A method, system, device and medium for identifying microbial occlusion

Abstract

This invention discloses a method, system, device, and medium for identifying microbial occlusion, specifically relating to the field of occlusion recognition technology. The key points are: inputting the original microbial image into an image segmentation model for foreground microbial segmentation to obtain a segmentation mask for the foreground microorganisms; using a deep learning classification network to identify the microbial category of each foreground microorganism to obtain an initial set of foreground microbial categories; inputting all segmentation masks and the original microbial image into a microbial missing generation model for image restoration; performing foreground microbial segmentation and microbial category recognition on the restored microbial image; if a foreground microorganism not present in the initial foreground microbial set is identified, then inputting the segmentation mask of that foreground microorganism and the restored microbial image into a pre-constructed microbial missing generation model for image restoration until no foreground microorganisms are identified, thus obtaining the final microbial occlusion recognition result.

Description

Technical Field

[0001] This invention relates to the field of occlusion recognition technology, specifically to a method, system, device, and medium for identifying microorganisms through occlusion. Background Technology

[0002] In the context of automated wastewater treatment and environmental monitoring, microscopic imaging combined with AI recognition is being used to identify sludge morphology and microbial communities online. However, in actual imaging, the same or different microorganisms often overlap, and factors such as flocs, bubbles, impurities, and uneven lighting lead to weak contrast and partial occlusion. Traditional recognition methods that rely solely on visible areas are prone to missed detections, false detections, and counting errors. For occluded background microorganisms, the lack of key morphological and texture information makes it difficult for general recognition or restoration models to provide reconstructions and discriminations that meet biological constraints.

[0003] For example, some existing technologies employ general object detection frameworks based on deep learning (such as YOLO, R-CNN series, Mask R-CNN, etc.) or instance segmentation networks (such as DeepLab, U-Net, etc.) to extract image features through convolutional neural networks for target localization, classification, and pixel-level segmentation. These frameworks demonstrate good recognition and counting capabilities for clear, unobstructed microbial samples. However, these general detection and segmentation models are primarily pre-trained on natural image datasets such as COCO and ImageNet. When encountering occlusion or weak contrast, they rely solely on the visible area of ​​the target for judgment, which can easily lead to missed or false detections. They lack specific optimization for small microbial targets and morphological diversity under microscopes, resulting in insufficient generalization ability. Alternatively, other existing technologies employ image inpainting models based on deep learning (such as Context Encoders, Partial...). Or traditional inpainting algorithms (such as Convolution, EdgeConnect, etc.) or sample block-based texture synthesis, Poisson fusion, etc., treat the occluded area as a "missing pixel" and automatically fill it in, reconstructing the occluded part by learning the texture patterns of surrounding pixels. These general image inpainting models are mainly trained on natural images and have not been adapted to the specific morphology, boundary features, and texture patterns of microorganisms. The generated results often lack biological rationality and are prone to artifacts such as morphological distortion and blurred boundaries, making it difficult to stably serve subsequent microbial identification and accurate counting. Alternatively, pre-trained deep neural networks (such as Amodal instance segmentation network, occlusion inference network, integrity completion network, etc.) are used to directly infer the complete shape, boundary contour, and spatial position of the occluded target from a single frame of microscope image, and achieve shape completion of the occluded area by learning a large number of occlusion-complete target pair mapping relationships. The training of this type of model requires a large amount of high-quality occlusion annotation data, which is costly. The generalization ability to new microbial species or microscope imaging conditions is limited. The inference results mainly output geometric contour masks, lacking texture detail information, and are difficult to support fine classification and morphological analysis that rely on texture features.

[0004] Therefore, the present invention aims to provide a method, system, device and medium for identifying microbial occlusion in order to solve the aforementioned problems. Summary of the Invention

[0005] The technical problem this invention aims to solve is that in general detection and segmentation models, information loss due to occlusion can lead to false detections and missed detections, resulting in low microbial counts or misclassification. The goal is to provide a method, system, device, and medium for identifying microbial occlusion. This involves using an image segmentation model for precise foreground segmentation to accurately locate occluded areas; then, utilizing a generative repair model based on a stable diffusion model and LoRA fine-tuning technology, specifically trained for microbial data, to generate high-quality repair content that conforms to the morphological and textural characteristics of microorganisms; finally, through a closed-loop processing mechanism of "initial identification—generative repair—secondary identification—iterative control," automatically determining whether new foreground microbial targets exist and deciding whether to continue iteration, significantly improving the identification recall and counting accuracy, and providing reliable technical support for automatic microscopic water quality detection.

[0006] This invention is achieved through the following technical solution:

[0007] A method for identifying microorganisms by occlusion, the method comprising:

[0008] The original microbial image is input into the image segmentation model to segment the foreground microorganisms, and a segmentation mask for all foreground microorganisms is obtained. Then, a pre-trained deep learning classification network is used to identify the microbial category of each foreground microorganism, and an initial set of foreground microbial categories is obtained.

[0009] All segmentation masks and the original microbial image are input into a pre-constructed microbial missing generation model for image inpainting to obtain the repaired microbial image;

[0010] The restored microbial image is segmented into foreground microorganisms and the microbial category is identified. If a foreground microorganism that does not appear in the initial foreground microorganism set is identified, the segmentation mask of the foreground microorganism and the restored microbial image are input into a pre-constructed microbial missing generation model for image restoration until no foreground microorganism that did not appear is identified, and the final microbial occlusion recognition result is obtained.

[0011] Furthermore, the original microbial image is input into the image segmentation model for foreground microbial segmentation, resulting in a segmentation mask for all foreground microorganisms, specifically:

[0012] Raw microbial images were acquired using a microscope imaging device, and the raw microbial images were preprocessed to obtain preprocessed raw microbial images.

[0013] The image segmentation model was used to segment the foreground microorganisms in the preprocessed original microbial image, and the segmentation mask of all foreground microorganisms was obtained.

[0014] Furthermore, the microbial deletion generation model was obtained by training the SD-XL Inpainting model, which was fine-tuned by LoRA.

[0015] Furthermore, the foreground microbial category set includes the coordinate information, category information, confidence level, and mask information of each foreground microorganism.

[0016] The present invention also provides a microbial occlusion identification system, which is used in any of the above-described microbial occlusion identification methods, the system comprising:

[0017] The initial identification module is used to input the original microbial image into the image segmentation model for foreground microbial segmentation, obtain the segmentation mask of all foreground microorganisms, and use a pre-trained deep learning classification network to identify the microbial category of each foreground microorganism, thereby obtaining an initial set of foreground microbial categories.

[0018] The image inpainting module is used to input all segmentation masks and the original microbial images into a pre-constructed microbial missing generation model for image inpainting, and obtain the repaired microbial images;

[0019] The secondary recognition module is used to segment foreground microorganisms and identify microorganism categories in the repaired microbial image. If a foreground microorganism that does not appear in the initial foreground microorganism set is identified, the segmentation mask of the foreground microorganism and the repaired microbial image are input into the pre-constructed microbial missing generation model for image repair until no foreground microorganisms that do not appear are identified, and the final microbial occlusion recognition result is obtained.

[0020] Furthermore, the original microbial image is input into the image segmentation model for foreground microbial segmentation, resulting in a segmentation mask for all foreground microorganisms, specifically:

[0021] Raw microbial images were acquired using a microscope imaging device, and the raw microbial images were preprocessed to obtain preprocessed raw microbial images.

[0022] The image segmentation model was used to segment the foreground microorganisms in the preprocessed original microbial image, and the segmentation mask of all foreground microorganisms was obtained.

[0023] Furthermore, the microbial deletion generation model was obtained by training the SD-XL Inpainting model, which was fine-tuned by LoRA.

[0024] The present invention also provides a computer device, including a system memory and a processor, wherein the system memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described above.

[0025] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the methods described above.

[0026] The present invention also provides a computer program product containing instructions that, when executed by a cluster of computer devices, cause the cluster of computer devices to perform the method described in any of the preceding claims.

[0027] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0028] In this invention, an image segmentation model is used for precise foreground segmentation to accurately locate occluded areas. Then, a generative repair model based on a stable diffusion model and LoRA fine-tuning technology is specifically trained for microbial data to generate high-quality repair content that conforms to the morphological and textural characteristics of microorganisms. Finally, through a closed-loop processing mechanism of "initial identification - generation repair - secondary identification - iterative control", the system automatically determines whether there are new foreground microbial targets and decides whether to continue iterating, which significantly improves the identification recall rate and counting accuracy, providing reliable technical support for automatic microscopic water quality detection. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0030] Figure 1 This is a schematic diagram of the process flow of a microbial occlusion identification method in this embodiment;

[0031] Figure 2 This is an example of a microbial image used to train the SD-XL Inpainting model in this embodiment;

[0032] Figure 3 This is an example of a microbial occlusion image used to train the SD-XL Inpainting model in this embodiment;

[0033] Figure 4 This is a schematic diagram of the module connections of a microbial occlusion recognition system in this embodiment;

[0034] Figure 5 This is a schematic diagram of the structure of a computer device in this embodiment. Detailed Implementation

[0035] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0036] In this disclosure, unless otherwise stated, the use of terms such as "first," "second," etc., to describe various elements is not intended to limit the positional, temporal, or importance relationships of these elements; such terms are merely used to distinguish one element from another. In some examples, the first element and the second element may refer to the same instance of that element, while in other cases, based on the context, they may refer to different instances.

[0037] The terminology used in the description of the various examples in this disclosure is for the purpose of describing particular examples only and is not intended to be limiting. Unless the context explicitly indicates otherwise, an element may be one or more unless the number of elements is specifically limited. Furthermore, the term "and / or" as used in this disclosure covers any one of the listed items and all possible combinations thereof.

[0038] As described in the background section, existing technologies for microscopic water quality detection commonly suffer from the problem of missed detection due to mutual occlusion of microbial targets. Traditional target detection algorithms cannot effectively handle occluded areas, resulting in low recall and inaccurate counting, which seriously affects the reliability of water quality detection. Furthermore, existing image restoration techniques are mostly general-purpose methods, lacking specific optimization for microbial morphological characteristics. The restoration results often produce artifacts that do not conform to biological characteristics, failing to meet the requirements for accurate identification. Therefore, this embodiment provides a method, system, device, and medium for microbial occlusion identification, as detailed in the following examples:

[0039] Example 1

[0040] See Figure 1 , Figure 1 A schematic flowchart of a method for identifying microorganism occlusion is shown, wherein the method includes:

[0041] S1: Input the original microbial image into the image segmentation model to segment the foreground microorganisms and obtain the segmentation mask of all foreground microorganisms; and use a pre-trained deep learning classification network to identify the microbial category of each foreground microorganism to obtain the initial set of foreground microorganism categories;

[0042] Specifically, in this embodiment, the original microbial images are first acquired using a microscope imaging device, and the original microbial images are preprocessed to obtain preprocessed original microbial images; then, the foreground microorganisms in the preprocessed original microbial images are segmented using an image segmentation model to obtain segmentation masks for all foreground microorganisms; finally, a pre-trained deep learning classification network is used to identify the microbial category of each foreground microorganism to obtain an initial set of foreground microbial categories.

[0043] It should be noted that, in this embodiment, during the early image acquisition process, parameters such as magnification, exposure time, and frame rate are adjusted according to sample characteristics to ensure that the image quality meets the requirements of subsequent processing. Simultaneously, pre-processing of the original microbial images includes denoising using a bilateral filter to preserve edge information, brightness and contrast correction using histogram equalization to improve image quality, standardizing the image size to a uniform specification while maintaining the aspect ratio, and finally normalizing pixel values ​​to adapt to the input requirements of the deep learning model. Simultaneously, a SAM (Segment Anything Model) segmentation model or other segmentation networks are used to segment the foreground targets in the pre-processed original microbial images. Accurate segmentation masks for all foreground microorganisms are obtained through automatically generated grid cue points. The segmentation process includes post-processing steps such as mask generation, small target filtering, and hole filling to ensure the integrity and accuracy of the segmentation results. Furthermore, conventional techniques in the field are used to train a deep learning classification network to obtain a deep learning classification network capable of recognizing microbial categories. The final initial set of foreground microbial categories includes the coordinate information, category information, confidence score, and mask information for each foreground microorganism.

[0044] In this embodiment, it should also be noted that the foreground microorganisms refer to microorganisms located above that have an overlapping occlusion relationship, as well as individual microorganisms that do not have an overlapping occlusion relationship.

[0045] S2: Input all segmentation masks and the original microbial image into the pre-constructed microbial missing generation model for image inpainting to obtain the repaired microbial image;

[0046] It should be noted that, in this embodiment, a microbial deletion generation model is first constructed, as follows:

[0047] Two images were selected from a microbial image library, and their main microorganisms were labeled A and B, respectively. Microorganisms A and B can be of the same species or different species. Figure 2 As shown; a mask for microorganism A is obtained using an instance segmentation algorithm or manual annotation. An affine transformation (rotation, translation, scaling) is applied to obtain image A', which is then superimposed on image B to construct an occluded sample, as shown. Figure 3As shown; the image of B after being occluded by A' + the occluded area mask is used as the input for inpainting, and the original unoccluded image of B is used as the inpainting target to form a microbial image inpainting fine-tuning pair; the mask boundary can be moderately expanded (1-3 pixels) to cover the edge aliasing area, and the blending and feathering strategies are used to reduce hard edge artifacts; the above process is repeated to generate sufficient fine-tuning samples, which are divided into training, validation and test sets according to the proportion;

[0048] The SD-XL Inpainting model was then trained using the training set. To achieve efficient domain adaptation, LoRA (Low-Rank Adaptation) was employed to fine-tune the pre-trained model. The core idea of ​​LoRA is to decompose weight updates into the product of low-rank matrices. ,in , , Where: d represents the input dimension of the original weight matrix, i.e., the dimension of the output feature vector of the previous layer of the neural network; k represents the output dimension of the original weight matrix, i.e., the dimension of the output feature vector of the current layer of the neural network; r is the rank parameter in low-rank adaptation, used to control the complexity and computational cost of adaptation, and satisfies that r is much smaller than the smaller value of d and k, so as to achieve efficient parameter updates. During forward propagation, the original weights... The combination with LoRA weights can be expressed as , where h is the output feature vector of the current layer, x is the input feature vector, and the scaling factor α is used to control the adaptation strength. The fine-tuning process mainly performs LoRA adaptation on the cross-attention layer and convolutional layer in the U-Net network of the SD-XL Inpainting model, keeping the VAE and text encoder parameters frozen, to ensure effective domain transfer while preserving general prior knowledge.

[0049] It should be noted that in this embodiment, the SD-XL Inpainting model uses the stable-diffusion-xl-1.0-inpainting-0.1 model as its basic architecture. This model is built on the diffusion probability model theory. Its core components include an encoder (variational autoencoder) based on the VAE architecture, which is responsible for compressing and encoding the input image into a latent space representation; a U-Net denoising network as the main body of the diffusion model, which realizes image generation through a stepwise denoising process; and a decoder component, which remaps the latent representation back to the original image space to complete the image reconstruction process.

[0050] S3: Perform foreground microbial segmentation and microbial category identification on the repaired microbial image. If a foreground microorganism that does not appear in the initial foreground microbial set is identified, the segmentation mask of the foreground microorganism and the repaired microbial image are input into the pre-constructed microbial missing generation model for image repair until no foreground microorganism that did not appear is identified, and the final microbial occlusion identification result is obtained.

[0051] It should be noted that in this embodiment, when a foreground microorganism not present in the initial foreground microorganism set is identified, the foreground mask of the newly identified foreground microorganism is re-entered into the repair step for the next round of iteration processing. At the same time, in order to avoid infinite loops, a maximum iteration limit (usually 3-5 times) is set in this embodiment, and the repair is forcibly terminated when the maximum iteration limit is reached. When no foreground microorganism not present in the initial foreground microorganism set is identified, the identification process ends and the final identification result is obtained. The identification result is the set of foreground microorganism categories obtained from the last identification.

[0052] Specifically, in this embodiment, an image segmentation model is used for precise foreground segmentation to accurately locate the occluded area. Then, a generative repair model based on a stable diffusion model and LoRA fine-tuning technology is specifically trained for microbial data to generate high-quality repair content that conforms to the morphological and textural characteristics of microorganisms. Finally, through a closed-loop processing mechanism of "initial identification - generation repair - secondary identification - iterative control", the system automatically determines whether there are new foreground microbial targets and decides whether to continue iterating, which significantly improves the identification recall rate and counting accuracy, providing reliable technical support for automatic microscopic water quality detection.

[0053] Example 2

[0054] See Figure 4 The present invention also provides a microbial occlusion identification system, which is used in any of the above-described microbial occlusion identification methods, the system comprising:

[0055] The initial identification module 100 is used to input the original microbial image into the image segmentation model for foreground microbial segmentation, obtain the segmentation mask of all foreground microorganisms, and use a pre-trained deep learning classification network to identify the microbial category of each foreground microorganism, thereby obtaining an initial set of foreground microbial categories.

[0056] The image inpainting module 200 is used to input all segmentation masks and the original microbial images into a pre-constructed microbial missing generation model for image inpainting, so as to obtain the repaired microbial images.

[0057] The secondary recognition module 300 is used to segment the foreground microorganisms and identify the microorganism categories in the repaired microbial image. If a foreground microorganism that does not appear in the initial foreground microorganism set is identified, the segmentation mask of the foreground microorganism and the repaired microbial image are input into the pre-constructed microbial missing generation model for image repair until no foreground microorganisms that do not appear are identified, and the final microbial occlusion recognition result is obtained.

[0058] Furthermore, the original microbial image is input into the image segmentation model for foreground microbial segmentation, resulting in a segmentation mask for all foreground microorganisms, specifically:

[0059] Raw microbial images were acquired using a microscope imaging device, and the raw microbial images were preprocessed to obtain preprocessed raw microbial images.

[0060] The image segmentation model was used to segment the foreground microorganisms in the preprocessed original microbial image, and the segmentation mask of all foreground microorganisms was obtained.

[0061] Furthermore, the microbial deletion generation model was obtained by training the SD-XL Inpainting model, which was fine-tuned by LoRA.

[0062] It should be noted that the modules in the system of Embodiment 2 correspond to the steps in the method of Embodiment 1. The steps in the method of Embodiment 1 have been described in detail in Embodiment 1, and the module content in the system will not be described in detail in this Embodiment 2.

[0063] Example 3

[0064] See Figure 5 This embodiment also provides a computer device, including a system memory 1005 and a processor 1001. The system memory 1005 stores a computer program, and the processor 1001 executes the computer program to implement the steps of any of the methods described above.

[0065] It should be noted that the processor 1001 is used to execute the steps in the above method embodiments according to the instructions in the program code. Alternatively, when the processor 1001 executes the computer program, it implements the functions of each module / unit in the above system / device embodiments.

[0066] Specifically, in this embodiment, the computer program can be divided into one or more modules / units, which are stored in the system memory 1005 and executed by the processor 1001 to complete this application. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the terminal device.

[0067] The terminal device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor 1001 and a system memory 1005. Those skilled in the art will understand that this does not constitute a limitation on the terminal device; it may include more or fewer components than shown in the figures, or a combination of certain components, or different components. For example, the terminal device may also include an input / output device 1003, a network access device 1002, a bus 1006, etc.

[0068] The processor 1001 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0069] System memory 1005 can be an internal storage unit of the terminal device, such as a hard drive or RAM. System memory 1005 can also be a storage device 1004 of the terminal device, such as an external hard drive, SmartMedia Card (SMC), Secure Digital (SD) card, or FlashCard. Furthermore, system memory 1005 can include both internal storage units and storage device 1004. System memory 1005 is used to store computer programs and other programs and data required by the terminal device. System memory 1005 can also be used to temporarily store data that has been output or will be output.

[0070] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0071] Example 4

[0072] This embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0073] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), registers, hard disks, optical fibers, compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof, or any other form of computer-readable storage medium in the art.

[0074] An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and storage medium can reside within an application-specific integrated circuit (ASIC). In embodiments of the invention, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device.

[0075] Example 5

[0076] This embodiment also provides a computer program product containing instructions that, when executed by a cluster of computer devices, cause the cluster of computer devices to perform the method described in Embodiment 1.

[0077] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying microorganisms by occlusion, characterized in that, The methods include: The original microbial image is input into the image segmentation model to segment the foreground microorganisms, and a segmentation mask of all foreground microorganisms is obtained. The pre-trained deep learning classification network is used to identify the microbial category of each foreground microorganism, thus obtaining an initial set of foreground microbial categories. All segmentation masks and the original microbial image are input into a pre-constructed microbial missing generation model for image inpainting to obtain the repaired microbial image; The restored microbial image is segmented into foreground microorganisms and the microbial category is identified. If a foreground microorganism that does not appear in the initial foreground microorganism set is identified, the segmentation mask of the foreground microorganism and the restored microbial image are input into a pre-constructed microbial missing generation model for image restoration until no foreground microorganisms that do not appear are identified, and the final microbial occlusion recognition result is obtained. The microbial deletion generation model was obtained by training the SD-XL Inpainting model, which was fine-tuned by LoRA.

2. The method for identifying microorganisms by occlusion according to claim 1, characterized in that, The original microbial image is input into the image segmentation model for foreground microbial segmentation, resulting in a segmentation mask for all foreground microorganisms, specifically: Raw microbial images were acquired using a microscope imaging device, and the raw microbial images were preprocessed to obtain preprocessed raw microbial images. The image segmentation model was used to segment the foreground microorganisms in the preprocessed original microbial image, and the segmentation mask of all foreground microorganisms was obtained.

3. The method for identifying microorganisms by occlusion according to claim 1, characterized in that, The set of foreground microbial categories includes the coordinates, category information, confidence level, and mask information of each foreground microorganism.

4. A microbial occlusion identification system, characterized in that, This system is used in a microbial occlusion identification method according to any one of claims 1-3, the system comprising: The initial identification module is used to input the original microbial image into the image segmentation model for foreground microbial segmentation, obtain the segmentation mask of all foreground microorganisms, and use a pre-trained deep learning classification network to identify the microbial category of each foreground microorganism, thereby obtaining an initial set of foreground microbial categories. The image inpainting module is used to input all segmentation masks and the original microbial images into a pre-constructed microbial missing generation model for image inpainting, and obtain the repaired microbial images; The secondary recognition module is used to segment foreground microorganisms and identify microorganism categories in the repaired microbial image. If a foreground microorganism that does not appear in the initial foreground microorganism set is identified, the segmentation mask of the foreground microorganism and the repaired microbial image are input into the pre-constructed microbial missing generation model for image repair until no foreground microorganisms that do not appear are identified, and the final microbial occlusion recognition result is obtained.

5. A microbial occlusion identification system according to claim 4, characterized in that, The original microbial image is input into the image segmentation model for foreground microbial segmentation, resulting in a segmentation mask for all foreground microorganisms, specifically: Raw microbial images were acquired using a microscope imaging device, and the raw microbial images were preprocessed to obtain preprocessed raw microbial images. The image segmentation model was used to segment the foreground microorganisms in the preprocessed original microbial image, and the segmentation mask of all foreground microorganisms was obtained.

6. A computer device comprising a system memory and a processor, wherein the system memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 3.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method described in any one of claims 1 to 3.

8. A computer program product containing instructions, characterized in that, When the instructions are executed by a cluster of computer devices, the cluster of computer devices causes the cluster of computer devices to perform the method as described in any one of claims 1 to 3.

Images