CN115578563A - Image segmentation method, system, storage medium and electronic equipment - Google Patents
Image segmentation method, system, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN115578563A CN115578563A CN202211281784.0A CN202211281784A CN115578563A CN 115578563 A CN115578563 A CN 115578563A CN 202211281784 A CN202211281784 A CN 202211281784A CN 115578563 A CN115578563 A CN 115578563A
- Authority
- CN
- China
- Prior art keywords
- training image
- image
- original training
- deep learning
- learning model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003709 image segmentation Methods 0.000 title claims abstract description 63
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000012549 training Methods 0.000 claims abstract description 280
- 238000013136 deep learning model Methods 0.000 claims abstract description 86
- 238000002372 labelling Methods 0.000 claims abstract description 22
- 238000012545 processing Methods 0.000 claims description 24
- 238000005070 sampling Methods 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 7
- 230000009467 reduction Effects 0.000 claims description 7
- 230000001427 coherent effect Effects 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 abstract description 5
- 230000008569 process Effects 0.000 description 10
- 230000009286 beneficial effect Effects 0.000 description 6
- 230000011218 segmentation Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 239000011800 void material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/7715—Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to an image segmentation method, a system, a storage medium and an electronic device, comprising: labeling each original training image in the training image set to obtain all first training images; training a preset deep learning model for image segmentation based on all original training images and all first training images to obtain a target deep learning model; and inputting the image to be detected into the target deep learning model to obtain an image segmentation result of the image to be detected. According to the image segmentation method, the image labeling characteristics are fused into the deep learning model, the strong learning capacity of the deep learning network is fully utilized, and the precision of image segmentation is improved through priori knowledge constraint.
Description
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to an image segmentation method, an image segmentation system, a storage medium, and an electronic device.
Background
With the development and maturity of computer vision technology, it is widely applied to various fields. Image segmentation is a key component in many visual understanding systems, and its main task is to convert a picture into some areas or objects. And still further, the example segmentation needs to completely trace out all interested objects in the picture. In recent years, as the learning ability of deep convolutional neural networks becomes stronger, the precision of example segmentation is gradually improved, and the neural networks can achieve good precision even on complex large data sets (such as COCO). However, when the existing deep learning model faces a relatively complex object overlapping scene, a lot of wrong segmentations are often generated due to the lack of prior knowledge. For example, the surface of the object has no obvious edge, and is similar to the background in imaging, which easily results in the occurrence of a void inside the object as a result of segmentation, and even divides the object into two discontinuous connected domains.
Therefore, it is desirable to provide a technical solution to the above problems.
Disclosure of Invention
In order to solve the technical problem, the invention provides an image segmentation method, an image segmentation system, a storage medium and electronic equipment.
The technical scheme of the image segmentation method is as follows:
s1, labeling each original training image in a training image set to obtain all first training images;
s2, training a preset deep learning model for image segmentation based on all original training images and all first training images to obtain a target deep learning model;
and S3, inputting the image to be detected into the target deep learning model to obtain an image segmentation result of the image to be detected.
The image segmentation method has the following beneficial effects:
according to the method, the image annotation characteristics are fused into the deep learning model, the strong learning capacity of the deep learning network is fully utilized, and the precision of image segmentation is improved through priori knowledge constraint.
On the basis of the above scheme, the image segmentation method of the present invention may be further improved as follows.
Further, the S1 includes:
judging whether any original training image has a shielded object or not to obtain a judgment result of any original training image;
when the judgment result of any original training image is yes, completely marking each occluded object in any original training image, marking each non-occluded object in any original training image, and obtaining a first training image corresponding to any original training image until obtaining a first training image corresponding to each original training image with the judgment result of yes;
and when the judgment result of any original training image is negative, labeling each object in any original training image to obtain a first training image corresponding to any original training image until the first training image corresponding to each original training image with the judgment result of negative is obtained.
The beneficial effect of adopting the further technical scheme is that: by carrying out the overlapping labeling on the training images, each object in the images keeps the morphological characteristics related to the category, and compared with a non-overlapping labeling method, the method is more consistent with the recognition process of human, so that the characteristics among the objects are more consistent, and the model learning of the characteristic relationship related to the objects is facilitated.
Further, the S2 includes:
s21, inputting any original training image into the preset deep learning model to obtain a prediction mask of an object to be segmented in any original training image; based on a continuous coherent method, acquiring a predicted bar code corresponding to a predicted mask of any original training image, and acquiring a prior bar code corresponding to a first training image corresponding to any original training image; calculating a target distance value between the predicted barcode and the prior barcode, and taking the target distance value as a topological loss value of the preset deep learning model until a topological loss value corresponding to each original training image is obtained;
and S22, adding all the topological loss values into a loss function of the preset deep learning model for back propagation to obtain and use the first deep learning model as the preset deep learning model, and returning to execute the S21 for iterative training until the preset deep learning model converges to obtain the target deep learning model.
The beneficial effect of adopting the further technical scheme is that: based on the characteristic that the topological features of the object are obvious, the topological features of the prediction mask are used as a normalization means, so that the prediction result of the model is close to the topological features of the labeled image, and the accuracy of image segmentation is improved by correcting the topological errors of the model in the training process.
Further, the preset deep learning model includes: at least two convolution layers and at least two down-sampling layers, wherein each convolution layer is correspondingly connected with one down-sampling layer; the step of inputting any original training image into the preset deep learning model to obtain the prediction mask of the object to be segmented in any original training image comprises the following steps:
extracting image features of any original training image through a first convolutional layer to obtain a first training image feature corresponding to any original training image;
performing dimensionality reduction processing on the first training image characteristic corresponding to any original training image through a first downsampling layer to obtain a second training image characteristic;
and processing the characteristics of the second training image sequentially through each residual convolution layer and each residual down-sampling layer to obtain a prediction mask of the object to be segmented in any original training image.
The technical scheme of the image segmentation system is as follows:
the method comprises the following steps: the system comprises a processing module, a training module and an operation module;
the processing module is used for: labeling each original training image in the training image set to obtain all first training images;
the training module is to: training a preset deep learning model for image segmentation based on all original training images and all first training images to obtain a target deep learning model;
the operation module is used for: and inputting the image to be detected into the target deep learning model to obtain an image segmentation result of the image to be detected.
The image segmentation system has the following beneficial effects:
the system of the invention fully utilizes the strong learning capability of the deep learning network by fusing the image annotation characteristics into the deep learning model, so as to improve the image segmentation precision through the prior knowledge constraint.
On the basis of the above scheme, an image segmentation system of the present invention may be further improved as follows.
Further, the processing module is specifically configured to:
judging whether a sheltered object exists in any original training image to obtain a judgment result of any original training image;
when the judgment result of any original training image is yes, completely marking each occluded object in any original training image, and marking each unoccluded object in any original training image to obtain a first training image corresponding to any original training image until a first training image corresponding to each original training image with the judgment result of yes is obtained;
and when the judgment result of any original training image is negative, labeling each object in any original training image to obtain a first training image corresponding to any original training image until the first training image corresponding to each original training image with the judgment result of negative is obtained.
The beneficial effect of adopting the further technical scheme is that: by carrying out the overlapping labeling on the training images, each object in the images keeps the morphological characteristics related to the category, and compared with a non-overlapping labeling method, the method is more consistent with the recognition process of human, so that the characteristics among the objects are more consistent, and the model learning of the characteristic relationship related to the objects is facilitated.
Further, the training module specifically includes: a first training module and a second training module;
the first training module is to: inputting any original training image into the preset deep learning model to obtain a prediction mask of an object to be segmented in any original training image; based on a continuous coherent method, acquiring a predicted bar code corresponding to a predicted mask of any original training image, and acquiring a prior bar code corresponding to a first training image corresponding to any original training image; calculating a target distance value between the predicted barcode and the prior barcode, and taking the target distance value as a topological loss value of the preset deep learning model until a topological loss value corresponding to each original training image is obtained;
the second training module is to: and adding all topological loss values into a loss function of the preset deep learning model for back propagation to obtain and use a first deep learning model as the preset deep learning model, and returning and calling the first training module for iterative training until the preset deep learning model is converged to obtain the target deep learning model.
The beneficial effect of adopting the further technical scheme is that: based on the characteristic that the topological features of the object are obvious, the topological features of the prediction mask are used as a normalization means, so that the prediction result of the model is close to the topological features of the labeled image, and the accuracy of image segmentation is improved by correcting the topological errors of the model in the training process.
Further, the preset deep learning model includes: at least two convolution layers and at least two down-sampling layers, wherein each convolution layer is correspondingly connected with one down-sampling layer; the first training module is specifically configured to:
extracting image features of any original training image through a first convolutional layer to obtain a first training image feature corresponding to any original training image;
performing dimensionality reduction processing on the first training image characteristic corresponding to any original training image through a first downsampling layer to obtain a second training image characteristic;
and processing the characteristics of the second training image sequentially through each residual convolution layer and each residual down-sampling layer to obtain a prediction mask of the object to be segmented in any original training image.
The technical scheme of the storage medium of the invention is as follows:
the storage medium has stored therein instructions which, when read by a computer, cause the computer to carry out the steps of an image segmentation method according to the invention.
The technical scheme of the electronic equipment is as follows:
comprising a memory, a processor and a computer program stored on the memory and being executable on the processor, characterized in that the processor, when executing the computer program, causes the computer to perform the steps of an image segmentation method according to the invention.
Drawings
FIG. 1 is a flowchart illustrating an image segmentation method according to an embodiment of the present invention;
fig. 2 is a structural diagram of a preset deep learning model in an image segmentation method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an image segmentation system according to an embodiment of the present invention.
Detailed Description
As shown in fig. 1, an image segmentation method according to an embodiment of the present invention includes the following steps:
s1, labeling each original training image in the training image set to obtain all first training images.
The training image set comprises a plurality of original training images, the original training images are images of any type and used for training a preset deep learning model, each original training image comprises one or more objects, and the types of the objects are not limited.
Wherein the first training image is: and (4) marking the object in the original training image to obtain an image.
It should be noted that the image in the present embodiment should have objects with uniform topological features (the same or similar shapes). Types of images include, but are not limited to: cells, balls, cell phones, and televisions, among others. These objects are characterized in that each object has similar topological features. For example, cells of a certain type may be round or bar-shaped, but cells of the same type are all similarly shaped.
And S2, training a preset deep learning model for image segmentation based on all the original training images and all the first training images to obtain a target deep learning model.
The preset deep learning model adopts a recent example segmentation algorithm, including but not limited to a one-stage algorithm and a two-stage algorithm, and only needs to be used for realizing image segmentation, which is not limited herein.
The target deep learning model is obtained after training.
And S3, inputting the image to be detected into the target deep learning model to obtain an image segmentation result of the image to be detected.
Wherein, the image to be measured is: randomly inputting images, wherein the image type is the same as or similar to the image type when the target deep learning model is trained. For example, when the training image is a circular cell, the image to be measured should also be a circular cell or a circular-like cell.
Wherein, the image segmentation result is: and (5) carrying out image segmentation on the image to be detected to obtain a result. For example, an image to be measured including a circular cell is input to the target deep learning model and subjected to image segmentation processing, and a result of segmenting the circular cell in the image to be measured is obtained.
Preferably, the S1 includes:
and judging whether the sheltered object exists in each original training image to obtain the judgment result of each original training image.
The occluded object refers to the fact that a part of an object is occluded due to object overlapping in an original training image and cannot display a complete object.
And when the judgment result of any original training image is yes, completely marking each occluded object in any original training image, marking each non-occluded object in any original training image, and obtaining a first training image corresponding to any original training image until obtaining a first training image corresponding to each original training image with the judgment result being yes.
And when the judgment result of any original training image is negative, labeling each object in any original training image to obtain a first training image corresponding to any original training image until the first training image corresponding to each original training image with the judgment result of negative is obtained.
It should be noted that, the process of completely labeling the occluded object is as follows: and restoring the complete form of the shielded object, labeling the shielded part of the object in the corresponding original training image, and acquiring the complete characteristics of the object as far as possible.
The process of normally labeling the object in the original training image is the prior art, and is not described herein in any greater detail.
Preferably, the S2 includes:
s21, inputting any original training image into the preset deep learning model to obtain a prediction mask of an object to be segmented in any original training image; based on a continuous coherent method, acquiring a predicted bar code corresponding to a predicted mask of any original training image, and acquiring a prior bar code corresponding to a first training image corresponding to any original training image; and calculating a target distance value between the predicted barcode and the prior barcode, and taking the target distance value as a topological loss value of the preset deep learning model until a topological loss value corresponding to each original training image is obtained.
The object to be segmented may be one object or multiple objects in the image, and no limitation is set herein.
Wherein, the prediction mask is: and the object to be segmented corresponds to the mask area in the image. The predicted barcode is: and the bar code corresponding to the prediction mask is drawn by a continuous tone method. The prior barcode is: and the bar code corresponding to the labeling feature in the first training image is drawn by a continuous modulation method. The topology loss values are: a target distance value between the barcode and the corresponding a priori barcode is predicted.
It should be noted that: (1) in the training process of the preset deep learning model, a first training image corresponding to each original training is used as prior knowledge and is fused into the model for training. (2) The process of obtaining the barcode corresponding to the image mask by the continuous coherence method is the prior art, and is not described herein in detail. (3) Besides the bar code form, the distance value between the prediction mask and the priori knowledge can be obtained by adopting a mode of calculating Betti number.
And S22, adding all the topological loss values into a loss function of the preset deep learning model for back propagation to obtain and use the first deep learning model as the preset deep learning model, and returning to execute the S21 for iterative training until the preset deep learning model converges to obtain the target deep learning model.
Wherein the loss function is composed of all topology loss values and the loss values (including but not limited to cross-entropy loss) generated by the training process.
It should be noted that, the process of performing back propagation on the loss function and updating the parameters of the model for iterative training is the prior art, and is not described herein in detail.
Preferably, the preset deep learning model comprises: at least two convolution layers and at least two down-sampling layers, wherein each convolution layer is correspondingly connected with one down-sampling layer; the step of inputting any original training image into the preset deep learning model to obtain the prediction mask of the object to be segmented in any original training image comprises the following steps:
and performing image feature extraction on any original training image through the first convolutional layer to obtain a first training image feature corresponding to any original training image.
The first training image feature is an image feature obtained by performing image feature extraction on any original training image.
As shown in fig. 2, the present embodiment takes two convolution layers and two downsampling layers as an example. The preset deep learning model comprises a first convolution layer, a first down-sampling layer, a second convolution layer and a second down-sampling layer which are sequentially connected. The convolution layer is used for extracting image characteristics of an object to be segmented in the image, and the down-sampling layer is used for performing dimensionality reduction processing on the image characteristics obtained by the first convolution layer by adopting a pooling operation.
And performing dimensionality reduction processing on the first training image characteristic corresponding to any original training image through a first downsampling layer to obtain a second training image characteristic.
And processing the characteristics of the second training image sequentially through each residual convolution layer and each residual down-sampling layer to obtain a prediction mask of the object to be segmented in any original training image.
According to the technical scheme, the image annotation features are fused into the deep learning model, the powerful learning capacity of the deep learning network is fully utilized, and the accuracy of image segmentation is improved through priori knowledge constraint.
As shown in fig. 3, an image segmentation system 200 according to an embodiment of the present invention includes: a processing module 210, a training module 220, and an execution module 230;
the processing module 210 is configured to: labeling each original training image in the training image set to obtain all first training images;
the training module 220 is configured to: training a preset deep learning model for image segmentation based on all original training images and all first training images to obtain a target deep learning model;
the execution module 230 is configured to: and inputting the image to be detected into the target deep learning model to obtain an image segmentation result of the image to be detected.
Preferably, the processing module 210 is specifically configured to:
judging whether any original training image has a shielded object or not to obtain a judgment result of any original training image;
when the judgment result of any original training image is yes, completely marking each occluded object in any original training image, marking each non-occluded object in any original training image, and obtaining a first training image corresponding to any original training image until obtaining a first training image corresponding to each original training image with the judgment result of yes;
and when the judgment result of any original training image is negative, labeling each object in any original training image to obtain a first training image corresponding to any original training image until the first training image corresponding to each original training image with the judgment result of negative is obtained.
Preferably, the training module 220 comprises: a first training module 221 and a second training module 222;
the first training module 221 is configured to: inputting any original training image into the preset deep learning model to obtain a prediction mask of an object to be segmented in any original training image; based on a continuous coherent method, acquiring a predicted bar code corresponding to a predicted mask of any original training image, and acquiring a prior bar code corresponding to a first training image corresponding to any original training image; calculating a target distance value between the predicted barcode and the prior barcode, and taking the target distance value as a topological loss value of the preset deep learning model until a topological loss value corresponding to each original training image is obtained;
the second training module 222 is configured to: adding all the topological loss values into a loss function of the preset deep learning model for back propagation to obtain a first deep learning model as the preset deep learning model, and returning and calling the first training module 221 for iterative training until the preset deep learning model converges to obtain the target deep learning model.
Preferably, the preset deep learning model comprises: at least two convolution layers and at least two down-sampling layers, wherein each convolution layer is correspondingly connected with one down-sampling layer; the first training module 221 is specifically configured to:
extracting image features of any original training image through a first convolutional layer to obtain a first training image feature corresponding to any original training image;
performing dimensionality reduction processing on the first training image characteristic corresponding to any original training image through a first downsampling layer to obtain a second training image characteristic;
and processing the characteristics of the second training image sequentially through each residual convolution layer and each residual down-sampling layer to obtain a prediction mask of the object to be segmented in any original training image.
According to the technical scheme, the image annotation features are fused into the deep learning model, the powerful learning capacity of the deep learning network is fully utilized, and the accuracy of image segmentation is improved through priori knowledge constraint.
The above steps for implementing the corresponding functions of the parameters and modules in the image segmentation system 200 of the present embodiment may refer to the parameters and steps in the above embodiment of the image segmentation method, which are not described herein again.
An embodiment of the present invention provides a storage medium, including: the storage medium stores instructions, and when the instructions are read by the computer, the computer is caused to execute the steps of the image segmentation method, which may specifically refer to the parameters and the steps in the above embodiment of the image segmentation method, and are not described herein again.
Computer storage media such as: flash disks, portable hard disks, and the like.
An electronic device provided in an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and is characterized in that when the processor executes the computer program, the computer executes a step of an image segmentation method, for example, specifically, reference may be made to each parameter and step in an embodiment of the image segmentation method, which is not described herein again.
Those skilled in the art will appreciate that the present invention may be embodied as methods, systems, storage media and electronic devices.
Thus, the present invention may be embodied in the form of: may be embodied entirely in hardware, entirely in software (including firmware, resident software, micro-code, etc.) or in a combination of hardware and software, and may be referred to herein generally as a "circuit," module "or" system. Furthermore, in some embodiments, the invention may also be embodied in the form of a computer program product in one or more computer-readable media having computer-readable program code embodied in the medium. Any combination of one or more computer-readable media may be employed. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Claims (10)
1. An image segmentation method, comprising:
s1, labeling each original training image in a training image set to obtain all first training images;
s2, training a preset deep learning model for image segmentation based on all original training images and all first training images to obtain a target deep learning model;
and S3, inputting the image to be detected into the target deep learning model to obtain an image segmentation result of the image to be detected.
2. The image segmentation method according to claim 1, wherein the S1 includes:
judging whether a sheltered object exists in any original training image to obtain a judgment result of any original training image;
when the judgment result of any original training image is yes, completely marking each occluded object in any original training image, and marking each unoccluded object in any original training image to obtain a first training image corresponding to any original training image until a first training image corresponding to each original training image with the judgment result of yes is obtained;
and when the judgment result of any original training image is negative, labeling each object in any original training image to obtain a first training image corresponding to any original training image until the first training image corresponding to each original training image with the judgment result of negative is obtained.
3. The image segmentation method according to claim 1, wherein the S2 includes:
s21, inputting any original training image into the preset deep learning model to obtain a prediction mask of an object to be segmented in any original training image; based on a continuous coherent method, acquiring a predicted bar code corresponding to a predicted mask of any original training image, and acquiring a prior bar code corresponding to a first training image corresponding to any original training image; calculating a target distance value between the predicted barcode and the prior barcode, and taking the target distance value as a topological loss value of the preset deep learning model until a topological loss value corresponding to each original training image is obtained;
and S22, adding all the topological loss values into a loss function of the preset deep learning model for back propagation to obtain and use the first deep learning model as the preset deep learning model, and returning to execute the S21 for iterative training until the preset deep learning model converges to obtain the target deep learning model.
4. The image segmentation method according to claim 3, wherein the preset deep learning model comprises: at least two convolution layers and at least two down-sampling layers, wherein each convolution layer is correspondingly connected with one down-sampling layer; the step of inputting any original training image into the preset deep learning model to obtain the prediction mask of the object to be segmented in any original training image comprises the following steps:
extracting image features of any original training image through a first convolutional layer to obtain a first training image feature corresponding to any original training image;
performing dimensionality reduction processing on the first training image characteristic corresponding to any original training image through a first downsampling layer to obtain a second training image characteristic;
and processing the characteristics of the second training image sequentially through each residual convolution layer and each residual down-sampling layer to obtain a prediction mask of the object to be segmented in any original training image.
5. An image segmentation system, comprising: the system comprises a processing module, a training module and an operation module;
the processing module is used for: labeling each original training image in the training image set to obtain all first training images;
the training module is configured to: training a preset deep learning model for image segmentation based on all original training images and all first training images to obtain a target deep learning model;
the operation module is used for: and inputting the image to be detected into the target deep learning model to obtain an image segmentation result of the image to be detected.
6. The image segmentation system of claim 5, wherein the processing module is specifically configured to:
judging whether a sheltered object exists in any original training image to obtain a judgment result of any original training image;
when the judgment result of any original training image is yes, completely marking each occluded object in any original training image, and marking each unoccluded object in any original training image to obtain a first training image corresponding to any original training image until a first training image corresponding to each original training image with the judgment result of yes is obtained;
and when the judgment result of any original training image is negative, labeling each object in any original training image to obtain a first training image corresponding to any original training image until the first training image corresponding to each original training image with the judgment result of negative is obtained.
7. The image segmentation system of claim 5, wherein the training module specifically comprises: a first training module and a second training module;
the first training module is to: inputting any original training image into the preset deep learning model to obtain a prediction mask of an object to be segmented in any original training image; based on a continuous coherent method, acquiring a predicted bar code corresponding to a predicted mask of any original training image, and acquiring a prior bar code corresponding to a first training image corresponding to any original training image; calculating a target distance value between the predicted barcode and the prior barcode, and taking the target distance value as a topological loss value of the preset deep learning model until a topological loss value corresponding to each original training image is obtained;
the second training module is to: and adding all topological loss values into a loss function of the preset deep learning model for back propagation to obtain and use a first deep learning model as the preset deep learning model, and returning and calling the first training module for iterative training until the preset deep learning model is converged to obtain the target deep learning model.
8. The image segmentation system of claim 7, wherein the pre-set deep learning model comprises: at least two convolution layers and at least two down-sampling layers, wherein each convolution layer is correspondingly connected with one down-sampling layer; the first training module is specifically configured to:
extracting image features of any original training image through a first convolutional layer to obtain a first training image feature corresponding to any original training image;
performing dimensionality reduction processing on the first training image characteristic corresponding to any original training image through a first downsampling layer to obtain a second training image characteristic;
and processing the characteristics of the second training image sequentially through each residual convolution layer and each residual down-sampling layer to obtain a prediction mask of the object to be segmented in any original training image.
9. A storage medium having stored therein instructions which, when read by a computer, cause the computer to execute the image segmentation method according to any one of claims 1 to 4.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor, when executing the computer program, causes the computer to perform the image segmentation method according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211281784.0A CN115578563A (en) | 2022-10-19 | 2022-10-19 | Image segmentation method, system, storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211281784.0A CN115578563A (en) | 2022-10-19 | 2022-10-19 | Image segmentation method, system, storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115578563A true CN115578563A (en) | 2023-01-06 |
Family
ID=84587179
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211281784.0A Pending CN115578563A (en) | 2022-10-19 | 2022-10-19 | Image segmentation method, system, storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115578563A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210004984A1 (en) * | 2018-09-25 | 2021-01-07 | Tsinghua University | Method and apparatus for training 6d pose estimation network based on deep learning iterative matching |
| CN113361533A (en) * | 2021-05-19 | 2021-09-07 | 山东师范大学 | Segmentation identification method and system for overlapped and shielded target object |
| CN114187491A (en) * | 2022-02-17 | 2022-03-15 | 中国科学院微电子研究所 | Method and device for detecting shielding object |
| CN114663655A (en) * | 2022-03-01 | 2022-06-24 | 新疆大学 | Image segmentation model training method, image semantic segmentation device and related equipment |
-
2022
- 2022-10-19 CN CN202211281784.0A patent/CN115578563A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210004984A1 (en) * | 2018-09-25 | 2021-01-07 | Tsinghua University | Method and apparatus for training 6d pose estimation network based on deep learning iterative matching |
| CN113361533A (en) * | 2021-05-19 | 2021-09-07 | 山东师范大学 | Segmentation identification method and system for overlapped and shielded target object |
| CN114187491A (en) * | 2022-02-17 | 2022-03-15 | 中国科学院微电子研究所 | Method and device for detecting shielding object |
| CN114663655A (en) * | 2022-03-01 | 2022-06-24 | 新疆大学 | Image segmentation model training method, image semantic segmentation device and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108710885B (en) | Target object detection method and device | |
| US11062210B2 (en) | Method and apparatus for training a neural network used for denoising | |
| CN112991278B (en) | Method and system for detecting Deepfake video by combining RGB (red, green and blue) space domain characteristics and LoG (LoG) time domain characteristics | |
| JP6773707B2 (en) | Signal processing equipment, signal processing methods and programs | |
| CN111340820B (en) | Image segmentation method and device, electronic equipment and storage medium | |
| CN111444807B (en) | Target detection method, device, electronic equipment and computer readable medium | |
| CN113436100B (en) | Method, apparatus, device, medium, and article for repairing video | |
| CN112749726B (en) | Training method and device for target detection model, computer equipment and storage medium | |
| JP2023531350A (en) | A method for incrementing a sample image, a method for training an image detection model and a method for image detection | |
| CN114898177B (en) | Defect image generation method, model training method, device, medium and product | |
| CN111767750A (en) | Image processing method and device | |
| CN111179276A (en) | Image processing method and device | |
| CN117649515A (en) | Digital twinning-based semi-supervised 3D target detection method, system and equipment | |
| CN115578563A (en) | Image segmentation method, system, storage medium and electronic equipment | |
| CN110852250A (en) | Vehicle weight removing method and device based on maximum area method and storage medium | |
| CN115861255A (en) | Model training method, device, equipment, medium and product for image processing | |
| CN116229535A (en) | Training method of face detection model, face detection method and device | |
| CN116363641A (en) | Image processing method and device and electronic equipment | |
| CN114022458A (en) | Skeleton detection method and device, electronic equipment and computer readable storage medium | |
| Azhar et al. | Texture-oriented image denoising technique for the removal of random-valued impulse noise | |
| CN112990046A (en) | Difference information acquisition method, related device and computer program product | |
| CN112487943A (en) | Method and device for removing duplicate of key frame and electronic equipment | |
| CN115511015B (en) | Sample screening method, device, equipment and computer readable storage medium | |
| CN116051935B (en) | Image detection method, training method and device of deep learning model | |
| CN115170618A (en) | Object tracking method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20230106 |