335 results about "Image labeling" patented technology

The invention relates to an automatic data labeling method of an ISP image signal processing visual sensor. The method comprises the following steps: carrying out noise reduction and signal enhancement on an original image data stream acquired by an ISP image signal processing visual sensor to obtain a standardized image sequence; performing time sequence feature extraction to obtain a time sequence feature tensor; performing spatial feature extraction to obtain spatial-temporal feature representation; inputting a double-fold teacher-student network for semi-supervised training to obtain an initial image annotation model; performing quality evaluation based on the prediction probability distribution and the time sequence consistency to obtain a labeling result with a quality score; and storing a high-quality sample in the labeling result with the quality score into a sample buffer pool, carrying out sample screening based on entropy and distance measurement, and carrying out parameter updating on the initial image labeling model through an incremental learning strategy to obtain a target image labeling model. According to the invention, the reliability evaluation of the labeling result is realized, and the accuracy of automatic labeling is ensured.

The invention discloses a pavement crack semantic segmentation method based on Transform and CNN architecture, and relates to the technical field of pavement crack detection, and the method comprises the steps: obtaining a pavement image, marking a crack disease image in the pavement image through an image marking tool, and constructing a data set based on the marked crack disease image; preprocessing the data set, and constructing and training a self-adaptive semantic segmentation model combining a self-attention neural network and a convolutional neural network based on the preprocessed data set; and deploying the trained adaptive semantic segmentation model to a local platform, and inputting the crack disease image into the adaptive semantic segmentation model for segmentation and marking. According to the method, a dynamic scale selector composed of a texture perceptron and a scale regulation and control unit is introduced, the size of a convolution kernel is dynamically adjusted according to the local texture complexity of an image, and fine modeling of different structure areas is achieved.

The invention discloses a rock slice mineral identification method, a medium and equipment, and relates to the technical field of mineral identification, the method comprises the following steps: obtaining single polarized light images and orthogonal polarized light images of a rock slice under a plurality of polarized light angles, and carrying out image labeling to obtain a label image set; a rock slice mineral recognition model is constructed based on a U-Net architecture, the left part of the rock slice mineral recognition model of the U-Net architecture is connected with five feature extraction blocks through four downsamples, each feature extraction block is formed by connecting a convolution and a Swing-Transform Stage, and features are output by the Swing-Transform Stage; and training and testing the model by using the label image set, and inputting a to-be-detected rock slice image into the trained model to obtain a mineral segmentation result. According to the scheme of the invention, the method is high in recognition capability for complex mineral images under polarized light of different angles.

The invention discloses a traffic image labeling method and device based on a D-S evidence theory and a medium, and relates to the technical field of image labeling. According to the method, two different types of real-time models are simultaneously utilized to process the same image in parallel, and two uncertainty detection results of the same image are obtained; a KM algorithm is utilized to match targets in two uncertainty detection results, then a prior credibility score is utilized to reduce a conflict coefficient, and two different types of small model prediction results are combined through an uncertainty theory. According to the traffic image labeling method and device based on the D-S evidence theory and the medium, the strong generalization ability of the open set model is utilized to make up for the defects of a closed set model, the use of a large model to improve the hardware dependence is avoided, the precision problem of a single small model is made up under the condition that the efficiency is ensured, and the iteration duration of the model is shortened.

The present application relates to the field of automatic data labeling, and particularly relates to an image labeling method, a computer device for implementing the method, and a computer storage medium for implementing the method. The image labeling method comprises: using an image segmentation model to perform segmentation processing on original images from multiple viewpoints, so as to output segmented mask images of the original images from the viewpoints, wherein each segmented mask image includes labeling information for indicating the contour of a target object; using a neural radiance field to perform multi-viewpoint fusion on the segmented mask images from the viewpoints, so as to correct single-viewpoint errors, and generating corrected first mask images; and performing multi-frame timing fusion on the first mask images on the basis of a timing relationship, so as to correct single-frame errors, and generating a second mask image from the viewpoint of a bird's eye view.

The invention provides a fracture morphology automatic identification and learning method based on multi-modal fusion learning, and relates to the field of metal fracture morphology analysis, and the method comprises the steps: S1, constructing a data set; s2, carrying out fracture image shooting and image annotation, and after annotation is completed, dividing a data set into a training set, a verification set and a test set; step S3, carrying out multi-modal fusion learning; and S4, feature importance analysis and SHAP analysis are carried out. According to the method, image recognition and machine learning are combined, accurate prediction of dimple fracture and brittle fracture areas of the aluminum alloy is achieved, meanwhile, key factors influencing the metal fracture morphology can be effectively obtained, and therefore an important reference basis is provided for effective optimization of material performance.

The invention provides a three-dimensional target marking method and device based on a point cloud data-assisted two-dimensional image, and the method comprises the steps: fusing precise three-dimensional point cloud information obtained by a laser radar and two-dimensional image data collected by a camera based on a point cloud data and image data cooperation three-dimensional marking method; and a dynamic high-precision three-dimensional target marking system is constructed. Specifically, a space-image relationship is established through three-dimensional point cloud information and a two-dimensional image plane, and three-dimensional information of a target is mapped into a two-dimensional image, so that geometric and semantic information labeling is carried out on the target in the two-dimensional image, information loss of two-dimensional visual data in dimensions such as depth, attitude and space structure is made up, and the accuracy of the three-dimensional information labeling is improved. And meanwhile, when the three-dimensional point cloud is unmeasurable, a three-dimensional target labeling mode is dynamically adjusted in combination with two-dimensional image labeling information. According to the method, through dynamic high-precision three-dimensional target labeling, the obtained image-label can be used for training and reasoning of a three-dimensional target detection network, and the method has a wide application prospect.

The invention discloses an intelligent bone marrow megakaryocyte morphology detection method which comprises the following steps: S1, data preparation: collecting and preprocessing a digital large map of a bone marrow smear, labeling megakaryocytes, and establishing a labeled sample; s2, generating and sorting a sample: extracting a small graph sample by taking megakaryocyte as a center; s3, constructing a deep learning model: constructing and training a deep convolutional neural network, and performing model optimization and performance improvement by using the generated small image sample set; s4, large image detection and reasoning: calling the trained model for reasoning by adopting a sliding window mechanism, and fusing detection results of a plurality of small windows back to an original large image through confidence weighting and a non-maximum suppression strategy; and S5, target cell segmentation: carrying out target segmentation on the detected megakaryocyte, and introducing a pyramid structure for cells with different sizes to obtain an accurate segmentation mask of each target cell. According to the method, the megakaryocyte detection and segmentation precision is improved through large image labeling, small image training and a sliding window reasoning strategy.

The invention discloses a semi-supervised medical image segmentation method based on causal uncertainty decomposition, and belongs to the field of medical image processing and artificial intelligence. According to the method, a segmentation model of teacher and student architectures is constructed, and a causal uncertainty decomposition module, a self-adaptive consistency learning module and a topology perception consistency loss module are integrated. The total uncertainty is decomposed into cognitive uncertainty and random uncertainty, so that targeted processing is realized; a dual-path weight fusion and differential modulation strategy is adopted to realize pixel-level adaptive learning; and a Betti number is introduced to calculate a topological distance, so that the integrity of an anatomical structure is kept. According to the method, under the condition that only 5%-20% of annotation data is used, the Dice coefficient on multiple medical image data sets is increased by 3.2%-4.8%, the segmentation precision and the boundary positioning accuracy are remarkably improved, the segmentation problem under the condition that medical image annotation is scarce is effectively solved, and the method has important clinical application value.

The invention discloses a detection method and system for non-inductive warehouse goods in and out of a warehouse, and the method is characterized in that the method comprises the following steps: collecting image information of an electric power fitting warehouse, and obtaining a to-be-detected image through the work of image labeling, image preprocessing, image enhancement and the like; analyzing the to-be-measured image through a polygon approximation method, and calculating the contour of the to-be-measured target; extracting a feature map by using a neural network, and identifying a to-be-detected target by using a coding and decoding method so as to obtain a plurality of parallax transformation modules; through the output of the plurality of parallax conversion modules, the to-be-detected image carries out non-inductive detection on the goods in the electric power fitting warehouse.

The invention discloses an image labeling method for oil and gas field station working condition recognition and a storage medium. The method comprises the steps that S1, an oil leakage target area in a two-dimensional image is expressed as a three-dimensional equivalent set function; s2, designing a binary step function to initialize an equivalent set function, and roughly positioning a potential target area; s3, designing an energy functional with expected difference, and defining the energy functional as a line integral along the zero-equivalence contour of the equivalence set function; s4, the equivalent set function is evolved, a gradient flow minimization energy functional is solved, and a zero equivalent set contour, namely the boundary of the optimal oil leakage target area is obtained and serves as a mask boundary; and S5, verifying whether the obtained mask boundary meets requirements or not. The method has the advantages of high efficiency, high stability and high precision in data labeling and training data set construction, and the problem of low labeling efficiency of the pipeline petroleum leakage sample can be effectively solved.

An AI auxiliary image annotation method based on semi-supervised learning comprises the following steps: acquiring a to-be-annotated image, extracting multilayer features of the image by using a deep learning network, and generating a feature map; the global context information and the local feature information are fused, and enhanced feature representation is generated and used for contour prediction; on the basis of enhanced feature representation, modeling a relationship between vertexes by adopting a graph convolutional network, and constructing a dynamic graph structure; according to the dynamic graph structure, calculating self-adaptive matching weights among the vertexes to realize a dynamic vertex pairing strategy; predicting an initial contour position by combining global and local features through a dynamic vertex pairing strategy; based on the predicted initial contour, a deformable convolutional network is adopted to realize progressive optimization of the contour; in the progressive optimization process, vertex position information is updated in real time, and the graph structure relation is adjusted; and finally, outputting an accurate object contour labeling result, and completing an image labeling task.

The embodiment of the invention relates to a multi-round medical image annotation method and device introducing a cross validation mechanism, and the method comprises the steps: transmitting a first image to a plurality of expert annotation interfaces, and merging the feedback annotation sets of all interfaces to obtain a first annotation set; setting cross validation items of every two annotation boxes in the first annotation set and constructing a corresponding cross validation matrix; on the basis of a cross validation matrix, dispute-free annotation box extraction, dispute-free annotation box aggregation and consistency score estimation are carried out; if the score is lower than a threshold value, a round of multi-person labeling is initiated again, otherwise, when the dispute labeling box set is empty, the non-dispute labeling box set serves as a final labeling result, when the dispute labeling box set is not empty, reexamination is conducted on the dispute labeling box set, and a set of the reexamination result and the non-dispute labeling box set serves as the final labeling result. According to the method, the labeling error of each image can be reduced, the overall labeling quality of all the images is improved, and the stability of the overall labeling quality is guaranteed.

The present application relates to an image annotation method, device, electronic device and storage medium, wherein the image annotation method includes: by obtaining prompt words, annotating the image to be annotated according to the prompt words, unifying the target category corresponding to the prompt words; extracting the text features of the target category; obtaining the image to be annotated, extracting the image features of the image to be annotated; solidifying the text features of the target category to obtain text-strongly correlated features; performing feature enhancement on the image features of the image to be annotated to obtain image enhancement features; inputting the text-strongly correlated features and the image enhancement features into a preset training model for detection and segmentation to obtain an initial detection and segmentation result; performing accuracy verification on the initial detection and segmentation result to obtain a target detection and segmentation result, and saving the target detection and segmentation result to the target annotation file corresponding to the image to be annotated. Through the present application, the accuracy of image annotation is improved and the workload of manual review is reduced.

The embodiment of the invention discloses an image auditing method and device based on a co-creation annotation mode, equipment and a medium. A specific embodiment of the method comprises the following steps: acquiring medical film and television data as initial data; preprocessing the initial data to obtain a task image set; distributing the task image set to at least one student terminal; in response to the detected annotation information sent by the student terminal, generating annotation feedback information according to the annotation information; sending the labeled feedback information to the student terminal; in response to the received annotation task image submitted by the student terminal, performing preliminary quality inspection on the annotation task image; sending the preliminary quality inspection result to an expert end; and in response to received auditing information corresponding to the preliminary quality inspection result sent by the expert terminal, sending the auditing information to the student terminal. According to the embodiment, a co-creation annotation mode is combined with a deep learning technology, so that efficient, accurate and high-quality image annotation auditing is realized, and the overall quality of annotation data is improved.

The invention relates to the technical field of brain tumor image segmentation, in particular to a brain tumor segmentation method and system based on anatomical perception symmetric comparison and cross-modal migration. The method comprises the following steps: carrying out data preprocessing on acquired multi-modal MRI image data; constructing a brain tumor segmentation model based on anatomical perception symmetric comparison and cross-modal migration; performing model training based on a two-stage decoupling training strategy; and performing model reasoning by using the trained model, and outputting a brain tumor segmentation result. Through a self-supervised learning framework, pre-training is carried out by using unmarked MRI data, dependence on a large-scale marked data set is greatly reduced, the problems of time consumption and high cost of medical image marking are solved, and the applicability of a model in a limited data scene is improved.

The invention discloses a semi-supervised image annotation method based on a limited label data set, and the method comprises the steps: taking a FixMatch frame as a basis, and integrating a learnable batch normalization module, a dual-scale parallel convolution module, a content and style separation dual-branch module and a dynamic residual gating module in a ResNet backbone network, the stability of feature extraction and the adaptive capacity to enhanced disturbance are improved. For a label-free sample, a multi-level pseudo-label fusion mechanism is provided, prediction distribution of weak, medium and strong enhanced views is synthesized, and high-confidence pseudo-labels are generated through confidence weighted fusion of multi-level enhanced views and comparison and screening with a category threshold. On the basis, a joint loss function composed of label supervision loss and pseudo label consistency loss is constructed, and a plurality of key control parameters in FixMatch + + are adjusted and optimized in a pre-experiment and grid search combined mode to obtain a group of optimal parameters of the model. Finally, a user inputs a label-free image into the trained FixMatch + + model, and the model can automatically generate a high-confidence pseudo label, so that the number of labeled samples in a limited labeled image set is increased, and the classification precision is improved. By implementing the method, the manual annotation cost can be reduced, and efficient and reliable support is provided for image analysis and recognition tasks.

The invention discloses an image annotation method, a client, terminal equipment and a storage medium, and relates to the technical field of image processing. The method is applied to a client and comprises the following steps: after a to-be-labeled image is rendered to a target container, calculating the scaling of the to-be-labeled image before and after rendering and the offset of the to-be-labeled image relative to the target container after the to-be-labeled image is rendered to the target container; performing target detection on the to-be-labeled image based on a preset target detection model to obtain a target detection result of the to-be-labeled image; correcting the target detection result through the offset and the scaling to obtain target annotation information of the to-be-annotated image; and according to the target annotation information, annotating the to-be-annotated image rendered into the target container. According to the invention, the accuracy of client image annotation can be improved.

The invention discloses an image annotation method and device, equipment and a storage medium. The method comprises the steps that a first image is acquired, and the first image comprises a plurality of objects to be labeled; performing image segmentation on the first image by using the first large segmentation model to obtain a segmented thermodynamic diagram, the segmented thermodynamic diagram comprising a plurality of pixel regions, and each pixel region corresponding to one to-be-labeled object; cutting the first image according to the plurality of pixel regions to obtain a second image corresponding to each to-be-labeled object; for each second image, carrying out image-text matching on the second image and a preset category label by utilizing a first image-text matching large model; and determining the preset category label matched with the second image as a labeling result of the to-be-labeled object in the second image. According to the image annotation method provided by the embodiment of the invention, the to-be-annotated objects of multiple categories can be annotated.

The invention discloses an information processing method and device for fish hemorrhagic disease detection, equipment and a medium, and relates to the technical field of intelligent agriculture, and the method comprises the steps: obtaining a first healthy fish image and a second healthy fish image, converting the second healthy fish image into a diseased fish image through a cyclic generative adversarial network, and obtaining a diseased fish image; constructing an initial training set by using the healthy fish image and the diseased fish image; labeling the fish individuals in the initial training set by using an image labeling model, and obtaining a target training set obtained after the expert performs health state classification on the fish individuals in the labeled training set; and training the initial image detection model based on the target training set to obtain a to-be-detected fish image, and performing hemorrhagic disease detection on fishes in the to-be-detected fish image by using the target image detection model. A healthy fish image sample is converted into a diseased fish image through the cyclic generative adversarial network, and the problem that the model cannot be trained due to insufficient abnormal samples is solved.

The invention discloses an accurate safflower picking method integrating a main camera technology and an auxiliary camera technology. According to the method, the process of picking operation of the safflower harvesting robot is divided into five parts of main shooting image acquisition and data set construction, auxiliary shooting image acquisition and filament area segmentation, global information acquisition and mechanical arm rapid positioning, filament feeding real-time monitoring and accurate picking, and continuous picking target. Specifically, a binocular camera installed at the top of a harvesting robot is used as a main camera, safflower image data is collected in a natural environment from an overlook angle, image labeling software is used for carrying out fine label labeling on the image data, a safflower data set is constructed, and a deep learning model is used for training and optimization; a small industrial camera is installed in the side direction of a cutting type end effector to serve as an auxiliary camera, safflower image data are collected in the natural environment from the side view angle, image segmentation processing is conducted on the image data through a clustering algorithm, and an accurate filament area is obtained; when the safflower harvesting robot enters a field to perform harvesting operation, a main-shooting binocular camera overviews global information of target plants in a picking space, and in combination with a pre-trained deep learning model, center point coordinates of top planes of filaments are obtained, and a mechanical arm is guided to drive a cutting type end effector to quickly reach the positions near the filaments; the mechanical arm drives the cutting type end effector to continuously and slowly descend, a negative pressure fan of the harvesting robot is started, the filaments are kept vertical under the suction effect of the negative pressure fan, a cutting plane is exposed, meanwhile, an auxiliary camera small industrial camera is started to monitor the filament feeding condition in real time, and the interior of the end effector is preset as a picking area; when the lowest points of the cut filaments enter a picking area, the mechanical arm stops moving, the end effector is triggered to start picking operation, and the end effector stops working after picking is completed; after one-time picking is completed, picking of the next target is carried out according to a pre-planned picking path until all the targets in the area are picked. The accurate safflower picking method integrating the main and auxiliary camera shooting technology can be provided for intelligent safflower picking, the accuracy and high efficiency of the picking action are ensured, and the quality and efficiency of safflower picking are effectively improved.

The invention provides a diabetic retinopathy total fundus dynamic labeling system, and relates to the technical field of medical image processing. The system comprises a data acquisition module, an image labeling module, a multi-mode integration display module and a user interaction module. According to the system, the diabetic retinopathy marking model is utilized to realize lesion area identification and lesion type marking on various types of fundus image data, so that a doctor can clearly and quickly know the progress of an illness state without artificial experience; the full fundus image data marked with the lesion area and the lesion type, the image data of the posterior pole and the macular area and the retinal blood vessel image data are fused, the lesion area and the lesion type of any area of the full fundus of the user to be detected can be displayed on a display interface, a visual blind area is eliminated, and the detection accuracy is improved. The user can view any interested area of the whole fundus, so that comprehensive analysis of the diabetic retinopathy is realized, and the diagnosis and treatment efficiency of the diabetic retinopathy is comprehensively improved.

The invention discloses a field core data feature automatic identification and extraction method and system. The method comprises the following steps: S1, acquiring a field core image and preprocessing the field core image; s2, performing image labeling on the preprocessed core image to obtain image classification and core attribute description; s3, training a core target detection and classification model by using the marked image data; s4, training a rock core attribute identification model by using a rock core target detection and classification result and rock core attribute information data; and S5, based on the trained core target detection and classification model and the core attribute identification model, carrying out target detection and classification on the core image, and carrying out attribute feature identification on a detected target object. According to the method, automation and intelligentization of the whole process from data collection, preprocessing and feature extraction to model training and application are realized, manual intervention is greatly reduced, and the working efficiency is improved.

The invention discloses a man-machine collaborative remote sensing image intelligent labeling method and device based on a visual large model. The method comprises the steps of obtaining remote sensing image data to be labeled, and performing image enhancement processing to obtain enhanced remote sensing image data; inputting the enhanced remote sensing image data into a visual large model for mask segmentation processing, and outputting a plurality of target segmentation masks and corresponding confidence coefficients; marking layer graphs with different colors on the to-be-marked remote sensing image data to form marked remote sensing image data, and displaying the marked remote sensing image data to a user; and the user performs confirmation operation, and stores the labeled remote sensing image data as a GeoTIFF file format image with geographic reference information. In the embodiment of the invention, the image labeling precision of complex ground objects, multiple regions and the like is obviously improved.

The invention relates to an electrical cabinet pressing plate description extraction method for a complex light environment, and the method comprises the steps: collecting an electrical cabinet pressing plate image, and carrying out the data enhancement processing; performing data annotation on the picture data of the pressing plate of the electrical cabinet by using an image annotation tool; making a character recognition training set and a test set for the pressing plate description text to obtain a text recognition model for pressing plate characters; inputting the to-be-detected picture into the system to carry out definition quantitative evaluation; the to-be-detected picture meeting the definition detection requirement is input into the pressing plate description detection module for positioning, and accurate coordinates of the to-be-detected picture are obtained; a text recognition model is used for accurately recognizing the detected pressing plate description area, and detailed pressing plate description text content is extracted; and based on the obtained pressing plate description text content, carrying out precise processing on pressing plate information through a dynamic space positioning algorithm of the pressing plate image of the system integrated electrical cabinet. The invention also relates to a corresponding system, device, processor and storage medium.

The invention relates to an interactive image annotation system and method based on facial posture and eye movement tracking and positioning, and solves the problems of poor annotation precision and low efficiency caused by the fact that two interactive technologies of facial posture detection and eye movement tracking are not organically combined with an automatic image segmentation algorithm in the existing image data annotation field. According to the invention, independent operation of two interaction modes of face posture detection and eye movement tracking is realized through modular design, a proper interaction mode is selected by judging the complexity of a scene, the subsequent processing time can be saved, and through secondary sliding mean filtering processing, the detection accuracy is improved. Errors caused by data noise of a depth camera or near-infrared eye movement tracking equipment based on iris reflection and micro movement of an operator are avoided, and the filtered stable coordinates are matched with a segmentation model of the interaction control module to carry out fine extraction on a target area. A traditional segmentation model and operator interaction information are organically combined in the interaction control module, and the manual labeling process is greatly simplified.

The invention discloses a three-dimensional interactive segmentation method and device, computer equipment and a storage medium, and is applied to the technical field of computers, and the method comprises the steps: obtaining a three-dimensional fluorescent microscopic image and interactive prompt information; wherein the interactive prompt information comprises a positive sample point, a negative sample point and a preorder segmentation mask; inputting the three-dimensional fluorescent microscopic image and the interactive prompt information into a constructed interactive segmentation model, and outputting a segmentation result; based on feedback of a user on the segmentation result, performing iterative optimization on the interactive segmentation model until the segmentation result meets preset precision; according to the invention, based on deep learning, the interaction prompt information and the three-dimensional fluorescence microscopic image are fused, accurate segmentation of the neuron structure is realized, and the precision and speed of neuron image labeling are improved.

The invention relates to a sample data generation method and device, equipment and a storage medium. The method comprises the following steps: acquiring configured text indication information; generating candidate images based on the text indication information; performing object labeling on the candidate image through a pre-trained image labeling model to obtain label information of the candidate image; selecting a target image meeting a screening condition from the candidate images based on the label information, and determining the target image and the label information corresponding to the target image as sample data; the sample data is used for training a target detection model. By adopting the method, the accuracy and acquisition efficiency of the sample data can be improved.

The invention discloses a skeletal muscle ultrasonic image processing system and method for ultrasonic detection, and belongs to the technical field of medical ultrasonic image processing.The skeletal muscle ultrasonic image processing method comprises the steps that after a skeletal muscle ultrasonic image is subjected to non-local mean value dynamic filtering denoising, the contrast ratio of the skeletal muscle ultrasonic image is enhanced through nonlinear gray level transformation, and the skeletal muscle ultrasonic image is obtained; skeletal muscle areas are segmented by using the spatial continuity and similarity of skeletal muscles; on the basis of the segmented skeletal muscle region, carrying out binaryzation on a skeletal muscle ultrasonic image, calculating fractal dimensions by utilizing grid sliding statistics and weighting processing, constructing and smoothing a tensor model, calculating a main characteristic value and a characteristic vector, and positioning a skeletal muscle atrophy region through threshold judgment and clustering analysis; and performing labeling post-processing on the skeletal muscle ultrasonic image to generate a skeletal muscle ultrasonic image labeling report. The objective of the invention is to solve the technical problems of denoising, contrast enhancement, region segmentation, texture complexity quantification, atrophy region positioning, annotation report generation and the like of skeletal muscle ultrasonic images.

The invention relates to a cellular structure segmentation method and system based on multi-round fluorescence in-situ hybridization images. The invention provides a method for automatically segmenting a cell structure in a fluorescence image. The method comprises the following steps: acquiring a fluorescence image to be analyzed as image data to be processed; training data are selected from the to-be-processed image, a cell truth value label data set is made through image labeling software, and the to-be-processed image is preprocessed; training a multi-task learning-based convolutional neural network based on the marked data set, and outputting a nearest neighbor affinity image between pixels and a probability image that the pixels belong to a boundary; on the basis of a network output result, a distance transformation watershed algorithm and a graph division algorithm are adopted to generate a cell instance segmentation result; and automatically identifying a segmentation error region in the segmentation result, artificially and selectively correcting the segmentation result, and obtaining a corrected cell boundary result through edge extraction.