Micro-plastic particle identification method based on deep learning and application

Abstract

The invention discloses a micro-plastic particle identification method based on deep learning and application. The method comprises the following steps: receiving an image to be analyzed; inputting a to-be-analyzed image into the micro-plastic particle recognition model to carry out feature extraction on the to-be-analyzed image, and generating a result map for marking the positions of the micro-plastic particles; and extracting the coordinates of the positions of the marked micro-plastic particles in the result map, drawing a bounding box on the image to be analyzed, obtaining the image marked with the micro-plastic particles, and obtaining the information of the micro-plastic particles based on an image processing technology. Through the technology, the micro plastic particles in the image can be automatically identified, the micro plastic particles existing in the image can be accurately identified and analyzed, detailed data of the micro plastic particles are provided, the detailed data comprise shape, size, number and other information, a user is helped to better understand distribution of micro plastic pollution, and the user experience is improved. Therefore, the distribution and development trend of micro-plastic pollution can be tracked more comprehensively, and an effective micro-plastic pollution prevention and control strategy can be formulated.

Description

Technical Field

[0001] The present invention relates to the technical field of microplastic identification and detection, and in particular to a microplastic particle identification method and application based on deep learning. Background Art

[0002] With the rapid development of global industrialization and urbanization, the production and use of plastic products have continued to increase, leading to an increasingly serious problem of plastic pollution. Microplastic particles, i.e. plastic particles with a diameter of less than 5mm, are particularly easy to be overlooked, but they have a wide and far-reaching negative impact on the environment and ecosystem. Microplastic particles may enter water bodies, soil and food chains, posing a potential threat to wildlife and human health.

[0003] The widely used microplastic identification technology at present is mainly to achieve quantitative research on microplastics with the help of professional instruments and equipment, but such professional equipment is expensive, complex to maintain and has a certain technical threshold for operation; some scientific research teams usually use visual analysis as the main research method, use microscopic imaging to obtain microplastic photos, and manually count microplastic individuals, measure particle size, count particle area and obtain microplastic concentration in samples, which further causes a lot of laboratory work and cumbersome processes. It brings a huge workload, resulting in huge manpower consumption, time cost and management cost; with the rapid development of machine vision and neural network technology, image recognition, target analysis and size calculation technologies have been widely used in the field of environmental monitoring. There are many characteristics in laboratory statistics of microplastic individuals, particle size, shape and counting. On the other hand, due to the huge differences in the size of microplastics and their huge volume, it is urgent to establish relatively unified and standardized microplastic sampling and analysis and operating procedures to form low-cost, easy-to-operate and efficient microplastic quantitative experimental technology and equipment. Summary of the invention

[0004] To this end, it is necessary to provide a microplastic particle identification method based on deep learning to solve the problems of high cost, cumbersome operation and low efficiency in identifying microplastic particles.

[0005] To achieve the above objectives, the present invention provides a method for identifying microplastic particles based on deep learning, which comprises the following steps:

[0006] receiving an image to be analyzed;

[0007] The image to be analyzed is input into the microplastic particle recognition model to extract its features and generate a result map marking the location of the microplastic particles;

[0008] The coordinates of the microplastic particle positions marked in the result image are extracted, a bounding box is drawn on the image to be analyzed, an image of the marked microplastic particles is obtained, and the information of the microplastic particles is obtained based on image processing technology.

[0009] Furthermore, the image to be analyzed is acquired through single image recognition, real-time recognition or batch image analysis.

[0010] Furthermore, before the image to be analyzed is input into the microplastic particle recognition model, the image to be analyzed is preprocessed.

[0011] Furthermore, the microplastic particle recognition model is pre-trained based on a deep neural network model.

[0012] Furthermore, the microplastic particle recognition model is pre-trained based on a deep neural network model, including the following steps:

[0013] Get a sample set of images;

[0014] Segment and label the images in the image sample set, and use the results as labels for training data;

[0015] Establish a deep neural network model and set the training parameters of the deep neural network model;

[0016] The deep neural network model is trained using images in the image sample set as training data. The output of the deep neural network model is calculated through forward propagation, and then the loss function is used to calculate the difference between the output and the label of the training data. The deep neural network model parameters are updated through the back propagation algorithm to obtain the microplastic particle recognition model.

[0017] Furthermore, the image of the marked microplastic particles and the information of the microplastic particles are displayed in a visual manner.

[0018] A microplastic particle identification system based on deep learning, comprising a data acquisition module, a data preprocessing module, an analysis module and a result display and output module; the analysis module comprises a microplastic particle identification model, wherein:

[0019] The data acquisition module receives the image to be analyzed;

[0020] The data preprocessing module preprocesses the image to be analyzed;

[0021] The analysis module inputs the image to be analyzed into the microplastic particle recognition model to extract its features and generate a result map marking the location of the microplastic particles;

[0022] The result display and output module displays the result graph in a visual form.

[0023] Furthermore, the data acquisition module includes a single image recognition part, a real-time recognition and analysis part, and an image batch analysis and recognition part; wherein

[0024] The single image recognition part receives a single image uploaded by a user;

[0025] The real-time recognition and analysis part receives the image recognized by the camera in real time;

[0026] The real-time recognition and analysis part receives images uploaded in batches by users.

[0027] A storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the above-mentioned method for identifying microplastic particles based on deep learning.

[0028] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the steps of the above-mentioned method for identifying microplastic particles based on deep learning are implemented.

[0029] Different from the existing technology, the above technical solution can realize the automatic identification of microplastic particles in images through the microplastic particle recognition model, avoiding the tedious and time-consuming manual identification; in the generated result map, the position of the microplastic particles is clearly marked, which is convenient for subsequent analysis and processing; at the same time, the model performs precise extraction to accurately identify and analyze the microplastic particles in the image, and provides detailed data of the microplastic particles, including quantitative research information such as shape, size, and quantity, to help users better understand the distribution of microplastic pollution, thereby helping to better and more comprehensively track the distribution and development trend of microplastic pollution, and formulate effective microplastic pollution prevention and control strategies, which can be applied to the analysis of microplastic particles in water bodies, soil, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A schematic diagram of a specific implementation process of a method for identifying microplastic particles based on deep learning is provided for the present invention;

[0031] Figure 2 This is a schematic diagram of the process of training a microplastic particle recognition model based on a deep neural network model in the present invention;

[0032] Figure 3 A schematic diagram of a process for preprocessing images in an image sample set of the present invention;

[0033] Figure 4 It is the relationship diagram between P_curve accuracy and confidence;

[0034] Figure 5 PR_curve is the relationship between recall rate (recall rate) and confidence;

[0035] Figure 6 R_curve is the relationship diagram of mean average precision;

[0036] Figure 7 It is the positioning loss box_loss, confidence loss obj_loss and classification loss cls_loss graph;

[0037] Figure 8 This is a visualization diagram of the microplastic particle identification system based on deep learning of the present invention;

[0038] Fig. 9 This is a visualization diagram of the microplastic particle identification system based on deep learning of the present invention;

[0039] Fig.10 This is a visualization diagram of the microplastic particle identification system based on deep learning of the present invention. DETAILED DESCRIPTION

[0040] In order to explain the technical content, structural features, achieved objectives and effects of the technical solution in detail, the following is a detailed description in conjunction with specific embodiments and accompanying drawings.

[0041] Reference to "embodiment" herein means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present application. The term "embodiment" appearing in various places in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or association with other embodiments. In principle, in the present application, as long as there is no technical contradiction or conflict, the various technical features mentioned in the embodiments can be combined in any way to form a corresponding implementable technical solution.

[0042] Unless otherwise defined, the technical terms used in this document have the same meanings as those generally understood by those skilled in the art to which this application belongs; the use of relevant terms in this document is only for describing specific embodiments and is not intended to limit this application.

[0043] In the description of this application, the term "and / or" is an expression used to describe the logical relationship between objects, indicating that three relationships may exist, for example, A and / or B, which means: A exists, B exists, and A and B exist at the same time. In addition, the character " / " in this article generally indicates that the objects before and after are in an "or" logical relationship.

[0044] In the present application, terms such as “first” and “second” are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship of quantity, priority or sequence between these entities or operations.

[0045] Without further limitations, in this application, the words "include", "comprises", "has" or other similar expressions used in the sentences are intended to cover non-exclusive inclusion. These expressions do not exclude the presence of additional elements in the process, method or product including the elements, so that the process, method or product including a series of elements may include not only those limited elements, but also other elements not explicitly listed, or also include elements inherent to such process, method or product.

[0046] Similar to the understanding in the Patent Examination Guidelines, in this application, expressions such as "greater than", "less than", "exceed" and the like are understood to exclude the number itself; expressions such as "above", "below", "within" and the like are understood to include the number itself. In addition, in the description of the embodiments of this application, "multiple" means more than two (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups", "multiple times", etc., unless otherwise clearly and specifically limited.

[0047] In the description of the embodiments of the present application, space-related expressions used, such as "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "vertical", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicate the orientation or position relationship based on the orientation or position relationship shown in the specific embodiments or drawings, and are only for the convenience of describing the specific embodiments of the present application or facilitating the reader's understanding, and do not indicate or imply that the referred device or component must have a specific position, a specific orientation, or be constructed or operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present application.

[0048] Unless otherwise expressly specified or limited, in the description of the embodiments of the present application, the terms such as "install", "connect", "connect", "fix", "set", etc. used should be understood in a broad sense. For example, the "connection" can be a fixed connection, a detachable connection, or an integrated setting; it can be a mechanical connection, an electrical connection, or a communication connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium; it can be the internal connection of two elements or the interaction relationship between two elements. For technicians in the technical field to which the present application belongs, the specific meanings of the above terms in the embodiments of the present application can be understood according to the specific circumstances.

[0049] See also Figure 1-Figure 7As shown, the present invention provides a microplastic particle identification method based on deep learning. Through the microplastic particle identification model, the automatic identification of microplastic particles in the image can be achieved, avoiding the tedious and time-consuming manual identification; in the generated result map, the position of the microplastic particles is clearly marked, which is convenient for subsequent analysis and processing; at the same time, the model performs precise extraction to accurately identify and analyze the microplastic particles in the image, and provides detailed data of the microplastic particles, including quantitative research information such as shape, size, and quantity, to help users better understand the distribution of microplastic pollution, thereby helping to better and more comprehensively track the distribution and development trend of microplastic pollution, and formulate effective microplastic pollution prevention and control strategies, which can be applied to analyze microplastic particles in water bodies, soil, etc.

[0050] In order to further illustrate the microplastic particle identification method based on deep learning of the present invention, the following implementation is provided, which includes the following steps:

[0051] receiving an image to be analyzed;

[0052] The image to be analyzed is input into the microplastic particle recognition model to extract its features and generate a result map marking the location of the microplastic particles;

[0053] The coordinates of the microplastic particle positions marked in the result image are extracted, a bounding box is drawn on the image to be analyzed, an image of the marked microplastic particles is obtained, and the information of the microplastic particles is obtained based on image processing technology.

[0054] See also Figure 1 As shown, the above-mentioned received image to be analyzed is an image containing microplastic particles, which can be obtained by single image recognition, real-time recognition or batch image recognition to meet the user needs of different application scenarios, and has stronger scene diversity and universality. The image to be analyzed can be obtained by uploading in a visual way, that is, the user uploads the image to be analyzed through a visual interface. In order to improve the accuracy of the analysis, the image to be analyzed can be preprocessed before inputting the microplastic particle recognition model. For example, the size and format of the image to be analyzed are adjusted according to the size of the input of the microplastic particle recognition model, and the format of the image to be analyzed after preprocessing is the same as the format and size of the image input to the microplastic particle recognition model. Among them, the size can usually be adjusted by scaling, cropping and other operations to adjust the size of the image in the image sample set. In order to ensure the important feature area of ​​the image to be analyzed, it is preferred to adjust the size of the image in the image sample set by scaling. The scaling operation can be achieved by methods such as bilinear interpolation and nearest neighbor interpolation.

[0055] Specifically, the image data set can be preprocessed using OpenCV technology, and then the processed data set is trained, and the particle shape of the image to be analyzed is identified according to the microplastic particle recognition model to generate a result map marking the position of the microplastic particles, and the bounding box coordinates are extracted from the result map generating the marked position of the microplastic particles, and the bounding box is drawn on the image to be analyzed (original image), the position of the microplastic particles is marked, and the image of the marked microplastic particles is obtained, and finally the image of the marked microplastic particles is saved to a specified location for further analysis of the information or display of the microplastic particles. The information of the microplastic particles includes but is not limited to size, shape, etc., such as calculating the diameter, area and perimeter of the microplastic particles based on the bounding box of the microplastic particles in the image of the marked microplastic particles based on the image processing technology, and obtaining the size of the microplastic particles; such as analyzing the shape of the microplastic particles, such as irregular shapes, quadrilaterals, strips, circles, long strips, etc.; counting the number of microplastic particles in the image to provide a basis for subsequent data analysis. Of course, when drawing the bounding box, non-maximum suppression (NMS) and other technologies can be used to optimize the bounding box to reduce overlap and redundancy.

[0056] The above-mentioned microplastic particle recognition model is pre-trained based on a deep neural network model. The above-mentioned deep learning network model can be based on a convolutional neural network model (CNN), a target analysis YOLOE series model or other deep learning models, or even a model improved based on the above-mentioned model, and then trained by a large number of images with microplastic particles, so that the trained microplastic particle recognition model can generate a result map marking the position of microplastic particles according to the image to be analyzed; to further illustrate the microplastic particle recognition model of the present invention, the deep learning network model is further described by taking the YOLOE series model as an example. The deep learning network model can be a YOLOv8 model, and the microplastic particle recognition model is obtained by training the YOLOv8 model, so that the trained microplastic particle recognition model can generate a result map marking the position of microplastic particles according to the image to be analyzed.

[0057] See also Figure 2 As shown, training a specific microplastic particle recognition model based on a deep neural network model may include the following steps:

[0058] Get a sample set of images;

[0059] Segment and label the images in the image sample set, and use the results as labels for training data;

[0060] Establish a deep neural network model and set the training parameters of the deep neural network model;

[0061] The deep neural network model is trained using images in the image sample set as training data. The output of the deep neural network model is calculated through forward propagation, and then the loss function is used to calculate the difference between the output and the label of the training data. The deep neural network model parameters are updated through the back propagation algorithm to obtain the microplastic particle recognition model.

[0062] The above-mentioned image sample set is used to train the deep neural network model, including several images, which are images of microplastic particles. Preferably, several images of microplastic particles of different shapes are included, so that the deep neural network model can learn more microplastic particle shape information during the training process, so that the generated microplastic particle recognition model can more accurately identify microplastic particles in scene applications. Its shape can include regular shapes (such as bars, triangles, circles, strips and quadrilaterals) and irregular shapes. Strips, etc., can even be microplastic particle images combined with the above multiple shapes, so that the deep neural network model can learn the characteristics of microplastic particles of various shapes during the training process, so as to increase the shape recognition ability of the generated microplastic particle recognition model in scene applications, so that it can quickly identify and match, so as to improve the generalization ability and adaptability of the microplastic particle recognition model.

[0063] In some embodiments, before the images in the image sample set are segmented and labeled and used as training data, the images in the image sample set may be preprocessed, such as by denoising, data enhancement, etc. For example, data enhancement may be performed on the images in the image sample set, such as by rotating, scaling, flipping, etc., to improve the generalization ability and adaptability of the visual tracking model. The size of the images in the image sample set may also be adjusted according to the size of the deep neural network model input. The size of the images in the image sample set may usually be adjusted by scaling, cropping, and other operations. In order to ensure the important feature areas of the images, it is preferred to adjust the size of the images in the image sample set by scaling. The scaling operation may be implemented by methods such as bilinear interpolation and nearest neighbor interpolation. See Figure 3 As shown in the figure, after obtaining the data set, a sample library is constructed to mark the morphological images of microplastic particles by categories, which include original images such as bars, triangles, circles, irregular shapes, long strips and quadrilaterals; secondly, image processing such as rotation and flipping is used to enhance the data set; finally, the images are screened to see if they meet the standards. If not, they are manually cropped; these processes are used to generate images with marked features of different shape categories.

[0064] The above-mentioned images in the image sample set are segmented and labeled, and the results are used as labels for training data. Specifically, the images in the image sample set can be segmented into microplastic particle parts and non-microplastic particle parts. The microplastic particles in the images can be labeled manually or using tools.

[0065] Specifically, after annotating and saving the image, the tool will generate an XML file containing all the annotation information. This file records the position, shape, size and label of the annotation. This file will contain image information (such as name, size, color mode) and specific information of the annotation. Taking the YOLOv8 model as an example, the generated dataset is generally stored in a folder in the VOC annotation method. Therefore, it is necessary to change the format of these test datasets from XML format to the YOLO annotation method supported by YOLOv8, that is, .txt format.

[0066] To begin, note the class names in the dataset you want to modify. Use the in_file() function to read the XML file stored in a specified folder. This function will deconstruct the various variables in the XML file, such as size, name, pose, truncated, difficult, bandbox, etc. Then, convert the .txt format that YOLOv8 can recognize through the convert() function and use the out_file() function to save the modified .txt format file to the YOLOv8 folder for training data and label division of training data.

[0067] After the YOLOv8 algorithm is trained, the distribution of P_curve, PR_curve, and R_curve results is obtained. P_curve is the relationship between accuracy and confidence, PR_curve is the relationship between recall (recall rate) and confidence, and R_curve is the relationship between mean average precision. By adding and subtracting the number of rounds of training data, when the number of training rounds is low, its accuracy is also relatively low. Only when the number of training times increases, the accuracy of P_curve, PR_curve, and R_curve is also increasing. When the confidence is greater, the category detection is more accurate. Only when the confidence is large can it be judged as a certain category.

[0068] When the confidence is set to a certain value, the accuracy of each category recognition. It can be seen that the greater the confidence, the more accurate the category detection. Only when the confidence is very large, it is judged to be a certain category, see Figure 4 shown.

[0069] When the confidence is set to a certain value, the probability of each category being fully checked. It can be seen that when the confidence is smaller, the category detection is more comprehensive, see Figure 5 shown.

[0070] The higher the precision, the lower the recall rate. However, the desired network detects all categories as much as possible under the premise of high accuracy. Therefore, the desired curve is close to the point (1, 1), that is, the area of ​​the mAP curve is expected to be as close to 1 as possible, as shown in Figure 6.

[0071] As shown in 7, the loss function is used to measure the degree of difference between the model's predicted value and the true value, which largely determines the performance of the model. After reproducing the YOLOv8 model on the data set and the validation set, the distribution of the positioning loss box_loss, the confidence loss obj_loss and the classification loss cls_loss graphs can be obtained. The model has a large loss in the initial training (within 50 times). However, as the number of training times increases, the loss decreases rapidly, indicating that the model performance is gradually optimized. In particular, at the 90th training, the bounding box loss (box_loss), the target loss (obj_loss) and the classification loss (cls_loss) have all been reduced to the lowest, indicating that the model has achieved a relatively excellent quantitative recognition state in terms of target recognition and classification.

[0072] The images of the marked microplastic particles and the information of the microplastic particles are displayed in a visual manner. It is convenient and quick to conduct quantitative statistical research and processing of microplastics, and improve the convenience and economy of scientific research and experimental efficiency. It can be presented in a visual interface.

[0073] The present invention also provides a microplastic particle identification system based on deep learning, which includes a data acquisition module, a data preprocessing module, an analysis module and a result display and output module; the analysis module includes a microplastic particle identification model, wherein:

[0074] The data acquisition module receives the image to be analyzed;

[0075] The data preprocessing module preprocesses the image to be analyzed;

[0076] The analysis module inputs the image to be analyzed into the microplastic particle recognition model to extract its features and generate a result map marking the location of the microplastic particles;

[0077] The result display and output module displays the result graph in a visual form.

[0078] The data acquisition module includes a single image recognition part, a real-time recognition and analysis part, and an image batch analysis and recognition part;

[0079] The single image recognition part receives a single image uploaded by a user;

[0080] The real-time recognition and analysis part receives the image recognized by the camera in real time;

[0081] The image batch analysis and recognition part receives images uploaded in batches by users.

[0082] By providing multiple functions such as single image recognition and analysis, real-time camera recognition and analysis, and batch recognition and analysis, the user needs of different application scenarios are met, making the method and device of the present invention have stronger scene diversity and universality.

[0083] See also Figure 8 As shown in the figure, through the single image recognition part, users can select a single image to be analyzed. After the user selects the image, click the start button, and the system will start and begin to identify microplastic particles. After the recognition process is completed, the system will display the detected shape and mark the location of the microplastic particles on the image. The recognition results will be displayed intuitively to the user.

[0084] See also Fig. 9 As shown, through the real-time recognition and analysis part, after the user clicks, the camera of the user's device (camera / microscope, etc.) will be called to recognize the image, and the system will start to recognize microplastic particles in real time. The detected shape will be displayed and the display frame of the microplastic particles will also be synchronously marked on the interface.

[0085] See also Fig.10 As shown, through the image batch analysis and recognition part, by clicking the upload multiple pictures button, the user will be asked to select a folder of images to be identified. After the user selects the folder, the system will start to identify microplastic particles and display the detected shape and number, preprocessing time and inference time, post-processing time, and the location of the microplastic particle display box in real time.

[0086] The present invention also provides a storage medium, which stores a computer program, and when the computer program is executed by the processor, the steps of the above-mentioned microplastic particle identification method based on deep learning are implemented. The computer program involved in the embodiment can be stored in a computer device readable storage medium, and the computer device readable storage medium includes but is not limited to a disk, a tape, a magnetic card, a floppy disk, a flash memory, an optical disk, an optical card, a read-only memory (ROM), a random access memory (RAM), an erasable programmable ROM (EPROM) and an electrically erasable programmable ROM (EEPROM), etc., and also includes other biological, physical or chemical structures that can achieve the same or equivalent functions as the storage media listed above, such as DNA, RNA, protein and other units with information storage capabilities. In a specific embodiment, the storage medium involved can be one of the above-mentioned media types, or a combination of the above-mentioned media types. In different embodiments, the computer program involved in the embodiment can be stored in a single medium in a centralized manner, or it can be stored in multiple media in a distributed manner. The memory containing a computer device readable storage medium can be a non-volatile memory or a random access memory. These computer device readable storage media may be built into the device, or may be connected to the device involved in the embodiment as an external device or part of an external device. In some embodiments, a memory with a computer device readable storage medium is deployed locally; in other embodiments, a solution of deploying the memory away from the processor may also be adopted, such as a network attached memory accessed via an RF circuit or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), etc., or a suitable combination thereof, as long as the computer device can access the memory. In addition, the computer program involved in the embodiment may be stored in plaintext / ciphertext form, or may be designed as training data, which may be integrated and reorganized through model training and implicitly stored in the parameter state of a deep neural network or other machine learning model.

[0087] The present invention also provides an electronic device, which includes a memory and a processor, wherein a computer program is stored on the memory, and the computer program is executed by the processor to implement the steps of the above-mentioned method for identifying microplastic particles based on deep learning. The processor described in the embodiment of the present application can be implemented by hardware, firmware, software or a combination thereof, and can use a circuit, a single or multiple application-specific integrated circuits (Application Specific Integrated Circuit, ASIC), a digital signal processor (Digital Signal Processor, DSP), a digital signal processing device (Digital Signal Processing Device, DSPD), a programmable logic device (Programmable Logic Device, PLD), a field programmable gate array (Field Programmable Gate Array, FPGA), a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, at least one of a microprocessor, and also includes other physical, biological or chemical structures that can realize functions similar to or equivalent to the above-listed processors, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in each embodiment of the present application or any combination of the steps mentioned therein.

[0088] It should be noted that, although the above embodiments have been described in this article, the patent protection scope of the present invention is not limited thereby. Therefore, based on the innovative concept of the present invention, changes and modifications made to the embodiments described herein, or equivalent structures or equivalent process changes made using the contents of the present invention specification and drawings, directly or indirectly applying the above technical solutions to other related technical fields, are all included in the patent protection scope of the present invention.

Claims

1. A method for identifying microplastic particles based on deep learning, characterized in that: The following steps are involved: receiving an image to be analyzed; The image to be analyzed is input into the microplastic particle recognition model to extract its features and generate a result map marking the location of the microplastic particles; The coordinates of the microplastic particle positions marked in the result image are extracted, a bounding box is drawn on the image to be analyzed, an image of the marked microplastic particles is obtained, and the information of the microplastic particles is obtained based on image processing technology.

2. The method for identifying microplastic particles based on deep learning according to claim 1, characterized in that: The image to be analyzed is obtained by single image recognition, real-time recognition or batch image analysis.

3. The method for identifying microplastic particles based on deep learning according to claim 1, characterized in that: Before the image to be analyzed is input into the microplastic particle recognition model, the image to be analyzed is preprocessed.

4. The method for identifying microplastic particles based on deep learning according to claim 1, characterized in that: The microplastic particle recognition model is pre-trained based on a deep neural network model.

5. The method for identifying microplastic particles based on deep learning according to claim 4, characterized in that: The microplastic particle recognition model is pre-trained based on a deep neural network model, which includes the following steps: Get a sample set of images; Segment and label the images in the image sample set, and use the results as labels for training data; Establish a deep neural network model and set the training parameters of the deep neural network model; The deep neural network model is trained using images in the image sample set as training data. The output of the deep neural network model is calculated through forward propagation, and then the difference between the output and the label of the training data is calculated using the loss function. The deep neural network model parameters are updated through the back propagation algorithm to obtain a microplastic particle recognition model.

6. The method for identifying microplastic particles based on deep learning according to claim 1, characterized in that: The images of the marked microplastic particles and the information of the microplastic particles are displayed in a visual manner.

7. A microplastic particle identification system based on deep learning, characterized in that: It includes a data acquisition module, a data preprocessing module, an analysis module and a result display and output module; the analysis module includes a microplastic particle identification model, wherein: The data acquisition module receives the image to be analyzed; The data preprocessing module preprocesses the image to be analyzed; The analysis module inputs the image to be analyzed into the microplastic particle recognition model to extract its features and generate a result map marking the location of the microplastic particles; The result display and output module displays the result graph in a visual form.

8. A microplastic particle identification system based on deep learning according to claim 7, characterized in that: The data acquisition module includes a single image recognition part, a real-time recognition and analysis part, and an image batch analysis and recognition part; The single image recognition part receives a single image uploaded by a user; The real-time recognition and analysis part receives the image recognized by the camera in real time; The real-time recognition and analysis part receives images uploaded in batches by users.

9. A storage medium, characterized in that The storage medium stores a computer program, which, when executed by a processor, implements the steps of the microplastic particle identification method based on deep learning as described in any one of claims 1 to 6.

10. An electronic device, characterized in that It includes a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the steps of the microplastic particle identification method based on deep learning as described in any one of claims 1 to 6 are implemented.

Images