A method for evaluating the results of microplastic pollution control

Abstract

The present invention discloses a method for evaluating the results of microplastic pollution control, which belongs to the technical field of pollution assessment, and specifically comprises: using unmanned aerial vehicles to automatically collect ground images of rural areas, identifying cultivated land areas in combination with cultivated land planning maps, and detecting plastic pollutants in cultivated areas and non-cultivated areas respectively, so as to count the total plastic pollution amount N and the total amount of plastic pollution in cultivated areas m; through multiple data collections, a line graph of N and m changing with time is drawn, and then the cultivated land pollution status and the trend of pollution diffusion are analyzed; at the same time, the system can identify the types of crops in the cultivated land, and calculate the pollution amount Pi per unit area of ​​each type, and construct the pollution index Q of the cultivated land in combination with the pollution weight ω and the slope k; when the Q value exceeds a preset threshold, the system will automatically issue a warning; the present invention helps to prevent the pollution problem from worsening and realize active intervention and effective control of microplastic pollution.

Description

Technical Field

[0001] The present invention relates to the technical field of pollution assessment, and in particular to a method for assessing the results of microplastic pollution control. Background Art

[0002] With the rapid development of modern agriculture and industry, plastic pollution has become one of the environmental problems that need to be solved urgently worldwide. In particular, microplastic pollution, due to its tiny particles and difficulty in degradation, has a long-term and far-reaching impact on soil, water sources and biodiversity. In rural areas, microplastic pollution not only threatens the growth quality of crops, but also may pose potential risks to human health through the food chain; therefore, how to effectively evaluate and control microplastic pollution in rural areas has become an important issue in the field of environmental protection.

[0003] Traditional pollution monitoring methods often rely on manual sampling and laboratory analysis, which is not only time-consuming and laborious, but also difficult to achieve large-scale, efficient monitoring. In recent years, with the rapid development of drone technology and computer vision technology, new solutions have been provided for the rapid identification and assessment of microplastic pollution. Drones can carry high-resolution cameras to conduct efficient aerial photography over a wide area and obtain a large amount of ground image data; while computer vision technology can automatically analyze these images through deep learning algorithms, accurately identify microplastic pollutants, and greatly improve monitoring efficiency and accuracy.

[0004] As a basic resource for agricultural production, the health of cultivated land is directly related to food security and farmers' income. However, the presence of microplastic pollution has seriously damaged the ecological environment of cultivated land and affected the normal growth of crops. Since microplastics are difficult to deal with, based on the principle of "blocking is worse than unblocking", it is necessary to detect plastic pollution in farmland in real time, so as to evaluate the results of pollution control and fundamentally solve the problem of microplastic pollution. Summary of the invention

[0005] The purpose of the present invention is to provide a method for evaluating the results of microplastic pollution control to solve the following technical problems:

[0006] As a basic resource for agricultural production, the health of cultivated land is directly related to food security and farmers' income. Since microplastics are difficult to deal with, based on the principle of "blocking is worse than unblocking", it is necessary to detect plastic pollution in farmland in real time, so as to evaluate the results of pollution control and fundamentally solve the problem of microplastic pollution.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A method for evaluating the results of microplastic pollution control comprises the following steps:

[0009] The ground images of the target rural area are collected based on the UAV platform, and the cultivated land area is identified based on computer vision technology and cultivated land planning maps. The plastic pollutants in the cultivated land area and the non-cultivated land area are detected respectively, and the total amount of plastic pollution N in the target area and the total amount of plastic pollution m in the cultivated land area are counted.

[0010] According to the total pollution amount N and the total pollution amount m collected for several times, the line graphs of the total pollution amount N and the total pollution amount m changing with time are drawn respectively. When the line graph of the total pollution amount N increases monotonically, if the line graph of the total pollution amount m increases monotonically, it means that the pollution of cultivated land is getting worse. If the line graph of the total pollution amount m decreases monotonically, it means that the pollution of cultivated land is decreasing, but the pollution diffusion is getting worse. When the total pollution amount N decreases monotonically, if the line graph of the total pollution amount m increases monotonically, it means that the pollution of cultivated land is getting worse, but the pollution diffusion is getting better. If the line graph of the total pollution amount m decreases monotonically, it means that the pollution of cultivated land is getting better.

[0011] Identify the crop types in the cultivated area and mark them as i, and calculate the pollution per unit area P for each crop i , the pollution index Q of cultivated land is calculated according to the following formula:

[0012] ;

[0013] Where n represents the number of crop types, ω represents the pollution weight corresponding to each crop, γ is the correction coefficient, and k j Indicates the slope of the total pollution m line graph between the current collection time and the previous collection time, k min Indicates the slope with the smallest value in the line graph segment of the total pollution m, k max It represents the slope with the largest value in the segment of the line graph of the total pollution amount m. When Q is greater than the preset threshold, a prompt is issued.

[0014] As a further solution of the present invention: when plastic pollution is identified in any cultivated land area, the polluter is controlled to fly above the cultivated land area and descend to a preset height.

[0015] As a further solution of the present invention: the computer vision technology includes a convolutional neural network model trained using a deep learning algorithm, which is used to automatically identify and distinguish between cultivated land and non-cultivated land areas. The specific training process is:

[0016] Collect ground image data including cultivated land and non-cultivated land areas as training sets and validation sets; annotate the images and design a convolutional neural network for image classification tasks; including several convolutional layers, pooling layers and fully connected layers;

[0017] The CNN model is trained using the training set. During the training process, the model's prediction results are calculated through forward propagation, and then compared with the true labels to calculate the loss function. Next, the gradient is calculated through the back-propagation algorithm, and the model parameters are updated to minimize the loss function. The above process is repeated until the model converges or the preset number of training rounds is reached.

[0018] Evaluate the performance of the trained CNN model on the validation set; determine the accuracy of the model by calculating the accuracy and recall rate; if the accuracy is lower than the preset requirement, retrain.

[0019] As a further solution of the present invention: the process of detecting microplastic pollutants is:

[0020] The collected ground images are preprocessed, including denoising and contrast enhancement; the preprocessed ground images are converted from RGB color space to HSV color space, and the grayscale threshold is set according to the color characteristics of plastic fragments; the ground images are converted into binary images, and the grayscale value of each pixel in the binary image is compared with the grayscale threshold to separate the plastic fragments from the background; shape analysis is performed on the binary image to extract the shape characteristics of the plastic fragments, including the area, perimeter, and edge contour of the plastic fragments; plastic fragments are further distinguished from other non-plastic objects; the size range of plastic fragments is set; by screening out fragments that meet the range, the extracted color and shape characteristics are matched with the preset plastic fragment feature library to identify them as plastic fragments; the identified plastic fragments are marked on the original ground image.

[0021] As a further solution of the present invention: the process of identifying the crop type in the cultivated area is:

[0022] Collect farmland image data containing different crop types; label the crop type in each image; extract crop features including color, texture, shape, and size from the labeled images; use principal component analysis to reduce the dimension of the feature space; store the extracted features and their corresponding crop types in a database to form a crop feature database; use the labeled image data and crop feature database to train a machine learning classifier;

[0023] When the drone collects real-time farmland images, the real-time farmland images are preprocessed and feature extracted; the features are input into the trained machine learning classifier for identification; and the corresponding crop type is output.

[0024] As a further solution of the present invention: the calculation process of the total amount of pollution is:

[0025] When the area of ​​any plastic fragment is larger than 1 square centimeter, the area value of the plastic fragment will be used as the count of the total pollution, in square centimeters; when the area of ​​any plastic fragment is less than 1 square centimeter, the count value of the plastic fragment will be 1.

[0026] As a further solution of the present invention: when the farmland pollution index Q exceeds a preset threshold, a report is generated to record in detail the location, area, pollution degree and recommended treatment measures of the farmland that exceeds the standard.

[0027] As a further solution of the present invention: the farmland planning map is used to demarcate registered farmland areas, and the computer vision technology is used to further identify unregistered farmland areas.

[0028] Beneficial effects of the present invention:

[0029] The present invention collects ground images through automated drone flight, which can quickly cover large rural areas. The convolutional neural network model trained by the deep learning algorithm, combined with computer vision technology, can automatically identify and distinguish between cultivated land and non-cultivated land areas, and accurately detect microplastic pollutants. It supports multiple data collection and draws a line graph of the total pollution over time, realizing real-time monitoring and dynamic analysis of microplastic pollution. By identifying the crop types in the cultivated land area and calculating the pollution per unit area, the method can formulate differentiated governance strategies for different crops. The present invention helps to prevent the pollution problem from worsening and realizes active intervention and effective control of microplastic pollution. It is expected to be promoted and applied in a wider range of rural areas, and help the development of environmental protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The present invention will be further described below in conjunction with the accompanying drawings.

[0031] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] See also Figure 1 As shown, the present invention is a method for evaluating the results of microplastic pollution control, comprising the following steps:

[0034] 1. Collect ground images of the target rural area based on the drone-mounted platform: Use the drone's high-resolution camera to conduct efficient and large-scale aerial photography in a wide area to obtain high-definition ground images covering the entire target area. These images can provide detailed geographic information and environmental details, laying the foundation for subsequent computer vision analysis.

[0035] 2. Identify cultivated land areas based on computer vision technology and cultivated land planning maps: The convolutional neural network model trained by the deep learning algorithm automatically identifies and distinguishes cultivated land and non-cultivated land areas. The model effectively captures complex features in the image through a deep architecture of multi-layer convolution, pooling, and fully connected layers, ensuring high accuracy and efficiency of the recognition process. At the same time, combined with the cultivated land planning map, the registered cultivated land areas are further accurately located, and the unregistered cultivated land areas are identified using computer vision technology.

[0036] 3. Detect plastic pollutants in cultivated land areas and non-cultivated land areas separately: The collected ground images are preprocessed, including denoising and contrast enhancement, to improve image quality. Subsequently, the image is converted from RGB color space to HSV color space, and the grayscale threshold is set according to the color characteristics of the plastic fragments to convert the image into a binary image. The shape characteristics of the plastic fragments (such as area, perimeter, edge contour) are extracted through shape analysis, and the plastic fragments are further distinguished from other non-plastic objects. The size range of plastic fragments is set, and the plastic pollutants are identified by screening out the fragments that meet the range and matching their color and shape characteristics with the preset plastic fragment feature library.

[0037] 4. Count the total amount of plastic pollution N in the target area and the total amount of plastic pollution m in the cultivated land area: Count the detected plastic pollutants and count the total amount of plastic pollution in the entire target area and the cultivated land area respectively.

[0038] 5. Draw a trend chart of the total amount of pollution over time through the data collected multiple times. When the total amount of pollution N line graph increases monotonically, if the total amount of pollution m line graph also increases monotonically, it means that the pollution of cultivated land has worsened; if the total amount of pollution m line graph decreases monotonically, it means that the pollution of cultivated land has decreased, but the pollution diffusion has worsened. When the total amount of pollution N decreases monotonically, if the total amount of pollution m line graph increases monotonically, it means that the pollution of cultivated land has worsened, but the pollution diffusion has decreased; if the total amount of pollution m line graph decreases monotonically, it means that the pollution of cultivated land has decreased.

[0039] 6. Preprocess and extract features of the cultivated land images collected in real time through machine learning classifiers, identify the types of crops in the cultivated land area, and assign them labels i.

[0040] 7. Based on the identified crop types, calculate the amount of plastic pollution per unit area of ​​each crop P i, the pollution index Q of cultivated land is calculated according to the formula:

[0041] ;

[0042] Where n represents the number of crop types, ω represents the pollution weight corresponding to each type of crop, which is determined according to the sensitivity of each type of crop to plastic pollution, and γ is the correction coefficient used to prevent the numerator of the formula from being 0. j Indicates the slope of the total pollution m line graph between the current collection time and the previous collection time, k min Indicates the slope with the smallest value in the line graph segment of the total pollution m, k max Indicates the slope with the largest value in the segment of the line graph of the total pollution m.

[0043] 8. Set a threshold value for the pollution index Q. When the calculated Q value exceeds the threshold, the system automatically issues a prompt message to remind relevant departments to take timely intervention measures.

[0044] This method uses drones to automatically fly to collect ground images of rural areas, and uses computer vision technology and deep learning algorithms to accurately identify and analyze the microplastic pollution status in cultivated and non-cultivated areas, thus achieving real-time monitoring and evaluation of microplastic pollution control results. At the same time, combined with cultivated land planning maps and crop type information, it provides a scientific basis for the formulation of targeted control measures. This method helps prevent the problem from worsening, achieves active control and effective intervention in microplastic pollution problems, and is expected to be promoted and applied in a wider range of rural areas, helping the development of environmental protection.

[0045] In another preferred embodiment of the present invention, when plastic pollution is identified in any cultivated land area, the polluter is controlled to fly above the cultivated land area and descend to an altitude of 50 to 100 meters.

[0046] In another preferred embodiment of the present invention, the computer vision technology includes a convolutional neural network model trained using a deep learning algorithm to automatically identify and distinguish between cultivated land and non-cultivated land areas. The specific training process is:

[0047] 1. Data Collection:

[0048] Collect ground image data covering both cultivated and non-cultivated areas. This data should cover images under various weather and lighting conditions to ensure the generalization ability of the model.

[0049] The collected image data is divided into a training set and a validation set. The training set is used to train the model, and the validation set is used to evaluate the performance of the model.

[0050] 2. Image Annotation:

[0051] Each image in the training set is labeled to clearly mark the cultivated land and non-cultivated land areas in the image. The accuracy of the labeling directly affects the learning effect of the model.

[0052] 3. Design Convolutional Neural Network (CNN):

[0053] Design a convolutional neural network (CNN) suitable for image classification tasks. The network usually includes several convolutional layers, pooling layers, and fully connected layers.

[0054] The convolution layer is used to extract features from the image and capture local information through convolution operations.

[0055] The pooling layer is used to reduce the spatial dimension of the feature map, reducing the amount of computation while retaining important features.

[0056] The fully connected layer flattens the feature map and makes classification predictions.

[0057] 4. Model training:

[0058] The CNN model is trained using the training set. During the training process, the model's prediction results are calculated through forward propagation, and then compared with the true labels to calculate the loss function.

[0059] The gradient is calculated through the back-propagation algorithm, and the model parameters are updated to minimize the loss function. Commonly used optimization algorithms include stochastic gradient descent (SGD) and Adam.

[0060] Repeat the above process until the model converges or reaches the preset number of training rounds. During the training process, data enhancement techniques (such as rotation, cropping, color transformation, etc.) can be used to increase the diversity of data and improve the robustness of the model.

[0061] 5. Model Evaluation:

[0062] Evaluate the performance of the trained CNN model on the validation set. Determine the accuracy and generalization ability of the model by calculating indicators such as precision and recall.

[0063] If the accuracy of the model is lower than the preset requirements, retrain it, or adjust the network structure and hyperparameters, and train and evaluate it again.

[0064] 6. Model deployment:

[0065] The trained CNN model is deployed on the UAV platform to process the ground images collected by the UAV in real time, and automatically identify and distinguish between cultivated land and non-cultivated land areas.

[0066] The present invention uses a convolutional neural network model trained by a deep learning algorithm to achieve automatic recognition and classification of cultivated and non-cultivated land areas, significantly improving the efficiency and accuracy of microplastic pollution assessment. The model effectively captures complex features in images through a deep architecture of multi-layer convolution, pooling, and fully connected layers. After sufficient training, it can show high accuracy and recall on the validation set. This technical means not only optimizes the data processing process and reduces manual intervention, but also ensures the objectivity and scientificity of the evaluation results, providing strong technical support for the precise management of microplastic pollution.

[0067] In another preferred embodiment of the present invention, the process of detecting microplastic pollutants is:

[0068] First, the collected ground images are preprocessed, including denoising and contrast enhancement. This step aims to improve image quality, reduce noise interference, and make subsequent feature extraction more accurate. Next, the preprocessed ground images are converted from RGB color space to HSV color space. This is because in the HSV color space, the separation of hue, saturation, and value makes the color features more obvious, which is convenient for identifying the color features of plastic fragments. According to the color features of plastic fragments, a grayscale threshold is set. By comparing the grayscale value of each pixel in the binary image with the grayscale threshold, the plastic fragments are separated from the background. In the binary image obtained in this way, the plastic fragments appear as white areas, while the background is black areas. Shape analysis is performed on the binary image to extract the shape features of the plastic fragments, including the area, perimeter, edge contour, etc. of the plastic fragments. These shape features help to further distinguish plastic fragments from other non-plastic objects, such as stones, leaves, etc. The size range of plastic fragments is set, and the extracted color and shape features are matched with the preset plastic fragment feature library by screening out the fragments that meet the range. If the match is successful, it is identified as plastic fragments. Finally, the identified plastic fragments are marked on the original ground image for subsequent statistics and analysis.

[0069] As can be seen from the above, the present invention achieves efficient and accurate detection of microplastic pollutants through a series of sophisticated image processing steps. First, the ground image quality is optimized using denoising and contrast enhancement techniques; then, the plastic fragments are successfully separated from the complex background through color space conversion and grayscale threshold setting. Further shape analysis and feature extraction, combined with preset feature library matching, ensure the high accuracy of the recognition process.

[0070] In another preferred embodiment of the present invention, the process of identifying the crop type of the cultivated area is:

[0071] First, collect farmland image data containing different types of crops. These data should cover images under various weather conditions and lighting conditions to ensure the generalization ability of the model. Then annotate the crop type in each image and clearly mark the crop type in the image. Extract the features of the crops from the annotated images, including color, texture, shape, size, etc. These features are key information to distinguish different crops. Use the principal component analysis method to reduce the dimension of the feature space to reduce the feature dimension and retain the main information. Store the extracted features and their corresponding crop types in the database to form a crop feature database. This database will serve as the basis for subsequent machine learning classifier training. Use the annotated image data and crop feature database to train the machine learning classifier. During the training process, the prediction results of the model are calculated by forward propagation, and then compared with the true label to calculate the loss function. Then the gradient is calculated by the back propagation algorithm, and the model parameters are updated to minimize the loss function. Repeat the above process until the model converges or the preset number of training rounds is reached. When the drone collects real-time farmland images, the real-time farmland images are preprocessed and feature extracted. The preprocessing step may include operations such as denoising and contrast enhancement to improve image quality. The extracted features are then input into a trained machine learning classifier for identification. The classifier will output the corresponding crop type based on the input features. In this way, the types of crops in the cultivated land area can be identified in real time, providing important reference information for subsequent microplastic pollution detection and governance results evaluation.

[0072] In another preferred embodiment of the present invention, the calculation process of the total amount of pollution is:

[0073] First, an area threshold of 1 square centimeter is set. Then the area of ​​each plastic fragment detected is measured. If the area of ​​any plastic fragment is larger than 1 square centimeter, the area value of the plastic fragment is directly used as the counting unit of the total pollution, that is, it is counted in square centimeters. This processing method can accurately reflect the contribution of larger plastic fragments to environmental pollution. On the other hand, the impact of smaller plastic fragments on pollution is not ignored. If the area of ​​any plastic fragment is less than 1 square centimeter, the count value of the plastic fragment is set to 1. This approach helps to simplify the statistical process. Through the above calculation process, the total amount of microplastic pollution in the target area can be obtained. This total amount not only reflects the number and size distribution of plastic fragments, but also provides an important basis for subsequent pollution trend analysis and governance effect evaluation.

[0074] In another preferred embodiment of the present invention, when the farmland pollution index Q exceeds the preset threshold, the system will automatically generate a detailed report. The report will first clearly indicate the location of the farmland that exceeds the standard, including the specific geographical coordinates and regional scope, so that the relevant departments can quickly locate the problem area. Then, the report will record in detail the area of ​​farmland that exceeds the standard, that is, the specific area of ​​land contaminated by microplastics, which will help to assess the severity of the pollution and the scope of impact. In addition, the report will also conduct an in-depth analysis of the degree of pollution, and by comparing historical data with current detection data, reveal the changes in pollution trends, as well as possible causes and influencing factors.

[0075] In another preferred embodiment of the present invention, by using the cultivated land planning map to mark the registered cultivated land areas, it is possible to ensure that the cultivated land information in these areas is accurately recorded and managed. At the same time, with the help of computer vision technology, the method can further identify unregistered cultivated land areas, thereby expanding the monitoring scope and improving the comprehensiveness of the data. This method, which combines traditional cultivated land planning maps with advanced computer vision technology, not only improves the accuracy and efficiency of cultivated land area identification, but also provides strong support for a comprehensive grasp of the status of cultivated land resources. It is helpful to timely discover the situation of newly added cultivated land or changes in cultivated land.

[0076] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for evaluating the results of microplastic pollution control, characterized in that: The following steps are involved: The ground images of the target rural area are collected based on the UAV platform, and the cultivated land area is identified based on computer vision technology and cultivated land planning maps. The plastic pollutants in the cultivated land area and the non-cultivated land area are detected respectively, and the total amount of plastic pollution N in the target area and the total amount of plastic pollution m in the cultivated land area are counted. According to the total pollution amount N and the total pollution amount m collected for several times, the line graphs of the total pollution amount N and the total pollution amount m changing with time are drawn respectively. When the line graph of the total pollution amount N increases monotonically, if the line graph of the total pollution amount m increases monotonically, it means that the pollution of cultivated land is getting worse. If the line graph of the total pollution amount m decreases monotonically, it means that the pollution of cultivated land is decreasing, but the pollution diffusion is getting worse. When the total pollution amount N decreases monotonically, if the line graph of the total pollution amount m increases monotonically, it means that the pollution of cultivated land is getting worse, but the pollution diffusion is getting better. If the line graph of the total pollution amount m decreases monotonically, it means that the pollution of cultivated land is getting better. Identify the crop types in the cultivated area and mark them as i, and calculate the pollution per unit area P for each crop i , the pollution index Q of cultivated land is calculated according to the following formula: ； Where n represents the number of crop types, ω represents the pollution weight corresponding to each crop, γ is the correction coefficient, and k j Indicates the slope of the total pollution m line graph between the current collection time and the previous collection time, k min Indicates the slope with the smallest value in the line graph segment of the total pollution m, k max It represents the slope with the largest value in the segment of the line graph of the total pollution amount m. When Q is greater than the preset threshold, a prompt is issued.

2. A method for evaluating the results of microplastic pollution control according to claim 1, characterized in that: When plastic pollution is identified in any cultivated land area, the polluter is controlled to fly over the cultivated land area and descend to a preset height.

3. A method for evaluating the results of microplastic pollution control according to claim 1, characterized in that: The computer vision technology includes a convolutional neural network model trained using a deep learning algorithm to automatically identify and distinguish between cultivated land and non-cultivated land areas. The specific training process is as follows: Collect ground image data including cultivated land and non-cultivated land areas as training sets and validation sets; annotate the images and design a convolutional neural network for image classification tasks; including several convolutional layers, pooling layers and fully connected layers; The CNN model is trained using the training set. During the training process, the model's prediction results are calculated through forward propagation, and then compared with the true labels to calculate the loss function. Next, the gradient is calculated through the back-propagation algorithm, and the model parameters are updated to minimize the loss function. The above process is repeated until the model converges or the preset number of training rounds is reached. Evaluate the performance of the trained CNN model on the validation set; determine the accuracy of the model by calculating the accuracy and recall rate; if the accuracy is lower than the preset requirement, retrain.

4. A method for evaluating the results of microplastic pollution control according to claim 1, characterized in that: The process of detecting microplastic pollutants is: The collected ground images are preprocessed, including denoising and contrast enhancement; the preprocessed ground images are converted from RGB color space to HSV color space, and the grayscale threshold is set according to the color characteristics of the plastic fragments; the ground images are converted into binary images, and the grayscale value of each pixel in the binary image is compared with the grayscale threshold to separate the plastic fragments from the background; the binary image is subjected to shape analysis to extract the shape characteristics of the plastic fragments; Including the area, perimeter, and edge contour of plastic fragments; further distinguishing plastic fragments from other non-plastic objects; setting the size range of plastic fragments; filtering out fragments that meet the range, matching the extracted color and shape features with the preset plastic fragment feature library, and identifying them as plastic fragments; The identified plastic fragments are marked on the original ground image.

5. A method for evaluating the results of microplastic pollution control according to claim 1, characterized in that: The process of identifying the crop type in the cultivated area is: Collect farmland image data containing different crop types; label the crop type in each image; extract crop features including color, texture, shape, and size from the labeled images; use principal component analysis to reduce the dimension of the feature space; store the extracted features and their corresponding crop types in a database to form a crop feature database; Use labeled image data and crop feature databases to train machine learning classifiers; When the UAV collects real-time farmland images, the real-time farmland images are preprocessed and feature extracted; Input the features into the trained machine learning classifier for identification; output the corresponding crop type.

6. A method for evaluating the results of microplastic pollution control according to claim 1, characterized in that: The calculation process of the total amount of pollution is: When the area of ​​any plastic fragment is larger than 1 square centimeter, the area value of the plastic fragment will be used as the count of the total pollution, in square centimeters; when the area of ​​any plastic fragment is less than 1 square centimeter, the count value of the plastic fragment will be 1.

7. A method for evaluating the results of microplastic pollution control according to claim 1, characterized in that: When the farmland pollution index Q exceeds the preset threshold, a report is generated to record in detail the location, area, degree of pollution and recommended treatment measures of the farmland that exceeds the standard.

8. A method for evaluating the results of microplastic pollution control according to claim 1, characterized in that: The farmland planning map is used to demarcate registered farmland areas, and the computer vision technology is used to further identify unregistered farmland areas.

Images