Tobacco plant growth and development stage and nutrition state identification method and device based on unmanned aerial vehicle technology

Abstract

The invention provides a tobacco plant growth and development stage and nutrition state identification method and device based on an unmanned aerial vehicle technology, and the method comprises the following steps: 1, obtaining tobacco plant visible light images, collected by an unmanned aerial vehicle, of different growth and development stages and different nutrition states, and constructing a tobacco plant data set; 2, establishing a tobacco plant growth and development stage classification model and nutrition state classification models of different growth and development stages, and training the development stage classification model and the nutrition state classification models of different development stages based on the tobacco plant data set in the step 1; step 3, obtaining a to-be-identified tobacco plant visible light image collected by the unmanned aerial vehicle, sending the to-be-identified tobacco plant visible light image into the trained tobacco plant growth and development stage classification model to obtain a tobacco plant growth and development stage, and sending the to-be-identified tobacco plant visible light image into the nutrition state classification model of the corresponding growth and development stage to obtain a nutrition state classification model of the tobacco plant growth and development stage; the nutrition state of the tobacco plants is obtained.

Description

Technical Field

[0001] The present invention relates to the field of intelligent identification of tobacco fields, and in particular to a method and device for identifying the growth and development stages and nutritional status of tobacco plants based on drone technology. Background Art

[0002] Tobacco leaves are the primary raw material for Chinese cigarette production. Scientifically and rationally grading tobacco leaves based on quality is crucial for raw tobacco leaf procurement and cigarette formulation design. Tobacco plants have significantly different requirements for nitrogen, phosphorus, potassium, and trace elements at different growth and development stages (e.g., seedling, root extension, growth, and maturity). For example, during the seedling stage, higher nitrogen levels are needed to promote leaf growth. During the growth stage, potassium is needed to enhance stress resistance and promote dry matter accumulation. During the maturity stage, nitrogen levels must be controlled to avoid excessive greening and late maturation, which can affect leaf quality. Therefore, observing the growth and development stage and nutritional status of tobacco plants during the growing season can better reflect and predict leaf quality. This observation facilitates timely adjustments to fertilizer and watering levels in tobacco fields. Strictly controlling the nutritional requirements at each stage of tobacco growth can effectively improve yield and quality.

[0003] At present, the growth and development stage and nutritional status of tobacco leaves mainly rely on manual experience, and manual grading is easily affected by subjective experience and environmental conditions, and there are problems such as low grading accuracy and poor consistency. To this end, CN119360207A proposes a field tobacco plant growth status classification method based on deep learning, which uses tobacco field camera equipment to collect tobacco leaf information image data in real time, establishes a tobacco field data set under a complex background, and uses a deep learning algorithm to discriminate the growth status of tobacco leaves. However, in the above scheme, the tobacco field camera needs to capture images in distant, medium and close shots, which takes a long time; and the background of the picture captured by the tobacco field camera is too messy and the clarity is not enough, so it is necessary to perform binary processing first to eliminate the complex background interference in the picture; in addition, the above scheme adopts a multi-output branch model to simultaneously identify the growth and development stage and nutritional status; however, this method has high requirements for the training data set. If the data distribution of the two tasks of growth and development stage and nutritional status is very different, or the amount of data of one task is much larger than the other, it may lead to poor performance of one task.

[0004] In recent years, drone remote sensing technology has been increasingly used for field tobacco plant counting and maturity assessment due to its low cost, high resolution, and timely nature. This technology is capable of rapidly acquiring large-scale image data. Therefore, it is necessary to develop a method for identifying tobacco plant growth and nutritional status based on drone technology that requires minimal training data and can quickly and accurately identify these stages and nutritional status.

[0005] In order to solve the above problems, people have been seeking an ideal technical solution. Summary of the Invention

[0006] Based on this, it is necessary to provide a method and device for identifying the growth and development stages and nutritional status of tobacco plants based on drone technology to address the above technical problems.

[0007] To achieve the above objectives, the present invention provides a method for identifying tobacco plant growth and development stages and nutritional status based on drone technology, comprising the following steps: Step 1: Obtain visible light images of tobacco plants at different growth and development stages and nutritional status collected by drones to construct a tobacco plant dataset; Step 2: Establishing a tobacco plant growth and development stage classification model and a nutritional status classification model at different growth and development stages, and training the growth stage classification model and the nutritional status classification model at different development stages based on the tobacco plant dataset in step 1; Step 3: Obtain the visible light image of the tobacco plant to be identified collected by the drone, send the visible light image of the tobacco plant to be identified into the trained tobacco plant growth and development stage classification model to obtain the tobacco plant growth and development stage, and send the visible light image of the tobacco plant to be identified into the nutritional status classification model corresponding to the growth and development stage to obtain the nutritional status of the tobacco plant.

[0008] In a possible embodiment of the first aspect, the growth and development stage classification model and the nutritional status classification model for different growth and development stages are both based on the EfficientNet-B0 model, the ordinary convolution of the MBConv module in the EfficientNet-B0 model is replaced by point-by-point grouped convolution, and a channel shuffling operation is introduced after the point-by-point grouped convolution, and the SE attention in the MBConv module is replaced by efficient channel attention.

[0009] In a possible embodiment of the first aspect, different growth and development stages include the clumping stage, the middle stage of vigorous growth, the dome stage, the lower leaf maturity stage, the middle leaf maturity stage and the upper leaf maturity stage, and different nutritional states include five nutritional states: extremely weak, weak, normal, strong and extremely strong.

[0010] In a possible embodiment of the first aspect, the step of constructing a tobacco plant dataset includes: The visible light images of tobacco plants at the same growth and development stage and different nutritional status collected by drones were stitched together to form images of the complete experimental field plots. The complete experimental field plot images were cropped according to 1024×1024 pixels, and the corresponding growth and development stages and nutritional status were annotated for the images to form the initial dataset. The initial data set is divided into training set and test set in a ratio of 8:2; Construct an image enhancement list that defines various image enhancement methods, including horizontal flipping, color conversion, adjusting image brightness, contrast, saturation, and hue, rotating an image by an angle between -30 and 30 degrees, partially blocking an image, grayscaling an image, and adding salt and pepper noise and Gaussian noise. Randomly select a preset number of image transformation methods from the image enhancement list to enhance the images in the training set to expand the training set; The expanded training set is divided into training set and validation set again according to 8:2, and the initially divided test set is used for testing after the model training is completed.

[0011] In order to achieve the above-mentioned objectives, the second aspect of the present invention provides a device for identifying tobacco plant growth and development stages and nutritional status based on drone technology, comprising: The tobacco plant dataset construction module is used to obtain visible light images of tobacco plants at different growth and development stages and in different nutritional states collected by drones to construct a tobacco plant dataset; A model building and training module is used to build a tobacco plant growth and development stage classification model and a nutritional status classification model at different growth and development stages, and train the development stage classification model and the nutritional status classification model at different development stages based on the tobacco plant dataset in step 1; The development stage recognition module has a built-in trained tobacco plant growth and development stage classification model. It is used to identify the visible light images of tobacco plants to be identified collected by drones using the tobacco plant growth and development stage classification model to obtain the tobacco plant growth and development stage; The nutritional status recognition module has a built-in trained nutritional status classification model for different developmental stages, and is used to use the nutritional status classification model of the aforementioned growth and development stage to identify the visible light image of the tobacco plant to be identified and obtain the nutritional status of the tobacco plant.

[0012] In an embodiment of the second aspect, the growth and development stage classification model and the nutritional status classification model for different growth and development stages are both based on the EfficientNet-B0 model, the ordinary convolution of the MBConv module in the EfficientNet-B0 model is replaced by point-by-point grouped convolution, and a channel shuffle operation is introduced after the point-by-point grouped convolution, and the SE attention in the MBConv module is replaced by efficient channel attention.

[0013] In order to achieve the above-mentioned purpose, the third aspect of the present invention provides a tobacco plant growth and development stage and nutritional status identification system based on drone technology, comprising a drone and the tobacco plant growth and development stage and nutritional status identification device described in the second aspect. The drone is used to collect visible light images of tobacco plants in the field and send them to the tobacco plant growth and development stage and nutritional status identification device; The tobacco plant growth and development stage and nutritional status recognition device recognizes the visible light image to obtain the tobacco plant growth and development stage and nutritional status.

[0014] In order to achieve the above-mentioned purpose, the fourth aspect of the present invention provides a tobacco plant growth and development stage and nutritional status identification device based on drone technology, which includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; the memory is used to store computer programs; the processor is used to implement the steps of the tobacco plant growth and development stage and nutritional status identification method based on drone technology when executing the program stored in the memory.

[0015] In order to achieve the above-mentioned purpose, the fifth aspect of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for identifying the growth and development stages and nutritional status of tobacco plants based on drone technology are implemented.

[0016] In order to achieve the above-mentioned purpose, the sixth aspect of the present invention provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for identifying the growth and development stages and nutritional status of tobacco plants based on drone technology.

[0017] The present invention has the following beneficial effects: by using drone technology to acquire visible light images, a tobacco plant dataset covering the complete tobacco plant development cycle in the field is established. The visible light images in this tobacco plant dataset are relatively neat and have high resolution. Furthermore, a method is proposed to first identify the tobacco plant's growth and development stage using a tobacco plant growth and development stage classification model, and then obtain the tobacco plant's nutritional status using a nutritional status classification model corresponding to the growth and development stage. This method can accurately identify the growth and development stage and nutritional status, and the models for each stage are independent and do not interfere with each other. Each classification model uses a lightweight neural network, which can quickly identify the developmental stage and nutritional status of field tobacco plants and is more suitable for scenarios with limited computing power on the drone's onboard side. Specifically, based on the EfficientNet-B0 model, the ordinary convolution in the MBConv module is replaced with point-by-point group convolution, and a channel shuffling operation is introduced after the point-by-point group convolution to enhance the cross-group interaction of local features. Furthermore, to enhance feature fusion and the comprehensive utilization of cross-channel information, the SE attention in the MBConv module in the EfficientNet-B0 model is replaced with efficient channel attention to better undertake the aforementioned group convolution feature flow process and make the feature information transmission between channels more flexible and comprehensive. This design realizes the organic combination of "local computing + global information", optimizes the cross-channel information that may be lost in traditional group convolution, thereby improving the expression ability of the convolution kernel in capturing global information, further improving the computing effect, and has the characteristics of lightweightness, which is more adaptable to edge distributed computing devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart for identifying tobacco plant growth and development stages and nutritional status in an embodiment of the present invention; Figure 2 These are images of tobacco plants at different growth and development stages according to an embodiment of the present invention; Figure 3 These are images of tobacco plants in different nutritional states according to an embodiment of the present invention; Figure 4 This is a data enhancement strategy diagram in an embodiment of the present invention.

[0019] Figure 5 This is a structural diagram of the MBConv module in ES-Efficientnet in an embodiment of the present invention.

[0020] Figure 6 This is a confusion matrix diagram for identifying growth and development stages in an embodiment of the present invention.

[0021] Figure 7 This is a confusion matrix diagram for identifying the nutritional status of the cluster stage in an embodiment of the present invention.

[0022] Figure 8 This is a confusion matrix diagram for identifying the nutritional status in the middle stage of vigorous growth in an embodiment of the present invention.

[0023] Figure 9 This is a confusion matrix diagram for identifying the nutritional status of the dome stage in an embodiment of the present invention.

[0024] Figure 10 This is a confusion matrix diagram for identifying the nutritional status of upper leaves at maturity in an embodiment of the present invention.

[0025] Figure 11This is a confusion matrix diagram for identifying the nutritional status of middle leaves at maturity in an embodiment of the present invention.

[0026] Figure 12 This is a confusion matrix diagram for identifying the nutritional status of lower leaves at maturity in an embodiment of the present invention.

[0027] Figure 13 This is a line chart comparing the accuracy, precision, recall, and F1 score for identifying growth and development stages in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The technical solution of the present invention is further described in detail below through specific implementation methods.

[0029] Example 1 This embodiment provides a method for identifying tobacco plant growth and development stages and nutritional status based on drone technology. Figure 1 As shown, the following steps are included: Step 1: Obtain visible light images of tobacco plants at different growth and development stages and nutritional status collected by drones to construct a tobacco plant dataset.

[0030] Specifically, using the experimental fields in the tobacco production area of ​​Fujian Province as an example, a drone equipped with a visible light lens was used to collect visible light images of tobacco leaves. As can be seen, since the drone is perpendicular to the ground, the visible light images it collects are relatively neat and have a high resolution.

[0031] Step 1.1: A fertilizer gradient experiment was conducted to create experimental plots with varying nutritional status. The test variety was CB-1, and the experimental sites were Chishan Village, Jianyang County, Nanping City, Fujian Province; Yanghou Village, Jianyang County, Nanping City, Fujian Province; Kaishan Village, Taining County, Sanming City, Fujian Province; and Xiaqu Village, Taining County, Sanming City, Fujian Province. A nitrogen fertilizer gradient experiment was conducted at each of the four experimental sites, consisting of five gradients (T0, T50, T100, T150, and T200), corresponding to 0%, 50%, 100%, 150%, and 200% of the local conventional nitrogen application rate, respectively. Each treatment consisted of four rows of tobacco plants, with 80-90 plants per row. Each treatment had three replicates, for a total of 15 plots. A guard row was established between each treatment, and an isolation zone was established between each plot.

[0032] In step 1.2, tobacco leaf images were collected using a drone equipped with a visible light camera. The drone model was a DJI Matrice 350 RTK, and the visible light camera was a Zenmuse P1. During the experiment, the drone was set to a fixed altitude of 20 meters to collect images.

[0033] In step 1.3, based on the expert's judgment of the tobacco plant's growth and development stage and nutritional status, tobacco leaf image data is collected when the tobacco field enters the clustering stage, the middle stage of vigorous growth, the dome stage, the lower leaf maturity stage, the middle leaf maturity stage, and the upper leaf maturity stage.

[0034] In step 1.4, the collected images of the four experimental fields were stitched together using DJI Zhitu software to create a complete image of the experimental field plot. Based on the results of the nitrogen fertilizer gradient experiment and the agronomic trait survey at each growth stage, the collected tobacco plant images were classified into five nutritional status categories: very weak, weak, normal, strong, and very strong.

[0035] Step 1.5: Crop the image according to 1024×1024 pixels, and divide the images of different growth and development stages and nutritional status as the initial data set. The images of different growth stages are as follows: Figure 2 As shown, images of different nutritional status are as follows Figure 3 shown.

[0036] In step 1.6, the initial dataset is divided into training set and test set in a ratio of 8:2.

[0037] Step 1.7, define an image enhancement list, including: horizontal flip, color conversion, adjusting the brightness, contrast, saturation and hue of the image, rotating the image between [-30, 30] degrees, partial occlusion of the image, grayscale, adding salt and pepper noise and Gaussian noise. The image contrast after each method is transformed is as follows: Figure 4 The images of the training set are randomly selected from the image enhancement list to enhance the image volume by using three image transformation methods. For example, in this embodiment, the images of the training set are expanded by 2 times, that is, the number of the training set is tripled.

[0038] During specific implementation, the number of types of image transformations selected and the final number of training set samples are determined according to the number of basic samples in the training set.

[0039] In step 1.8, the expanded training set was split again into training and validation sets at an 8:2 ratio. The initial test set was used for testing after model training. The initial image count for each experiment was 3627 images, 1210 images for the expanded training set, 1210 images for the expanded validation set, and 1210 images for the initial test set.

[0040] Step 2: Establish a tobacco plant growth and development stage classification model and a nutritional status classification model at different growth and development stages. The development stage classification model and the nutritional status classification model at different development stages are trained based on the tobacco plant dataset in step 1.

[0041] Specifically, the growth and development stage classification model and the nutritional status classification model for different growth and development stages are both ES-EfficientNet-B0 models improved based on the EfficientNet-B0 model.

[0042] Among them, the EfficientNet-B0 model is composed of different levels of mobile inverted bottleneck convolution (MBConv) architecture stacking, specifically, including 16 mobile flip bottleneck convolution modules, 2 convolution layers, 1 global average pooling layer and 1 classification layer.

[0043] In this embodiment, Figure 5 As shown in the figure, the normal convolution in MBConv is replaced with point-by-point grouped convolution. A new channel shuffle operation is added after the point-by-point grouped convolution in MBConv to improve computational efficiency. Then, the efficient channel attention mechanism (ECA) is used to replace the SE module in the MBConv module in the original model.

[0044] It's understandable that replacing conventional convolution with grouped convolution can reduce the number of parameters, enhance local feature learning, adapt to the computing power limitations of drones, and improve the ability to capture subtle features. Adding channel shuffling breaks information isolation and promotes cross-group feature interaction; it also enhances multi-feature collaboration in complex field scenarios and suppresses background interference.

[0045] However, it should be noted that grouped convolution features only flow between groups. Although adding channel shuffling can make them flow across groups, it is still impossible to learn all group features together. Therefore, although grouped convolution has high computational efficiency, it still has certain limitations in feature fusion and the comprehensive utilization of cross-channel information. This application further replaces the SE module in the MBConv module with the ECA attention mechanism, which can better undertake the aforementioned grouped convolution feature flow process and make the feature information transfer between channels more flexible and comprehensive.

[0046] This design enables each channel to perform calculations based on its own local information during the convolution process while receiving weighted information from other channels, realizing the organic combination of "local calculation + global information" and optimizing the cross-channel information that may be lost in traditional grouped convolution, thereby improving the expressive power of the convolution kernel in capturing global information and further improving the computational effect.

[0047] Specifically, ECA adaptively weights each channel and dynamically adjusts the relative importance of channels, enabling a more effective global integration of all features, enabling the network to better learn and express information. In contrast, while the SE module can effectively compress features, its further compression of feature information often limits the ability to express features, failing to fully utilize the richness of features.

[0048] To verify the effectiveness of the tobacco plant growth status classification model, we conducted ablation experiments to compare the model results with commonly used image classification models. Specifically, the following were performed: Step 1), ES-EfficientNet recognizes the growth and development stages with an accuracy of 97.9%, a recall rate of 97.8%, and an F1 score of 97.9%. The confusion matrix is ​​as follows: Figure 6 ES-EfficientNet has an accuracy of 82.5%, a recall rate of 83.5%, and an F1 score of 83.3% in identifying the nutritional status of clusters. The confusion matrix is ​​shown in Figure 7 ES-EfficientNet has an accuracy of 85.9%, a recall rate of 87.0%, and an F1 score of 86.8% in identifying the nutritional status of mid-stage vigorous growth. The confusion matrix is ​​shown in Figure 2. Figure 8 ES-EfficientNet has an accuracy of 72.0%, a recall rate of 74.8%, and an F1 score of 74.0% in recognizing the nutritional status of the dome stage. The confusion matrix is ​​shown in Figure 3 ES-EfficientNet has an accuracy of 85.9%, a recall rate of 86.4%, and an F1 score of 86.4% in identifying the nutritional status of upper leaves at maturity. The confusion matrix is ​​shown in Figure 10 ES-EfficientNet has an accuracy of 87.3%, a recall rate of 87.7%, and an F1 score of 87.4% in identifying the nutritional status of the middle leaf at maturity. The confusion matrix is ​​shown in Figure 2. Figure 11 ES-EfficientNet recognizes the nutritional status of lower leaves at maturity with an accuracy of 79.0%, a recall rate of 79.3%, and an F1 score of 79.0%. The confusion matrix is ​​shown in Figure 2. Figure 12 shown.

[0049] In step 2, new components were gradually removed from the model in ablation experiments to evaluate the contribution of each component to performance. The original EfficientNet model for identifying growth and development stages achieved 95.5% accuracy, 95.3% recall, and 3.82MB of parameters. Replacing SE attention within the MBConv module of the original model with efficient channel attention (ECA) achieved 96.4% accuracy, 96.3% recall, and 3.22MB of parameters. Replacing ordinary convolution within the MBConv module of the original model with pointwise grouped convolution and adding a new channel shuffling operation achieved 90.8% accuracy, 90.6% recall, and 2.17MB of parameters. The overall improved ES-EfficientNet achieved 97.9% accuracy, 97.8% recall, and 1.56MB of parameters.

[0050] Step 3) We compared the model with those that have achieved good results in image classification tasks in recent years, such as EfficientNet-B0, EfficientNet V2, Mobile ViT, MobileNet V2, MobileNet V3, ResNet34, and ShuffleNet.

[0051] The experimental results show that although MobileViT, MobileNet V3 and ShuffleNet have smaller parameters, the accuracy of ES-EfficientNet is 25.35%, 37.22% and 2.81% higher than the other three respectively. The comparison line chart of the experimental results for identifying growth and development stages is shown below. Figure 13 Experiments have shown that the improved ES-EfficientNet is a model with fewer parameters and the highest accuracy, demonstrating excellent performance and effectiveness. Multiple experimental comparisons have further verified the generalizability of ES-EfficientNet.

[0052] Step 3: Obtain the visible light images of the tobacco plants to be identified collected by the drone, and send the visible light images of the tobacco plants to be identified into the trained tobacco plant growth and development stage classification model to obtain the tobacco plant growth and development stage, and send the visible light images of the tobacco plants to be identified into the nutritional status classification model corresponding to the growth and development stage to obtain the nutritional status of the tobacco plants, so that the growth and development stage and nutritional status of the tobacco plants in the tobacco field can be understood in real time during the tobacco planting stage, and effective measures can be taken to increase the quality and yield of the tobacco plants by adjusting the amount of fertilizer, etc.

[0053] It can be understood that this embodiment uses a step-by-step approach to achieve the classification and identification of the growth and development stages and nutritional status of tobacco plants. When identifying the nutritional status, it also utilizes the characteristics of the growth and development stages of tobacco plants, selects the nutritional status classification model corresponding to the growth and development stage of tobacco plants, and obtains the nutritional status of tobacco plants, thereby improving the accuracy of nutritional status identification.

[0054] It should be understood that, although the steps in the flowcharts of the above embodiments are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the flowcharts of the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily to be performed in sequence, but can be performed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0055] Example 2 Based on the same inventive concept, embodiments of the present application also provide a tobacco plant growth and development stage and nutritional status identification device for implementing the aforementioned method for identifying tobacco plant growth and development stages and nutritional status using drone technology. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more tobacco plant growth and development stage and nutritional status identification device embodiments provided below can be found in the above-mentioned limitations of the tobacco plant growth and development stage and nutritional status identification method, and will not be repeated here.

[0056] Specifically, the tobacco plant growth and development stage and nutritional status identification device based on drone technology includes: The tobacco plant dataset construction module is used to obtain visible light images of tobacco plants at different growth and development stages and in different nutritional states collected by drones to construct a tobacco plant dataset; A model building and training module is used to build a tobacco plant growth and development stage classification model and a nutritional status classification model at different growth and development stages, and train the development stage classification model and the nutritional status classification model at different development stages based on the tobacco plant dataset in step 1; The development stage recognition module has a built-in trained tobacco plant growth and development stage classification model. It is used to identify the visible light images of tobacco plants to be identified collected by drones using the tobacco plant growth and development stage classification model to obtain the tobacco plant growth and development stage; The nutritional status recognition module has a built-in trained nutritional status classification model for different developmental stages, and is used to use the nutritional status classification model of the aforementioned growth and development stage to identify the visible light image of the tobacco plant to be identified and obtain the nutritional status of the tobacco plant.

[0057] Specifically, the growth and development stage classification model and the nutritional status classification model for different growth and development stages are both based on the EfficientNet-B0 model. The ordinary convolution of the MBConv module in the EfficientNet-B0 model is replaced by point-by-point grouped convolution, and a channel shuffling operation is introduced after the point-by-point grouped convolution. The SE attention in the MBConv module is replaced by efficient channel attention.

[0058] Example 3 This embodiment provides a tobacco plant growth and development stage and nutritional status identification system based on drone technology, including a drone and the tobacco plant growth and development stage and nutritional status identification device described in Example 2. The drone is used to collect visible light images of tobacco plants in the field and send them to the tobacco plant growth and development stage and nutritional status identification device; The tobacco plant growth and development stage and nutritional status recognition device recognizes the visible light image to obtain the tobacco plant growth and development stage and nutritional status.

[0059] Example 4 This embodiment provides a synchronous in-situ measurement device for the starch content of tobacco leaves in the middle of a field tobacco plant during its maturity period. The device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the display unit, and the input device are connected to the system bus via the input / output interface. The processor of the device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the device is used to exchange information between the processor and an external device. When the computer program is executed by the processor, it implements the method for identifying the growth and development stage and nutritional status of tobacco plants based on drone technology as described in Example 1.

[0060] Example 5 Based on the above embodiments, this embodiment provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for identifying the growth and development stages and nutritional status of tobacco plants based on drone technology described in Example 1 are implemented.

[0061] Example 6 Based on the above embodiments, this embodiment provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the method for identifying tobacco plant growth and development stages and nutritional status based on drone technology described in Example 1.

[0062] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to a memory, database, or other medium used in the embodiments provided herein may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, and the like.

[0063] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0064] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or some technical features can be replaced by equivalents without departing from the spirit of the technical solutions of the present invention. They should all be included in the scope of the technical solutions claimed for protection by the present invention.

Claims

1. A method for identifying tobacco plant growth and development stages and nutritional status based on drone technology, characterized in that: The following steps are involved: Step 1: Obtain visible light images of tobacco plants at different growth and development stages and nutritional status collected by drones to construct a tobacco plant dataset; Step 2: Establishing a tobacco plant growth and development stage classification model and a nutritional status classification model at different growth and development stages, and training the growth stage classification model and the nutritional status classification model at different development stages based on the tobacco plant dataset in step 1; Step 3: Obtain the visible light image of the tobacco plant to be identified collected by the drone, send the visible light image of the tobacco plant to be identified into the trained tobacco plant growth and development stage classification model to obtain the tobacco plant growth and development stage, and send the visible light image of the tobacco plant to be identified into the nutritional status classification model corresponding to the growth and development stage to obtain the nutritional status of the tobacco plant.

2. The method for identifying tobacco plant growth and development stages and nutritional status based on drone technology according to claim 1, characterized in that: The growth and development stage classification model and the nutritional status classification model for different growth and development stages are both based on the EfficientNet-B0 model. The ordinary convolution of the MBConv module in the EfficientNet-B0 model is replaced by point-by-point grouped convolution, and a channel shuffle operation is introduced after the point-by-point grouped convolution. The SE attention in the MBConv module is replaced by efficient channel attention.

3. The method for identifying tobacco plant growth and development stages and nutritional status based on drone technology according to claim 1 or 2, characterized in that: Different growth and development stages include the clumping stage, the middle stage of vigorous growth, the dome stage, the lower leaf maturity stage, the middle leaf maturity stage and the upper leaf maturity stage; different nutritional states include five nutritional states: extremely weak, weak, normal, strong and extremely strong.

4. The method for identifying tobacco plant growth and development stages and nutritional status based on drone technology according to claim 3, characterized in that: The steps to construct the tobacco plant dataset include: The visible light images of tobacco plants at the same growth and development stage and different nutritional status collected by drones were stitched together to form images of the complete experimental field plots. The complete experimental field plot images were cropped according to 1024×1024 pixels, and the corresponding growth and development stages and nutritional status were annotated for the images to form the initial dataset. The initial data set is divided into training set and test set in a ratio of 8:2; Construct an image enhancement list that defines various image enhancement methods, including horizontal flipping, color conversion, adjusting image brightness, contrast, saturation, and hue, rotating an image by an angle between -30 and 30 degrees, partially blocking an image, grayscaling an image, and adding salt and pepper noise and Gaussian noise. Randomly select a preset number of image transformation methods from the image enhancement list to enhance the images in the training set to expand the training set; The expanded training set is divided into training set and validation set again according to 8:2, and the initially divided test set is used for testing after the model training is completed.

5. A tobacco plant growth and development stage and nutritional status identification device based on drone technology, characterized in that: include: The tobacco plant dataset construction module is used to obtain visible light images of tobacco plants at different growth and development stages and in different nutritional states collected by drones to construct a tobacco plant dataset; A model building and training module is used to build a tobacco plant growth and development stage classification model and a nutritional status classification model at different growth and development stages, and train the development stage classification model and the nutritional status classification model at different development stages based on the tobacco plant dataset in step 1; The development stage recognition module has a built-in trained tobacco plant growth and development stage classification model. It is used to identify the visible light images of tobacco plants to be identified collected by drones using the tobacco plant growth and development stage classification model to obtain the tobacco plant growth and development stage; The nutritional status recognition module has a built-in trained nutritional status classification model for different developmental stages, and is used to use the nutritional status classification model of the aforementioned growth and development stage to identify the visible light image of the tobacco plant to be identified and obtain the nutritional status of the tobacco plant.

6. The device for identifying tobacco plant growth and development stages and nutritional status based on drone technology according to claim 5, characterized in that: The growth and development stage classification model and the nutritional status classification model for different growth and development stages are both based on the EfficientNet-B0 model. The ordinary convolution of the MBConv module in the EfficientNet-B0 model is replaced by point-by-point grouped convolution, and a channel shuffle operation is introduced after the point-by-point grouped convolution. The SE attention in the MBConv module is replaced by efficient channel attention.

7. A tobacco plant growth and development stage and nutritional status identification system based on drone technology, comprising a drone and the tobacco plant growth and development stage and nutritional status identification device according to claim 5 or 6, The drone is used to collect visible light images of tobacco plants in the field and send them to the tobacco plant growth and development stage and nutritional status identification device; The tobacco plant growth and development stage and nutritional status recognition device recognizes the visible light image to obtain the tobacco plant growth and development stage and nutritional status.

8. A device for identifying tobacco plant growth and development stages and nutritional status based on drone technology, characterized by: The processor, the communication interface, the memory and the communication bus are connected to each other via the communication bus. Memory for storing computer programs; The processor is configured to implement the steps of the method for identifying the growth and development stage and nutritional status of tobacco plants based on drone technology as described in any one of claims 1 to 4 when executing the program stored in the memory.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for identifying the growth and development stage and nutritional status of tobacco plants based on drone technology according to any one of claims 1 to 4 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for identifying the growth and development stage and nutritional status of tobacco plants based on drone technology according to any one of claims 1 to 4 are implemented.

Images