of low drug delivery efficiency and poor intelligence in the existing technology for monitoring fruit tree leaf diseases.

[0006] According to one aspect of the present invention, an intelligent drug delivery method for fruit tree leaves is provided. The method includes: receiving fruit tree leaf photos captured by a high-definition camera installed in a tree-side monitoring drug delivery device; inputting the fruit tree leaf photos into a preset EfficientNet-B1 convolutional neural network model to obtain a matching recognition result of the fruit tree leaf photos, wherein the preset EfficientNet-B1 convolutional neural network model is used to identify leaf disease symptoms corresponding to the fruit tree leaf photos, and the preset EfficientNet-B1 convolutional neural network model is configured in an artificial intelligence recognition chip or human-computer interface installed in the tree-side monitoring drug delivery device. In the AI ​​recognition software, the recognition result is sent to the host computer, which is network-connected to the tree-side monitoring and drug delivery device. The software receives a drug delivery command corresponding to the recognition result from the host computer. Based on the drug delivery command, the software controls the drug delivery module to spray the drug. The drug delivery module is located in the tree-side monitoring and drug delivery device and includes at least a drug storage box, an electrically adjustable nozzle, a piezoelectric micro-pressure pump, and a control circuit. The drug storage box contains at least one drug bottle. The control circuit controls the direction of the electrically adjustable nozzle and controls the piezoelectric micro-pressure pump to push the drug from the drug bottle into the electrically adjustable nozzle.

[0007] Furthermore, the receiving of fruit tree leaf photos acquired by the high-definition camera installed in the above-mentioned tree-side monitoring and drug delivery device includes: receiving the fruit tree leaf photos acquired by the high-definition camera in a multi-angle shooting mode within a preset acquisition period, wherein the preset acquisition period dynamically corresponds to the fruit tree growth stage, and the multi-angle shooting mode includes a horizontal angle shooting mode, an upward angle shooting mode, and a downward angle shooting mode; preprocessing the fruit tree leaf photos, wherein the preprocessing includes at least one of the following: cropping the leaf area through an edge detection algorithm, removing image noise through Gaussian filtering, and balancing the image contrast under different lighting conditions through adaptive brightness adjustment.

[0008] Furthermore, the above method also includes: constructing an ImageNet training dataset, which contains multiple images of the aforementioned fruit tree leaves, each image of which is labeled with leaf disease type parameters and leaf disease severity parameters; performing data augmentation processing on the ImageNet training dataset, including random rotation, random scaling, random horizontal flipping, random cropping, and adding Gaussian noise to obtain a target ImageNet training dataset; fine-tuning and validating the preset EfficientNet-B1 convolutional neural network model pre-trained on the target ImageNet training dataset using transfer learning; and, if the preset EfficientNet-B1 convolutional neural network model passes the model performance validation, deploying the preset EfficientNet-B1 convolutional neural network model into the aforementioned artificial intelligence recognition chip or the aforementioned artificial intelligence recognition software.

[0009] Furthermore, the above method also includes: inputting a disease-pesticide matching database into the host computer, wherein the disease-pesticide matching database stores dosing instruction parameters corresponding to the leaf disease symptoms, and the dosing instruction parameters include at least one of the following: recommended agent type, optimal concentration ratio, single dose, dosing frequency, and safety interval; the host computer retrieves the dosing instruction corresponding to the identification result from the disease-pesticide matching database based on the leaf disease type parameter and the leaf disease severity parameter in the identification result.

[0010] Furthermore, the above-mentioned drug spraying control module according to the above-mentioned drug administration command includes: the control circuit controls the electric valve of the drug bottle corresponding to the drug type parameter in the drug storage box to open according to the drug type parameter in the above-mentioned drug administration command, and then monitors the drug output in real time through the flow sensor built into the drug bottle; when the above-mentioned drug administration command is a multi-drug mixing command, the control circuit controls multiple target drug bottles to synchronously output drugs to the mixing chamber according to a preset ratio, and the piezoelectric micro pressure pump pushes the drugs in the mixing chamber to the electrically controlled adjustable nozzle for spraying, wherein the mixing chamber is set in the above-mentioned drug administration module and is respectively connected to the multiple target drug bottles and the electrically controlled adjustable nozzle.

[0011] According to another aspect of the present invention, an intelligent drug delivery system for fruit tree leaves is also provided. The system includes: a first receiving unit for receiving fruit tree leaf photos captured by a high-definition camera installed in a tree-side monitoring drug delivery device; and a recognition unit for inputting the fruit tree leaf photos into a preset EfficientNet-B1 convolutional neural network model to obtain a recognition result matching the fruit tree leaf photos. The preset EfficientNet-B1 convolutional neural network model is used to identify leaf disease symptoms corresponding to the fruit tree leaf photos. The preset EfficientNet-B1 convolutional neural network model is configured within an artificial intelligence recognition chip or artificial intelligence recognition device installed in the tree-side monitoring drug delivery device. The software includes: a sending unit for sending the identification result to a host computer, wherein the host computer is network-connected to the tree-side monitoring drug delivery device; a second receiving unit for receiving a drug delivery instruction corresponding to the identification result sent by the host computer; and a first processing unit for controlling the drug delivery module to spray the drug according to the drug delivery instruction. The drug delivery module is installed in the tree-side monitoring drug delivery device and includes at least a drug storage box, an electrically adjustable nozzle, a piezoelectric micro pressure pump, and a control circuit. The drug storage box contains at least one drug bottle, and the control circuit controls the direction of the electrically adjustable nozzle and controls the piezoelectric micro pressure pump to push the drug from the drug bottle into the electrically adjustable nozzle for spraying.

[0012] Furthermore, the first receiving unit includes: a receiving subunit, used to receive the fruit tree leaf photos acquired by the high-definition camera in a multi-angle shooting mode within a preset acquisition period, wherein the preset acquisition period dynamically corresponds to the fruit tree growth stage, and the multi-angle shooting mode includes a horizontal angle shooting mode, an upward angle shooting mode, and a downward angle shooting mode; and a processing subunit, used to preprocess the fruit tree leaf photos, wherein the preprocessing includes at least one of the following: cropping the leaf area using an edge detection algorithm, removing image noise using Gaussian filtering, and balancing image contrast under different lighting conditions through adaptive brightness adjustment.

[0013] Furthermore, the system further includes: a construction unit for constructing an ImageNet training dataset, wherein the ImageNet training dataset contains multiple images of the aforementioned fruit tree leaves, each image of which is labeled with leaf disease type parameters and leaf disease severity parameters; a second processing unit for performing data augmentation processing on the ImageNet training dataset, wherein the data augmentation processing includes random rotation, random scaling, random horizontal flipping, random cropping, and adding Gaussian noise to obtain a target ImageNet training dataset; a verification unit for fine-tuning and verifying the performance of the preset EfficientNet-B1 convolutional neural network model pre-trained on the target ImageNet training dataset using a transfer learning method; and a deployment unit for deploying the preset EfficientNet-B1 convolutional neural network model to the aforementioned artificial intelligence recognition chip or the aforementioned artificial intelligence recognition software, provided that the preset EfficientNet-B1 convolutional neural network model has passed the aforementioned model performance verification.

[0014] Furthermore, the system further includes: an input unit for inputting a disease-pesticide matching database into the host computer, wherein the disease-pesticide matching database stores dosing instruction parameters corresponding to the leaf disease symptoms, and the dosing instruction parameters include at least one of the following: recommended agent type, optimal concentration ratio, single dose, dosing frequency, and safety interval; and a retrieval unit for retrieving, based on the leaf disease type parameter and leaf disease severity parameter in the identification result, the dosing instruction corresponding to the identification result from the disease-pesticide matching database.

[0015] Furthermore, the first processing unit includes: a first control subunit, used by the control circuit to control the electric valve of the medicine bottle corresponding to the medicine type parameter in the medicine storage box to open according to the medicine type parameter in the medicine administration instruction, and then monitor the medicine output in real time through the flow sensor built into the medicine bottle; a second control subunit, used by the control circuit to control multiple target medicine bottles to synchronously output medicine to the mixing chamber according to a preset ratio when the medicine administration instruction is a multi-medication mixing instruction, and the piezoelectric micro pressure pump pushes the medicine in the mixing chamber to the electrically controlled adjustable nozzle for spraying, wherein the mixing chamber is disposed in the medicine administration module and is respectively connected to the multiple target medicine bottles and the electrically controlled adjustable nozzle.

[0016] In this embodiment of the invention, a tree-side monitoring and drug delivery device is used to collect photos of fruit tree leaves using its own high-definition camera. These photos are then input into a preset EfficientNet-B1 convolutional neural network model in the artificial intelligence recognition chip or software installed in the device. The identification results of leaf disease symptoms are obtained and sent to a network-connected host computer. Subsequently, the device receives the corresponding drug delivery command from the host computer. Under the control of the control circuit, the drug delivery module in the tree-side monitoring and drug delivery device pushes the corresponding agent through a piezoelectric micro-pressure pump and sprays it through an adjustable nozzle. This achieves the technical effect of improving the monitoring and drug delivery efficiency of fruit tree leaf diseases and optimizing the intelligence of drug delivery, thereby solving the technical problems of low monitoring and drug delivery efficiency and poor intelligence of drug delivery in the prior art. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a schematic flowchart of an optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating another optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention. Figure 3 This is a schematic flowchart of another optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention; Figure 4 This is a schematic flowchart of another optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention; Figure 5 This is a schematic flowchart of another optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention; Figure 6 This is a schematic diagram of an optional intelligent drug delivery system for fruit tree leaves according to an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0019] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0020] Example 1

[0021] According to an embodiment of the present invention, an embodiment of an intelligent drug delivery method for fruit tree leaves is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0022] Figure 1 This is a schematic flowchart of an optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes the following steps: Step S102: Receive photos of fruit tree leaves captured by a high-definition camera installed in the tree-side monitoring and drug delivery device; Step S104: Input the fruit tree leaf photos into the preset EfficientNet-B1 convolutional neural network model to obtain the recognition results of the fruit tree leaf photos matching. The preset EfficientNet-B1 convolutional neural network model is used to identify the leaf disease symptoms corresponding to the fruit tree leaf photos. The preset EfficientNet-B1 convolutional neural network model is set in the artificial intelligence recognition chip or artificial intelligence recognition software installed in the tree-side monitoring and drug delivery device. Step S106: Send the identification result to the host computer, wherein the host computer is connected to the tree-side monitoring and drug delivery device network; Step S108: Receive the drug administration instruction corresponding to the identification result sent by the host computer; Step S110: Control the drug delivery module to spray the drug according to the drug delivery instruction. The drug delivery module is set in the tree-side monitoring drug delivery device. The drug delivery module includes at least a drug storage box, an electrically adjustable nozzle, a piezoelectric micro pressure pump and a control circuit. The drug storage box contains at least one drug bottle. The control circuit is used to control the direction of the electrically adjustable nozzle and control the piezoelectric micro pressure pump to push the drug in the drug bottle to the electrically adjustable nozzle for spraying.

[0023] In this embodiment of the invention, a tree-side monitoring and drug delivery device is used to collect photos of fruit tree leaves using its own high-definition camera. These photos are then input into a preset EfficientNet-B1 convolutional neural network model in the artificial intelligence recognition chip or software installed in the device. The identification results of leaf disease symptoms are obtained and sent to a network-connected host computer. Subsequently, the device receives the corresponding drug delivery command from the host computer. Under the control of the control circuit, the drug delivery module in the tree-side monitoring and drug delivery device pushes the corresponding agent through a piezoelectric micro-pressure pump and sprays it through an adjustable nozzle. This achieves the technical effect of improving the monitoring and drug delivery efficiency of fruit tree leaf diseases and optimizing the intelligence of drug delivery, thereby solving the technical problems of low monitoring and drug delivery efficiency and poor intelligence of drug delivery in the prior art.

[0024] In this embodiment of the invention, by using a high-definition camera in the tree-side monitoring and drug delivery device and a locally deployed EfficientNet-B1 model, leaf diseases can be automatically and quickly identified without manual inspection, improving identification efficiency and accuracy and solving the problem of early disease detection in large-scale orchards. The local deployment of the model avoids cloud transmission delays and network dependence, ensuring real-time disease identification. The upper-level computer matches the drug delivery instructions, and the drug delivery module in the tree-side monitoring and drug delivery device executes them precisely, achieving seamless linkage between identification and drug delivery. This allows for precise application of pesticides to diseases, reducing pesticide waste and environmental pollution. By controlling the selection of pesticides, the direction of the nozzle, and the operation of the pressure pump through the control circuit, the drug delivery plan can be dynamically adapted according to the disease situation, improving the control effect.

[0025] Optionally, Figure 2 This is a schematic flowchart of another optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention, such as... Figure 2 As shown, in step S102, receiving photos of fruit tree leaves captured by the high-definition camera installed in the tree-side monitoring and drug delivery device includes: Step S202: Receive photos of fruit tree leaves obtained by a high-definition camera within a preset acquisition period using a multi-angle shooting mode. The preset acquisition period dynamically corresponds to the growth stage of the fruit tree, and the multi-angle shooting mode includes a horizontal angle shooting mode, an upward angle shooting mode, and a downward angle shooting mode. Step S204: Preprocess the fruit tree leaf photos. The preprocessing includes at least one of the following: cropping the leaf area using an edge detection algorithm, removing image noise using Gaussian filtering, and balancing image contrast under different lighting conditions by adaptive brightness adjustment.

[0026] Optionally, the preset collection cycle can be dynamically adjusted according to the growth stage of the fruit trees. The collection cycle during the seedling stage can be 12 hours / time, and the collection cycle during the mature stage can be 24 hours / time. The multi-angle shooting mode can include horizontal angle shooting, 30° upward angle shooting and 30° downward angle shooting, and each angle can take 3-5 photos.

[0027] Optionally, Figure 3 This is a flowchart illustrating another optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention, as shown in Figure 3. The method further includes: Step S302: Construct the ImageNet training dataset. The ImageNet training dataset contains multiple fruit tree leaf images, and each fruit tree leaf image is labeled with leaf disease type parameters and leaf disease severity parameters. Step S304: Perform data augmentation on the ImageNet training dataset. The data augmentation process includes random rotation, random scaling, random horizontal flipping, random cropping, and adding Gaussian noise to obtain the target ImageNet training dataset. Step S306: Fine-tune and verify the performance of the pre-trained EfficientNet-B1 convolutional neural network model based on the target ImageNet training dataset according to the transfer learning method. Step S308: If the preset EfficientNet-B1 convolutional neural network model passes the model performance verification, deploy the preset EfficientNet-B1 convolutional neural network model into the artificial intelligence recognition chip or artificial intelligence recognition software.

[0028] For example, the ImageNet training dataset can contain at least 50,000 images of fruit tree leaves labeled with disease types and severity, covering 15 common leaf diseases and 3 severity levels (mild infection, moderate infection, and severe infection), and each disease image includes samples from different seasons, time periods, and shooting environments; data augmentation processing is performed on the training dataset, including random rotation (-30° to 30°), random scaling (0.8x to 1.2x), random horizontal flipping, random cropping, and adding Gaussian noise.

[0029] Optionally, transfer learning is employed to fine-tune a pre-trained EfficientNet-B1 convolutional neural network model based on the ImageNet dataset. For example, the parameters of the first 80% of the convolutional layers are frozen, and only the last 20% of the convolutional and fully connected layers are trained. The initial learning rate is set to 0.001, and a cosine annealing learning rate scheduling strategy is used for 100 training iterations. The model performance is verified using a test dataset. When the model's recognition accuracy reaches above 95% and its F1-score reaches above 0.94, the pre-trained EfficientNet-B1 convolutional neural network model is considered complete and can be deployed to an AI recognition chip or AI recognition software.

[0030] Optionally, Figure 4 This is a flowchart illustrating another optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention, as shown in Figure 4. The method further includes: Step S402: Input the disease-pesticide matching database into the host computer. The disease-pesticide matching database stores dosing instruction parameters corresponding to leaf disease symptoms. The dosing instruction parameters include at least one of the following: recommended agent type, optimal concentration ratio, single dosing dose, dosing frequency, and safety interval. In step S404, the host computer retrieves the dosing instruction corresponding to the identification result from the disease agent matching database based on the leaf disease type parameter and leaf disease severity parameter in the identification result.

[0031] Optionally, the host computer has a built-in disease-pesticide matching database, which stores recommended pesticide types, optimal concentration ratios, single dosages, dosing frequencies, and safety intervals for each leaf disease. Based on the disease type and severity identified in the results, the host computer retrieves the corresponding parameters from the disease-pesticide matching database and dynamically adjusts the dosing parameters in conjunction with the current growth stage of the fruit tree and real-time environmental data (including air temperature, relative humidity, wind speed, and sunlight intensity). When the identification result indicates a complex infection of multiple diseases, the host computer generates a complex dosing instruction, which includes the ratio of multiple pesticides, the staged spraying sequence, and pesticide compatibility verification results. The dosing instruction can be sent to the tree-side monitoring and dosing device using encrypted transmission based on the RSA asymmetric encryption algorithm.

[0032] Optionally, Figure 5 This is a flowchart illustrating another optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention. As shown in Figure 5, step S110, controlling the drug delivery module to spray the drug according to the drug delivery command, includes: In step S502, the control circuit controls the electric valve of the medicine bottle corresponding to the medicine type parameter in the medicine storage box to open according to the medicine type parameter in the medicine administration instruction, and then monitors the output amount of medicine in real time through the flow sensor built into the medicine bottle. In step S504, when the drug administration command is a multi-drug mixing command, the control circuit controls multiple target drug bottles to synchronously output drugs to the mixing chamber according to a preset ratio. The piezoelectric micro pressure pump pushes the drugs in the mixing chamber to the electrically controlled adjustable nozzle for spraying. The mixing chamber is set in the drug administration module and is connected to multiple target drug bottles and the electrically controlled adjustable nozzle respectively.

[0033] Optionally, the control circuit can control the opening of the electric valve of the corresponding medicine bottle in the medicine storage box according to the medicine type in the dosing instruction, and monitor the medicine output in real time through a flow sensor; when the dosing instruction requires mixing multiple medicines, the control circuit controls different medicine bottles to synchronously output medicines to the mixing chamber according to a preset ratio. The mixing chamber has an internal stirring rod to achieve uniform mixing of the medicines. After the mixed medicines are pressured by a pressure sensor, they are delivered to the electrically adjustable nozzle; the piezoelectric micro pressure pump adjusts its operating frequency according to the pressure parameters in the dosing instruction to achieve continuous pressure adjustment within the range of 0.1MPa to 0.5MPa, so as to control the switching of the medicine atomized particle diameter between 50μm and 200μm; the electrically adjustable nozzle is driven by a stepper motor to achieve 360° horizontal rotation and vertical rotation. The spray angle can be adjusted from 45° to 90°. The control circuit combines the leaf position information captured by the high-definition camera to calculate the optimal spray angle, so that the pesticide coverage rate reaches more than 90%. After spraying, the control circuit controls the drug delivery module to execute the cleaning program, which flushes the pipes and nozzles through the clean water box, and records the pesticide consumption, spraying time and equipment status parameters of this drug delivery, and feeds them back to the host computer for storage.

[0034] In this embodiment of the invention, a tree-side monitoring and drug delivery device is used to collect photos of fruit tree leaves using its own high-definition camera. These photos are then input into a preset EfficientNet-B1 convolutional neural network model in the artificial intelligence recognition chip or software installed in the device. The identification results of leaf disease symptoms are obtained and sent to a network-connected host computer. Subsequently, the device receives the corresponding drug delivery command from the host computer. Under the control of the control circuit, the drug delivery module in the tree-side monitoring and drug delivery device pushes the corresponding agent through a piezoelectric micro-pressure pump and sprays it through an adjustable nozzle. This achieves the technical effect of improving the monitoring and drug delivery efficiency of fruit tree leaf diseases and optimizing the intelligence of drug delivery, thereby solving the technical problems of low monitoring and drug delivery efficiency and poor intelligence of drug delivery in the prior art.

[0035] Example 2

[0036] According to another aspect of the present invention, an intelligent drug delivery system for fruit tree leaves is also provided. Figure 6 This is a schematic diagram of an optional intelligent drug delivery method for fruit tree leaves according to an embodiment of the present invention, as shown below. Figure 6 As shown, the system includes: The first receiving unit 601 is used to receive photos of fruit tree leaves collected by a high-definition camera set in the tree-side monitoring and drug delivery device; The recognition unit 603 is used to input fruit tree leaf photos into a preset EfficientNet-B1 convolutional neural network model to obtain the recognition result of matching fruit tree leaf photos. The preset EfficientNet-B1 convolutional neural network model is used to identify leaf disease symptoms corresponding to fruit tree leaf photos. The preset EfficientNet-B1 convolutional neural network model is set in the artificial intelligence recognition chip or artificial intelligence recognition software installed in the tree-side monitoring and drug delivery device. The sending unit 605 is used to send the identification result to the host computer, wherein the host computer is connected to the tree-side monitoring drug delivery device network; The second receiving unit 607 is used to receive the drug administration instruction sent by the host computer corresponding to the recognition result; The first processing unit 609 is used to control the drug delivery module to spray drugs according to the drug delivery instruction. The drug delivery module is set in the tree-side monitoring drug delivery device. The drug delivery module includes at least a drug storage box, an electrically adjustable nozzle, a piezoelectric micro pressure pump and a control circuit. The drug storage box contains at least one drug bottle. The control circuit is used to control the direction of the electrically adjustable nozzle and control the piezoelectric micro pressure pump to push the drug in the drug bottle to the electrically adjustable nozzle for spraying.

[0037] Optionally, the first receiving unit 601 includes: a receiving subunit, used to receive fruit tree leaf photos acquired by a high-definition camera in a multi-angle shooting mode within a preset acquisition period, wherein the preset acquisition period dynamically corresponds to the fruit tree growth stage, and the multi-angle shooting mode includes a horizontal angle shooting mode, an upward angle shooting mode, and a downward angle shooting mode; and a processing subunit, used to preprocess the fruit tree leaf photos, wherein the preprocessing includes at least one of the following: cropping the leaf area using an edge detection algorithm, removing image noise using Gaussian filtering, and balancing image contrast under different lighting conditions by adaptive brightness adjustment.

[0038] Optionally, the system further includes: a construction unit for constructing an ImageNet training dataset, which contains multiple fruit tree leaf images, each labeled with leaf disease type parameters and leaf disease severity parameters; a second processing unit for performing data augmentation on the ImageNet training dataset, including random rotation, random scaling, random horizontal flipping, random cropping, and adding Gaussian noise to obtain a target ImageNet training dataset; a validation unit for fine-tuning and validating the performance of a pre-trained EfficientNet-B1 convolutional neural network model based on the target ImageNet training dataset using transfer learning; and a deployment unit for deploying the pre-trained EfficientNet-B1 convolutional neural network model to an AI recognition chip or AI recognition software, provided that the model performance has been validated.

[0039] Optionally, the system further includes: an input unit for inputting a disease-pesticide matching database into a host computer, wherein the disease-pesticide matching database stores dosing instruction parameters corresponding to leaf disease symptoms, and the dosing instruction parameters include at least one of the following: recommended agent type, optimal concentration ratio, single dose, dosing frequency, and safety interval; and a retrieval unit for the host computer to retrieve the dosing instruction corresponding to the identification result from the disease-pesticide matching database based on the leaf disease type parameter and leaf disease severity parameter in the identification result.

[0040] Optionally, the first processing unit 609 includes: a first control subunit, used to control the electric valve of the medicine bottle corresponding to the medicine type parameter in the medicine storage box to open according to the medicine type parameter in the medicine administration instruction, and then monitor the output amount of medicine in real time through the flow sensor built into the medicine bottle; a second control subunit, used to control the control circuit to output medicine to the mixing chamber synchronously according to a preset ratio when the medicine administration instruction is a multi-medication mixing instruction, and the piezoelectric micro pressure pump pushes the medicine in the mixing chamber to the electrically controlled adjustable nozzle for spraying, wherein the mixing chamber is set in the medicine administration module and is respectively connected to the multiple target medicine bottles and the electrically controlled adjustable nozzle.

[0041] In this embodiment of the invention, a tree-side monitoring and drug delivery device is used to collect photos of fruit tree leaves using its own high-definition camera. These photos are then input into a preset EfficientNet-B1 convolutional neural network model in the artificial intelligence recognition chip or software installed in the device. The identification results of leaf disease symptoms are obtained and sent to a network-connected host computer. Subsequently, the device receives the corresponding drug delivery command from the host computer. Under the control of the control circuit, the drug delivery module in the tree-side monitoring and drug delivery device pushes the corresponding agent through a piezoelectric micro-pressure pump and sprays it through an adjustable nozzle. This achieves the technical effect of improving the monitoring and drug delivery efficiency of fruit tree leaf diseases and optimizing the intelligence of drug delivery, thereby solving the technical problems of low monitoring and drug delivery efficiency and poor intelligence of drug delivery in the prior art.

[0042] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0043] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0044] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The system embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.

[0045] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0046] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0047] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0048] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An intelligent method for administering medication to fruit tree leaves, characterized in that, include: Receive photos of fruit tree leaves captured by a high-definition camera installed in the tree-side monitoring and drug delivery device; The fruit tree leaf photos are input into a preset EfficientNet-B1 convolutional neural network model to obtain the recognition results of the fruit tree leaf photos. The preset EfficientNet-B1 convolutional neural network model is used to identify the leaf disease symptoms corresponding to the fruit tree leaf photos. The preset EfficientNet-B1 convolutional neural network model is set in the artificial intelligence recognition chip or artificial intelligence recognition software installed in the tree-side monitoring and drug delivery device. The identification result is sent to a host computer, wherein the host computer is network-connected to the tree-side monitoring and drug delivery device; Receive the drug administration command sent by the host computer corresponding to the identification result; The drug delivery module is controlled to spray drugs according to the drug delivery command. The drug delivery module is set in the tree-side monitoring drug delivery device. The drug delivery module includes at least a drug storage box, an electrically adjustable nozzle, a piezoelectric micro pressure pump, and a control circuit. The drug storage box contains at least one drug bottle. The control circuit is used to control the direction of the electrically adjustable nozzle and to control the piezoelectric micro pressure pump to push the drug in the drug bottle out of the electrically adjustable nozzle.

2. The intelligent drug delivery method for fruit tree leaves according to claim 1, characterized in that, The high-definition camera installed in the receiving tree-side monitoring and drug delivery device captures images of fruit tree leaves, including: The system receives photos of fruit tree leaves acquired by the high-definition camera within a preset acquisition period using a multi-angle shooting mode. The preset acquisition period dynamically corresponds to the growth stage of the fruit tree, and the multi-angle shooting mode includes a horizontal angle shooting mode, an upward angle shooting mode, and a downward angle shooting mode. The fruit tree leaf photographs are preprocessed, and the preprocessing includes at least one of the following: cropping the leaf area using an edge detection algorithm, removing image noise using Gaussian filtering, and balancing image contrast under different lighting conditions by adaptive brightness adjustment.

3. The intelligent drug delivery method for fruit tree leaves according to claim 1, characterized in that, The method further includes: An ImageNet training dataset is constructed, which contains multiple images of the fruit tree leaves, each image of the fruit tree leaves is labeled with leaf disease type parameters and leaf disease severity parameters; The ImageNet training dataset is subjected to data augmentation processing, which includes random rotation, random scaling, random horizontal flipping, random cropping, and adding Gaussian noise to obtain the target ImageNet training dataset. The preset EfficientNet-B1 convolutional neural network model, pre-trained on the target ImageNet training dataset, is fine-tuned and its performance is verified using transfer learning methods. If the preset EfficientNet-B1 convolutional neural network model passes the model performance verification, the preset EfficientNet-B1 convolutional neural network model is deployed to the artificial intelligence recognition chip or the artificial intelligence recognition software.

4. The intelligent drug delivery method for fruit tree leaves according to claim 3, characterized in that, The method further includes: Input the disease-pesticide matching database into the host computer. The disease-pesticide matching database stores dosing instruction parameters corresponding to the leaf disease symptoms. The dosing instruction parameters include at least one of the following: recommended agent type, optimal concentration ratio, single dosing dose, dosing frequency, and safety interval. The host computer retrieves the drug administration instruction corresponding to the identification result from the disease agent matching database based on the leaf disease type parameter and the leaf disease severity parameter in the identification result.

5. The intelligent drug delivery method for fruit tree leaves according to claim 1, characterized in that, The step of controlling the drug delivery module to spray the drug according to the drug delivery command includes: The control circuit controls the electric valve of the medicine bottle corresponding to the medicine type parameter in the medicine administration instruction to open according to the medicine type parameter in the medicine storage box, and then monitors the output amount of medicine in real time through the flow sensor built into the medicine bottle; When the drug administration command is a multi-drug mixing command, the control circuit controls multiple target drug bottles to synchronously output drugs to the mixing chamber according to a preset ratio. The piezoelectric micro pressure pump pushes the drugs in the mixing chamber to the electrically controlled adjustable nozzle for spraying. The mixing chamber is located in the drug administration module and is connected to the multiple target drug bottles and the electrically controlled adjustable nozzle respectively.

6. An intelligent drug delivery system for fruit tree leaves, characterized in that, include: The first receiving unit is used to receive photos of fruit tree leaves collected by a high-definition camera installed in the tree-side monitoring and drug delivery device; The recognition unit is used to input the fruit tree leaf photos into a preset EfficientNet-B1 convolutional neural network model to obtain the recognition result of the fruit tree leaf photos matching. The preset EfficientNet-B1 convolutional neural network model is used to identify the leaf disease symptoms corresponding to the fruit tree leaf photos. The preset EfficientNet-B1 convolutional neural network model is set in the artificial intelligence recognition chip or artificial intelligence recognition software installed in the tree-side monitoring and drug delivery device. A sending unit is used to send the identification result to a host computer, wherein the host computer is network-connected to the tree-side monitoring and drug delivery device; The second receiving unit is used to receive the drug administration instruction sent by the host computer corresponding to the identification result; The first processing unit is used to control the drug delivery module to spray drugs according to the drug delivery instruction. The drug delivery module is set in the tree-side monitoring drug delivery device. The drug delivery module includes at least a drug storage box, an electrically adjustable nozzle, a piezoelectric micro pressure pump, and a control circuit. The drug storage box contains at least one drug bottle. The control circuit is used to control the direction of the electrically adjustable nozzle and to control the piezoelectric micro pressure pump to push the drug in the drug bottle out of the electrically adjustable nozzle.

7. The intelligent drug delivery system for fruit tree leaves according to claim 6, characterized in that, The first receiving unit includes: The receiving subunit is used to receive photos of the fruit tree leaves acquired by the high-definition camera in a multi-angle shooting mode within a preset acquisition period. The preset acquisition period dynamically corresponds to the growth stage of the fruit tree, and the multi-angle shooting mode includes a horizontal angle shooting mode, an upward angle shooting mode, and a downward angle shooting mode. The processing subunit is used to preprocess the fruit tree leaf photograph, and the preprocessing includes at least one of the following: cropping the leaf area by edge detection algorithm, removing image noise by Gaussian filtering, and balancing image contrast under different lighting conditions by adaptive brightness adjustment.

8. The intelligent drug delivery system for fruit tree leaves according to claim 6, characterized in that, The system also includes: The construction unit is used to construct the ImageNet training dataset, which contains multiple images of the fruit tree leaves, and each image of the fruit tree leaves is labeled with leaf disease type parameters and leaf disease severity parameters; The second processing unit is used to perform data augmentation processing on the ImageNet training dataset. The data augmentation processing includes random rotation, random scaling, random horizontal flipping, random cropping, and adding Gaussian noise to obtain the target ImageNet training dataset. The verification unit is used to fine-tune and verify the performance of the preset EfficientNet-B1 convolutional neural network model pre-trained on the target ImageNet training dataset according to the transfer learning method. The deployment unit is used to deploy the preset EfficientNet-B1 convolutional neural network model to the artificial intelligence recognition chip or the artificial intelligence recognition software when it is determined that the preset EfficientNet-B1 convolutional neural network model has passed the model performance verification.

9. The intelligent drug delivery system for fruit tree leaves according to claim 8, characterized in that, The system also includes: An input unit is used to input a disease-pesticide matching database into the host computer. The disease-pesticide matching database stores dosing instruction parameters corresponding to the leaf disease symptoms. The dosing instruction parameters include at least one of the following: recommended agent type, optimal concentration ratio, single dosing dose, dosing frequency, and safety interval. The retrieval unit is used by the host computer to retrieve the drug administration instruction corresponding to the identification result from the disease agent matching database based on the leaf disease type parameter and the leaf disease severity parameter in the identification result.

10. The intelligent drug delivery system for fruit tree leaves according to claim 6, characterized in that, The first processing unit includes: The first control subunit is used by the control circuit to control the electric valve of the medicine bottle corresponding to the medicine type parameter in the medicine storage box to open according to the medicine type parameter in the medicine administration instruction, and then monitor the medicine output in real time through the flow sensor built into the medicine bottle; The second control subunit is used to control multiple target drug bottles to synchronously output drugs to the mixing chamber in a preset ratio when the drug administration command is a multi-drug mixing command. The piezoelectric micro pressure pump pushes the drugs in the mixing chamber to the electrically controlled adjustable nozzle for spraying. The mixing chamber is located in the drug administration module and is connected to the multiple target drug bottles and the electrically controlled adjustable nozzle respectively.

Images