CN113762263A - Semantic segmentation method and system for small-scale similar structure - Google Patents
Semantic segmentation method and system for small-scale similar structure Download PDFInfo
- Publication number
- CN113762263A CN113762263A CN202110943961.6A CN202110943961A CN113762263A CN 113762263 A CN113762263 A CN 113762263A CN 202110943961 A CN202110943961 A CN 202110943961A CN 113762263 A CN113762263 A CN 113762263A
- Authority
- CN
- China
- Prior art keywords
- small
- segmentation
- scale similar
- segmented
- image
- 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
- 230000011218 segmentation Effects 0.000 title claims abstract description 117
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 41
- 238000010586 diagram Methods 0.000 claims description 16
- 238000005070 sampling Methods 0.000 claims description 16
- 238000001514 detection method Methods 0.000 claims description 10
- 238000007670 refining Methods 0.000 claims description 6
- 230000001537 neural effect Effects 0.000 claims description 5
- 238000000605 extraction Methods 0.000 abstract description 7
- 238000013528 artificial neural network Methods 0.000 abstract description 5
- 230000037182 bone density Effects 0.000 description 11
- 238000013135 deep learning Methods 0.000 description 3
- 238000001739 density measurement Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 239000000047 product Substances 0.000 description 3
- 210000000988 bone and bone Anatomy 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 239000011800 void material Substances 0.000 description 1
Images
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Abstract
The invention discloses a semantic segmentation method and a semantic segmentation system for small-scale similar structures, and relates to the technical field of neural networks. The method comprises the following steps: acquiring an image to be segmented; inputting the image to be segmented into a convolutional neural network to obtain a segmentation result of the small-scale similar structure and key points; and processing the segmentation result according to the key point. The method is suitable for semantic segmentation of the small-scale similar structure, the network structure is designed according to the characteristics of the small-scale similar structure image, region segmentation and key point extraction are carried out on the small-scale similar structure, image information and key point information are fully utilized, and therefore segmentation accuracy is improved.
Description
Technical Field
The invention relates to the technical field of neural networks, in particular to a semantic segmentation method and a semantic segmentation system for small-scale similar structures.
Background
The deep learning network is applied more and more in various fields, and at present, the purposes of image detection, parameter prediction, image recognition and the like can be realized by using the deep learning network, and the purposes are realized based on semantic segmentation.
However, for a small-scale structure with a similar structure, the existing semantic segmentation method does not consider the structural characteristics thereof, the segmentation accuracy is not high, the region segmentation of a non-target object is often easily performed, and misjudgment is easily caused, which affects the subsequent function implementation.
Disclosure of Invention
The invention aims to solve the technical problem of providing a semantic segmentation method and a semantic segmentation system for small-scale similar structures aiming at the defects of the prior art.
The technical scheme for solving the technical problems is as follows:
a semantic segmentation method for small-scale similar structures, comprising:
acquiring an image to be segmented;
inputting the image to be segmented into a convolutional neural network to obtain a segmentation result of a small-scale similar structure and key points;
and processing the segmentation result according to the key point.
Another technical solution of the present invention for solving the above technical problems is as follows:
a semantic segmentation system for small-scale similar structures, comprising: collection equipment, processing apparatus and display device, wherein:
the acquisition equipment is used for acquiring an image to be segmented;
the processing equipment is used for inputting the image to be segmented into a convolutional neural network to obtain a segmentation result of a small-scale similar structure and key points; processing the segmentation result according to the key point;
the display device is used for displaying the segmentation result of the image to be segmented.
The invention has the beneficial effects that: the method and the system provided by the invention are suitable for semantic segmentation of small-scale similar structures, and the small-scale similar structures are subjected to region segmentation and key point extraction by designing the feature extraction network structure according to the characteristics of small-scale similar structure images, so that the image information and the key point information are fully utilized, and the segmentation precision is improved.
Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
FIG. 1 is a schematic flow chart of a process according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a spine segmentation result provided by an embodiment of the method of the present invention, wherein a is a schematic diagram of a key point, and b is a schematic diagram of a segmentation region;
FIG. 3 is a schematic diagram of a convolutional neural network structure provided by another embodiment of the method of the present invention;
FIG. 4 is a schematic structural flow chart of a convolutional neural network for measuring bone density according to another embodiment of the method of the present invention;
fig. 5 is a schematic structural framework diagram provided by an embodiment of the system of the present invention.
Detailed Description
The principles and features of this invention are described below in conjunction with the following drawings, which are set forth to illustrate, but are not to be construed to limit the scope of the invention.
As shown in fig. 1, a schematic flow chart provided for an embodiment of the method of the present invention is used to improve the semantic segmentation accuracy of a small-scale similar structure, and includes:
s1 acquires an image to be segmented.
It should be noted that the image to be segmented may be image data containing small-scale similar structures. The small-scale similar structure refers to an object structure with a smaller size and a more similar structure, for example, human spines, petals, densely arranged architectural decorations and the like can be used. The objects have common characteristics that the objects are usually small in size and similar in shape, and are easily interfered when being segmented by using a convolutional neural network, and the existing convolutional neural network is not optimized according to the characteristics of the structures, so that the semantic segmentation accuracy is low.
And S2, inputting the image to be segmented into a convolutional neural network to obtain a segmentation result of the small-scale similar structure and key points.
It should be noted that the key point may be a central point of each small-scale similar structure, and since the small-scale similar structure is relatively fixed in shape compared to other objects, the central point may be labeled in advance, which is convenient for machine learning and identification.
The structure of the convolutional neural network can be set according to actual requirements, the input of the convolutional neural network can be an image to be segmented, and the output of the convolutional neural network can be a feature map of a segmentation region of each small-scale similar structure and a key point feature map of each small-scale similar structure.
And S3, processing the segmentation result according to the key point.
After obtaining the feature map of each small-scale similar structure segmentation region and the feature map of each small-scale similar structure keypoint, the segmentation region containing the keypoint can be taken as a final segmentation result, so that the segmentation result of a non-target object can be eliminated.
The range of the segmentation region can be re-determined according to the key point, for example, assuming that semantic segmentation is performed on the spine, and the spine and the adjacent bone tissues are segmented together in the obtained segmentation result, the position of the key point in the range of the segmentation region is not near the central point, so that the range of the segmentation region can be re-determined according to the key point, the key point is in the preset range of the central point of the spine, and the accuracy of region segmentation is improved.
The following description will take semantic segmentation of the spine as an example.
For example, as shown in b of fig. 2, when an image is segmented, an erroneous determination may occur or a non-spinal bone tissue is segmented, and an obtained segmentation map may include a non-spinal region, then segmented regions may be screened by key points, and as can be seen from a of fig. 2, although a segmented region is not present, a segmented region including key points is used as a final segmentation result by combining the key points with the segmented regions, thereby obtaining a spinal region segmentation result with higher accuracy.
That is, after the segmentation result and the key points of each vertebra are obtained, the segmentation region including the key points is used as the final segmentation result, and the segmentation accuracy can be improved.
After obtaining a more accurate spine segmentation map, the following application can be made based on the spine segmentation map: for example, the spine segmentation map is used to detect the bone density of the spine, and the spine segmentation map may be processed using a convolutional neural network to obtain the bone density of each spine.
The structure of the convolutional neural network for detecting bone density can be set according to actual requirements, and can be realized by the existing neural network, for example, a CNN network can be selected.
An exemplary convolutional neural network structure for detecting bone density is provided as shown in fig. 4, and the convolutional neural network structure for detecting bone density is described below with reference to fig. 4.
The convolutional neural network includes: the number of convolution kernels of the 3 layers 3 x 3 of the convolution layer C7 is 512, 256 and 128 respectively, the convolution kernels are used for extracting features, the 2 layers of the full-connected layer F and the final layer O are provided, the number of units of the final layer O is the number of corresponding vertebral blocks, and the bone density of the corresponding vertebral blocks is measured respectively.
T11, T12, L1 and L2 are different vertebrae, respectively.
The workflow of the convolutional neural network for detecting bone density is explained below.
Inputting a vertebra image containing a segmentation region into a convolutional neural network, extracting features of the image to be segmented through a convolutional layer, classifying the extracted features through a full-link layer, and outputting a bone density measured value of each vertebra through a final layer.
It should be understood that when the deep learning method is used for realizing the automatic diagnosis of the bone density measurement, the designated vertebral body can be firstly segmented, regression measurement is carried out according to the segmented vertebral body, and the segmentation and positioning of the spine are of great importance, so that the spine is segmented by combining key points, the segmentation precision is improved, and the bone density measurement precision is further improved.
The following will further describe the semantic segmentation of chips on a certain production line.
After the production of the chip is finished, the chip needs to be attached, the detection of the attachment position is important, the yield of the chip is related, the chip on the production line has the characteristics of small size and similar structure, and the chip can be considered as a small-scale similar structure, so that the image containing the chip can be shot, the image is input into a convolutional neural network, the characteristic extraction is carried out on the chip, the result characteristic diagram of the segmentation area of each chip and the key point characteristic diagram of each chip are obtained, the segmentation area containing the key points is used as the final segmentation result, the semantic segmentation precision of the chip is improved, and the subsequent detection and processing are facilitated.
The method provided by the embodiment is suitable for semantic segmentation of the small-scale similar structure, the network structure is designed according to the characteristics of the small-scale similar structure image, region segmentation and key point extraction are carried out on the small-scale similar structure, image information and key point information are fully utilized, and therefore segmentation precision is improved.
Optionally, in some possible embodiments, the step of inputting the image to be segmented into the convolutional neural network to obtain the segmentation result of the small-scale similar structure and the keypoint specifically includes:
inputting an image to be segmented into a convolutional neural network, and extracting the characteristics of the image to be segmented through a convolutional layer;
dividing the extracted features into two paths, wherein one path is subjected to downsampling processing through a downsampling layer to obtain shallow semantic features, and the other path is subjected to cavity convolution processing of different scales through a convolution layer to obtain deep semantic features;
combining the deep semantic features with the shallow semantic features after up-sampling processing to obtain combined features;
and refining the combination characteristics through the convolution layer, and performing up-sampling treatment on the refined combination characteristics through the up-sampling layer to obtain a characteristic diagram containing a small-scale similar structure segmentation region and a characteristic diagram containing key points.
An exemplary schematic diagram of a convolutional neural network structure for extracting segmentation results and key points of a small-scale similar structure is provided as shown in fig. 3, and the convolutional neural network structure is described below with reference to fig. 3.
The convolutional neural network includes: and the 5 layers of 3-by-3 convolution layers C1 are used for extracting the features of the image to be segmented, and the obtained features are divided into two paths.
And for the first path, the convolution layer C2 with 1 x 1 is used for refining the features, and the down-sampling layer D is used for down-sampling 4 times of the refined features as shallow semantic information and retaining texture information of a small-scale similar structure.
For the second path, 2 convolutional layers C3 and C4 of 1 × 1 and 4 convolutional layers C5 of different scales are included to extract deep semantic features, where rates of 4 convolutional layers C5 of different scales are 6, 12, and 18 and global, and information of small-scale similar structures is not excessively lost while the receptive field is expanded. And the up-sampling layer U1 is used for up-sampling the features extracted by the multi-scale void convolution by 4 times.
Further comprising: and 3 × 3 convolutional layer C6 for combining the deep semantic features with the shallow semantic features and then refining.
Further comprising: and the upsampling layer U2 is used for upsampling the thinned combined features by 4 times to obtain a feature map with the same size as the original image, and the feature map is used as the output feature of the final small-scale similar structure image.
It should be noted that, the combination of the deep semantic features and the shallow semantic features refers to: the deep semantic features and the shallow semantic features have a plurality of feature maps, and the feature maps can be used as input of the next layer together.
Optionally, in some possible embodiments, before the step of inputting the image to be segmented into the convolutional neural network to obtain the segmentation result of the small-scale similar structure and the keypoints, the method further includes:
and performing multi-task training on the convolutional neural network, and simultaneously performing region segmentation and key point detection training.
The region segmentation and the key point detection are trained simultaneously, although the final purpose is to segment a target region, each key point with a small-scale similar structure is valuable, and the branch of the key point and the segmented branch supplement each other, so that the model convergence is accelerated, and the segmentation accuracy is improved.
Optionally, in some possible embodiments, the step of processing the segmentation result according to the key point specifically includes:
and comparing the feature maps of the segmented regions with the feature maps of the key points, and taking the segmented regions containing the key points as final segmentation results.
By selecting the segmented regions including the key points as the final segmentation result, interference can be eliminated, and the segmented regions without key points are not reserved, thereby improving the accuracy of region segmentation.
Optionally, in some possible embodiments, before the step of inputting the image to be segmented into the convolutional neural network to obtain the segmentation result of the small-scale similar structure and the keypoints, the method further includes:
and detecting small-scale similar structures in the image to be segmented.
By detecting the image to be segmented in advance, the approximate position of the small-scale similar structure can be determined in advance, so that interference can be eliminated conveniently, and the segmentation speed and accuracy are improved.
For example, the detection can be performed by an existing image detection algorithm, which is not described herein.
It is to be understood that some or all of the various embodiments described above may be included in some embodiments.
Referring to FIG. 5, a schematic structural framework is provided for an embodiment of the system of the present invention for improving spinal density measurement accuracy, comprising: acquisition device 10, processing device 20 and display device 30, wherein:
the acquisition equipment 10 is used for acquiring an image to be segmented;
the processing device 20 is configured to input the image to be segmented into the convolutional neural network to obtain a segmentation result of the small-scale similar structure and a key point; processing the segmentation result according to the key point;
the display device 30 is used for displaying the segmentation result of the image to be segmented.
The system provided by the embodiment is suitable for semantic segmentation of small-scale similar structures, the feature extraction network structure is designed according to the characteristics of small-scale similar structure images, region segmentation and key point extraction are carried out on the small-scale similar structures, image information and key point information are fully utilized, and therefore segmentation precision is improved.
Optionally, in some possible embodiments, the processing device 20 comprises: the device comprises a processor and a neural chip, wherein a convolution neural network is arranged on the neural chip, and the convolution neural network comprises:
the processor is used for inputting the image to be segmented into the convolutional neural network;
the neural chip is used for extracting the features of the image to be segmented through the convolution layer; dividing the extracted features into two paths, wherein one path is subjected to downsampling processing through a downsampling layer to obtain shallow semantic features, and the other path is subjected to cavity convolution processing of different scales through a convolution layer to obtain deep semantic features; combining the deep semantic features with the shallow semantic features after up-sampling processing to obtain combined features; and refining the combination characteristics through the convolution layer, and performing up-sampling treatment on the refined combination characteristics through the up-sampling layer to obtain a characteristic diagram containing a small-scale similar structure segmentation region and a characteristic diagram containing key points.
Optionally, in some possible embodiments, the processing device 20 further comprises: and the training system is used for carrying out multi-task training on the convolutional neural network and simultaneously carrying out training on region segmentation and key point detection.
Optionally, in some possible embodiments, the processing device 20 further comprises: and the first graphic processor is used for comparing the feature map of the divided region with the feature map of the key point and taking the divided region containing the key point as a final divided result.
Optionally, in some possible embodiments, the processing device 20 further comprises: and the second graphic processor is used for detecting the small-scale similar structure in the image to be segmented.
It is to be understood that some or all of the various embodiments described above may be included in some embodiments.
It should be noted that the above embodiments are product embodiments corresponding to previous method embodiments, and for the description of the product embodiments, reference may be made to corresponding descriptions in the above method embodiments, and details are not repeated here.
The reader should understand that in the description of this specification, reference to the description of the terms "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described method embodiments are merely illustrative, and for example, the division of steps into only one logical functional division may be implemented in practice in another way, for example, multiple steps may be combined or integrated into another step, or some features may be omitted, or not implemented.
The above method, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention essentially or partially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
While the invention has been described with reference to specific embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A semantic segmentation method for small-scale similar structures, comprising:
acquiring an image to be segmented;
inputting the image to be segmented into a convolutional neural network to obtain a segmentation result of a small-scale similar structure and key points;
and processing the segmentation result according to the key point.
2. The semantic segmentation method for the small-scale similar structure according to claim 1, wherein the step of inputting the image to be segmented into a convolutional neural network to obtain a segmentation result and key points of the small-scale similar structure specifically comprises:
inputting the image to be segmented into a convolutional neural network, and extracting the characteristics of the image to be segmented through a convolutional layer;
dividing the extracted features into two paths, wherein one path is subjected to downsampling processing through a downsampling layer to obtain shallow semantic features, and the other path is subjected to cavity convolution processing of different scales through a convolution layer to obtain deep semantic features;
the deep semantic features are combined with the shallow semantic features after being subjected to up-sampling processing to obtain combined features;
and refining the combination characteristics through a convolutional layer, and performing up-sampling treatment on the refined combination characteristics through an up-sampling layer to obtain a characteristic diagram containing a small-scale similar structure segmentation region and a characteristic diagram containing key points.
3. The semantic segmentation method for the small-scale similar structure according to claim 2, wherein before the step of inputting the image to be segmented into a convolutional neural network to obtain the segmentation result and the key points of the small-scale similar structure, the method further comprises:
and performing multi-task training on the convolutional neural network, and simultaneously performing region segmentation and key point detection training.
4. The semantic segmentation method for small-scale similar structures according to claim 2, wherein the step of processing the segmentation result according to the key point specifically comprises:
and comparing the feature map of the segmentation region with the feature map of the key point, and taking the segmentation region containing the key point as a final segmentation result.
5. The semantic segmentation method for the small-scale similar structure according to any one of claims 1 to 4, wherein before the step of inputting the image to be segmented into a convolutional neural network to obtain the segmentation result and the keypoints of the small-scale similar structure, the method further comprises:
and detecting the small-scale similar structure in the image to be segmented.
6. A semantic segmentation system for small-scale similar structures, comprising: collection equipment, processing apparatus and display device, wherein:
the acquisition equipment is used for acquiring an image to be segmented;
the processing equipment is used for inputting the image to be segmented into a convolutional neural network to obtain a segmentation result of a small-scale similar structure and key points; processing the segmentation result according to the key point;
the display device is used for displaying the segmentation result of the image to be segmented.
7. The semantic segmentation system for small-scale similar structures according to claim 6, wherein the processing device comprises: a processor and a neural chip having a convolutional neural network disposed thereon, wherein:
the processor is used for inputting the image to be segmented into a convolutional neural network;
the neural chip is used for extracting the features of the image to be segmented through the convolutional layer; dividing the extracted features into two paths, wherein one path is subjected to downsampling processing through a downsampling layer to obtain shallow semantic features, and the other path is subjected to cavity convolution processing of different scales through a convolution layer to obtain deep semantic features; the deep semantic features are combined with the shallow semantic features after being subjected to up-sampling processing to obtain combined features; and refining the combination characteristics through a convolutional layer, and performing up-sampling treatment on the refined combination characteristics through an up-sampling layer to obtain a characteristic diagram containing a small-scale similar structure segmentation region and a characteristic diagram containing key points.
8. The semantic segmentation system for small-scale similar structures according to claim 7, wherein the processing device further comprises: and the training system is used for carrying out multi-task training on the convolutional neural network and simultaneously carrying out training on region segmentation and key point detection.
9. The semantic segmentation system for small-scale similar structures according to claim 6, wherein the processing device further comprises: and the first graphic processor is used for comparing the feature map of the segmentation region with the feature map of the key point and taking the segmentation region containing the key point as a final segmentation result.
10. The semantic segmentation system for small-scale similar structures according to any of claims 6 to 9, characterized in that the processing device further comprises: and the second graphic processor is used for detecting the small-scale similar structure in the image to be segmented.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110943961.6A CN113762263A (en) | 2021-08-17 | 2021-08-17 | Semantic segmentation method and system for small-scale similar structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110943961.6A CN113762263A (en) | 2021-08-17 | 2021-08-17 | Semantic segmentation method and system for small-scale similar structure |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113762263A true CN113762263A (en) | 2021-12-07 |
Family
ID=78790163
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110943961.6A Pending CN113762263A (en) | 2021-08-17 | 2021-08-17 | Semantic segmentation method and system for small-scale similar structure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113762263A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116030016A (en) * | 2023-01-10 | 2023-04-28 | 广州市易鸿智能装备有限公司 | Product image defect detection method and device, electronic equipment and storage medium |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086770A (en) * | 2018-07-25 | 2018-12-25 | 成都快眼科技有限公司 | A kind of image, semantic dividing method and model based on accurate scale prediction |
| CN110232394A (en) * | 2018-03-06 | 2019-09-13 | 华南理工大学 | A kind of multi-scale image semantic segmentation method |
| CN110889449A (en) * | 2019-11-27 | 2020-03-17 | 中国人民解放军国防科技大学 | Edge-enhanced multi-scale remote sensing image building semantic feature extraction method |
| CN111210435A (en) * | 2019-12-24 | 2020-05-29 | 重庆邮电大学 | Image semantic segmentation method based on local and global feature enhancement module |
| US20200202533A1 (en) * | 2018-12-24 | 2020-06-25 | Adobe Inc. | Identifying target objects using scale-diverse segmentation neural networks |
| CN111681273A (en) * | 2020-06-10 | 2020-09-18 | 创新奇智(青岛)科技有限公司 | Image segmentation method and device, electronic equipment and readable storage medium |
| CN112396607A (en) * | 2020-11-18 | 2021-02-23 | 北京工商大学 | Streetscape image semantic segmentation method for deformable convolution fusion enhancement |
| US20210097692A1 (en) * | 2019-10-01 | 2021-04-01 | Volvo Car Corporation | Data filtering of image stacks and video streams |
| CN112733919A (en) * | 2020-12-31 | 2021-04-30 | 山东师范大学 | Image semantic segmentation method and system based on void convolution and multi-scale and multi-branch |
| CN112861911A (en) * | 2021-01-10 | 2021-05-28 | 西北工业大学 | RGB-D semantic segmentation method based on depth feature selection fusion |
| CN113052180A (en) * | 2021-03-23 | 2021-06-29 | 中国人民解放军战略支援部队信息工程大学 | Encoding and decoding network port image segmentation method fusing semantic flow fields |
| CN113139502A (en) * | 2021-05-11 | 2021-07-20 | 大连理工大学 | Unsupervised video segmentation method |
| CN113191367A (en) * | 2021-05-25 | 2021-07-30 | 华东师范大学 | Semantic segmentation method based on dense scale dynamic network |
| US20210319265A1 (en) * | 2020-11-02 | 2021-10-14 | Zhengzhou University | Method for segmentation of underground drainage pipeline defects based on full convolutional neural network |
-
2021
- 2021-08-17 CN CN202110943961.6A patent/CN113762263A/en active Pending
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110232394A (en) * | 2018-03-06 | 2019-09-13 | 华南理工大学 | A kind of multi-scale image semantic segmentation method |
| CN109086770A (en) * | 2018-07-25 | 2018-12-25 | 成都快眼科技有限公司 | A kind of image, semantic dividing method and model based on accurate scale prediction |
| US20200202533A1 (en) * | 2018-12-24 | 2020-06-25 | Adobe Inc. | Identifying target objects using scale-diverse segmentation neural networks |
| US20210097692A1 (en) * | 2019-10-01 | 2021-04-01 | Volvo Car Corporation | Data filtering of image stacks and video streams |
| CN110889449A (en) * | 2019-11-27 | 2020-03-17 | 中国人民解放军国防科技大学 | Edge-enhanced multi-scale remote sensing image building semantic feature extraction method |
| CN111210435A (en) * | 2019-12-24 | 2020-05-29 | 重庆邮电大学 | Image semantic segmentation method based on local and global feature enhancement module |
| CN111681273A (en) * | 2020-06-10 | 2020-09-18 | 创新奇智(青岛)科技有限公司 | Image segmentation method and device, electronic equipment and readable storage medium |
| US20210319265A1 (en) * | 2020-11-02 | 2021-10-14 | Zhengzhou University | Method for segmentation of underground drainage pipeline defects based on full convolutional neural network |
| CN112396607A (en) * | 2020-11-18 | 2021-02-23 | 北京工商大学 | Streetscape image semantic segmentation method for deformable convolution fusion enhancement |
| CN112733919A (en) * | 2020-12-31 | 2021-04-30 | 山东师范大学 | Image semantic segmentation method and system based on void convolution and multi-scale and multi-branch |
| CN112861911A (en) * | 2021-01-10 | 2021-05-28 | 西北工业大学 | RGB-D semantic segmentation method based on depth feature selection fusion |
| CN113052180A (en) * | 2021-03-23 | 2021-06-29 | 中国人民解放军战略支援部队信息工程大学 | Encoding and decoding network port image segmentation method fusing semantic flow fields |
| CN113139502A (en) * | 2021-05-11 | 2021-07-20 | 大连理工大学 | Unsupervised video segmentation method |
| CN113191367A (en) * | 2021-05-25 | 2021-07-30 | 华东师范大学 | Semantic segmentation method based on dense scale dynamic network |
Non-Patent Citations (3)
| Title |
|---|
| ZHANG, XIUWEI，等: "ICENETv2: A Fine-Grained River Ice Semantic Segmentation Network Based on UAV Images", 《REMOTE SENSING》, vol. 13, no. 4, pages 633 * |
| 张云: "结合上下文特征与CNN多层特征融合的语义分割", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 1, pages 138 - 1401 * |
| 李鑫: "基于深度学习的图像语义分割方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 1, pages 138 - 1340 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116030016A (en) * | 2023-01-10 | 2023-04-28 | 广州市易鸿智能装备有限公司 | Product image defect detection method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106056595B (en) | Based on the pernicious assistant diagnosis system of depth convolutional neural networks automatic identification Benign Thyroid Nodules | |
| CN109493308A (en) | The medical image synthesis and classification method for generating confrontation network are differentiated based on condition more | |
| CN112184617B (en) | Spine MRI image key point detection method based on deep learning | |
| CN107993221B (en) | Automatic identification method for vulnerable plaque of cardiovascular Optical Coherence Tomography (OCT) image | |
| CN108257135A (en) | The assistant diagnosis system of medical image features is understood based on deep learning method | |
| CN110766051A (en) | Lung nodule morphological classification method based on neural network | |
| CN110245657B (en) | Pathological image similarity detection method and detection device | |
| CN112150428A (en) | Medical image segmentation method based on deep learning | |
| CN113743463B (en) | Tumor benign and malignant recognition method and system based on image data and deep learning | |
| CN114693933A (en) | Medical image segmentation device based on generation of confrontation network and multi-scale feature fusion | |
| CN112215217B (en) | Digital image recognition method and device for simulating doctor to read film | |
| Fashandi et al. | An investigation of the effect of fat suppression and dimensionality on the accuracy of breast MRI segmentation using U‐nets | |
| CN115546605A (en) | Training method and device based on image labeling and segmentation model | |
| CN114332572B (en) | Method for extracting breast lesion ultrasonic image multi-scale fusion characteristic parameters based on saliency map-guided hierarchical dense characteristic fusion network | |
| CN113762263A (en) | Semantic segmentation method and system for small-scale similar structure | |
| CN114170473A (en) | Method and system for classifying dMMR subtypes based on pathological images | |
| CN112861881A (en) | Honeycomb lung recognition method based on improved MobileNet model | |
| WO2021183765A1 (en) | Automated detection of tumors based on image processing | |
| Mamun et al. | Recognition of human skin diseases using inception-v3 with transfer learning | |
| CN113379770A (en) | Nasopharyngeal carcinoma MR image segmentation network construction method, image segmentation method and device | |
| Kumar et al. | Performance analysis of machine learning-based breast cancer detection algorithms | |
| CN112734707A (en) | Auxiliary detection method, system and device for 3D endoscope and storage medium | |
| Srilakshmi et al. | Performance enhancement of content based medical image retrieval for MRI brain images based on hybrid approach | |
| Ambily et al. | Classification of Brain MRI Images Using Convolution Neural Network and Transfer Learning | |
| Arefin et al. | Deep learning approach for detecting and localizing brain tumor from magnetic resonance imaging images |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 100192 A206, 2 / F, building B-2, Dongsheng Science Park, Zhongguancun, 66 xixiaokou Road, Haidian District, Beijing Applicant after: Huiying medical technology (Beijing) Co.,Ltd. Address before: 100192 A206, 2 / F, building B-2, Dongsheng Science Park, Zhongguancun, 66 xixiaokou Road, Haidian District, Beijing Applicant before: HUIYING MEDICAL TECHNOLOGY (BEIJING) Co.,Ltd. |