CN109190626A - A kind of semantic segmentation method of the multipath Fusion Features based on deep learning - Google Patents
A kind of semantic segmentation method of the multipath Fusion Features based on deep learning Download PDFInfo
- Publication number
- CN109190626A CN109190626A CN201810842201.4A CN201810842201A CN109190626A CN 109190626 A CN109190626 A CN 109190626A CN 201810842201 A CN201810842201 A CN 201810842201A CN 109190626 A CN109190626 A CN 109190626A
- Authority
- CN
- China
- Prior art keywords
- network
- feature
- image
- multipath
- deep learning
- 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
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
- G06V10/267—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 by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06V10/32—Normalisation of the pattern dimensions
Abstract
The semantic segmentation method of the present invention relates to a kind of multipath Fusion Features based on deep learning, comprising the following steps: multipath Feature fusion is used to extract the depth of foundation feature of image；The depth of foundation feature of extraction is passed through into decoding end network, restores original image resolution information, and generate segmentation result；Using cross entropy loss function as target training network, network performance is evaluated using accuracy rate and mIoU.The present invention has rational design, it has fully considered local message and global information, feature extraction end and classification end in a network is added to many paths, the output of network is and original image resolution segmentation figure of the same size, segmentation accuracy rate is calculated using the existing label of image, network is trained to minimize cross entropy loss function as target, effectively improves image, semantic dividedly accuracy rate.
Description
Technical field
The invention belongs to computer visual image semantic segmentation technical field, especially a kind of multichannel based on deep learning
The semantic segmentation method of diameter Fusion Features.
Background technique
Image, semantic segmentation, which refers to, is divided into different semantic classes for each pixel in image by certain method, real
The now reasoning process from bottom to high-level semantic finally obtains the segmentation for showing the semantic tagger pixel-by-pixel of different cut zone
Figure.The streetscape identification and target detection, the detection of unmanned plane pick-up point, field that image, semantic segmentation is widely used in automatic driving
Application in terms of many Computer Vision Tasks such as scape understanding, robot vision.From machine learning side based on computer vision
Method is to the method currently based on deep learning, and the research of image, semantic partitioning algorithm has obtained very big progress, still, due to work
Industry demand continues to increase, and image, semantic divides one of the research hotspot being still in Computer Vision Task.
The image, semantic segmentation of early stage is using mark feature by hand, such as histograms of oriented gradients HOG and scale invariant feature
Convert SIFT.Based on the method for machine learning from simplest pixel scale threshold method, based on the dividing method of pixel cluster to
The dividing method divided based on graph theory.These methods excessively rely on the feature database marked by hand, it is difficult to characteristics of image is indicated extensively,
There is significant limitation in practical applications.In recent years, the development of convolutional neural networks (CNN), makes in Computer Vision Task
Many problems obtained huge breakthrough.Since depth convolutional network can extract the spy of image from great amount of samples data
Sign, it is more preferable than mark feature by hand, it is obtained in image classification and target detection contour level Computer Vision Task huge
Success.Appoint in addition, also all introducing image, semantic segmentation in the contests such as Pascal VOC, Cityscapes, Microsoft COCO
Business, image is complicated, object classification is more, difficulty is big, has attracted the strong interest of a large amount of researchers at home and abroad, and emerges a large amount of
Classical effective method.
Although image, semantic dividing method has had good development at present, because its complexity, still has very
More problems have to be solved.The challenge of image, semantic segmentation is mainly reflected in: the uncertainty of object level, this is because image
Light levels, fog-level, the influence of the factors such as size, direction of objects in images；Object category level obscures
Property, the same species it is different classes of, it is also bad to open respectively.In order to reduce the shadow of the extraneous factors such as uncertain and ambiguity
It rings, (such as image pixel value can therefrom propose many features, such as color characteristic, picture in image to the information for making full use of in image
The contextual informations such as the connection between element, object), better character representation is obtained, is an important method.In order to improve figure
As the accuracy of semantic segmentation, people, which also have been working hard, uses advanced algorithm.Therefore, image, semantic segmentation property how is improved
It can be problem in the urgent need to address at present.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art, the base that a kind of design is reasonable and recognition accuracy is high is proposed
In the semantic segmentation method of the multipath Fusion Features of deep learning.
The present invention solves its technical problem and adopts the following technical solutions to achieve:
A kind of semantic segmentation method of the multipath Fusion Features based on deep learning, comprising the following steps:
Step 1, the depth of foundation feature that image is extracted using multipath Feature fusion；
The depth of foundation feature of extraction is passed through decoding end network by step 2, restores original image resolution information, and raw
At segmentation result；
Step 3 trains network by target of cross entropy loss function, and accuracy rate and mIoU is used to evaluate network performance.
The depth of foundation feature includes global information and local message and is merged by the output of different convolutional layers
It arrives.
The concrete methods of realizing of the step 1 comprises the steps of:
(1) input picture is zoomed into uniform sizes 256 × 256, using VGG16 convolutional neural networks framework as pre-training
Basic convolutional neural networks, according to output feature resolution size network is divided into 4 modules, each module is original
Two paths are added on the basis of network, the quantity of convolutional layer is 1 and 2 respectively；
(2) the image after scaling is input in modified VGG16 network structure, image passes through a series of convolution, batch
After normalization, Chi Hua, ReLU operation, each module exports a kind of feature, and resolution ratio is followed successively by 64 × 64,32 × 32,16 ×
16 and 8 × 8 local feature.
The concrete methods of realizing of the step 2 comprises the steps of:
(1) 8 × 8 features are up-sampled into layer by convolution sum, 16 × 16 characteristic pattern is obtained, by this feature and coding side 16
× 16 convolutional layer cascade, obtains more 16 × 16 characteristic patterns；
(2) 16 × 16 features are also passed through into the processing in (1), obtain 32 × 32 features；
(3) successively execute, until obtaining 128 × 128 characteristic patterns；
(4) 128 × 128 characteristic patterns are generated into 256 × 256 segmentation figure by up-sampling and process of convolution.
The concrete methods of realizing of the step 3 comprises the steps of:
(1) calculate prediction segmentation figure and the segmentation figure that has marked intersects entropy loss, utilizes back-propagation algorithm update power
Weight.
(2) after the completion of network training, its estimated performance is measured using accuracy rate and mIoU.
The advantages and positive effects of the present invention are:
The present invention has rational design, has fully considered local message and global information, feature extraction end in a network
Be added to many paths with classification end, the output of network be with original image resolution segmentation figure of the same size, use image
Existing label calculates segmentation accuracy rate, trains network to minimize cross entropy loss function as target, effectively improves
Image, semantic dividedly accuracy rate.
Detailed description of the invention
Fig. 1 is overall network frame diagram proposed by the present invention；
Fig. 2 is the multipath module working principle diagram (applying in coding side) in inventive network structure；
Fig. 3 is the up-sampling module working principle diagram (applying in decoding end) in inventive network structure；
Fig. 4 is test result of the present invention on public data collection CamVid.
Specific embodiment
The embodiment of the present invention is further described below in conjunction with attached drawing.
The present invention proposes aiming at the problem that how to make full use of global information and local message in image, semantic segmentation
It is a kind of to carry out semantic segmentation using multipath Fusion Features network.As shown in Figures 1 to 3, the present invention changes network structure, in feature
End, i.e. coding side (Encoder) are extracted, a paths of each of network structure convolutional layer are become into mulitpath,
Each layer of convolution output end, the feature that mulitpath is extracted are added, are input in next layer network.At classification end, that is, solve
Code end, feature restore image original resolution size, keep classification results more credible by convolutional layer and up-sampling layer.It is this
Method is equivalent to feature extraction end in a network and end of classifying is added to many paths, and different paths can make in same layer
The receptive field of convolution kernel is also different, and the dimensional information that the feature that such each path obtains includes is different, has finally obtained a system
It arranges from part to global feature.Such fusion results have fully considered local message and global information.The output of network
Be with original image resolution segmentation figure of the same size, segmentation accuracy rate is calculated using the existing label of image, finally with minimum
Changing cross entropy loss function is target to train network.
In the present embodiment, a kind of semantic segmentation method of the multipath Fusion Features based on deep learning includes following step
It is rapid:
Step S1, in coding side, the depth of foundation feature of image is extracted using the multipath Feature fusion of proposition, this
A little features are obtained by the output fusion of different convolutional layers, therefore contain global information and local message.The tool of this step
Body processing method is as follows:
Input picture is zoomed to uniform sizes 256 × 256 by step S1.1, using VGG16 convolutional neural networks framework as
Then network is divided into 4 modules according to the size of output feature resolution by the basic convolutional neural networks of pre-training, each
Module adds two paths on the basis of primitive network, and the quantity of convolutional layer is 1 and 2 respectively；
Image after scaling is input in modified VGG16 network structure by step S1.2, and image passes through a series of volumes
After product, batch normalization, Chi Hua, ReLU are operated, each module exports a kind of feature, and resolution ratio is followed successively by 64 × 64,32 ×
32,16 × 16,8 × 8 local feature；
Step S2, the feature of extraction is passed through into decoding end network, restores original image resolution information, and generate segmentation knot
Fruit.The specific processing method of this step is as follows:
8 × 8 features are up-sampled layer by convolution sum by step S2.1, obtain 16 × 16 characteristic pattern, by this feature and are compiled
The convolutional layer cascade at code end 16 × 16, obtains more 16 × 16 characteristic patterns.
16 × 16 features are also passed through the processing in (1) by step S2.2, obtain 32 × 32 features.
Step S2.3, it successively executes, until obtaining 128 × 128 characteristic patterns.
Step S2.4,128 × 128 characteristic pattern is generated into 256 × 256 segmentation figure by up-sampling and convolutional layer.
Step S3, using cross entropy loss function as target training network, network performance is evaluated using accuracy rate and mIoU.This
The specific processing method of step is as follows:
Step S3.1, calculate prediction segmentation figure and the segmentation figure that has marked intersects entropy loss, is calculated using backpropagation
Method updates weight.
Step S3.2, after the completion of network training, accuracy rate and mIoU (Mean Intersection over are used
Union is handed over and is compared) measure its estimated performance.
It is tested below as method of the invention, illustrates recognition effect of the invention.
Test environment: python2.7；PyTorch frame；Ubuntu16.04 system；NVIDIA GTX 1070p GPU
Cycle tests: selected data collection is the image data set CamVid and CityScapes for image segmentation.Wherein
CamVid data set includes 701 images, and CityScapes data set includes 5000 images.
Test index: the present invention is Performance Evaluating Indexes using accuracy rate (Global Accuracy) and mIoU.Accuracy rate
Refer to pixel classifications accuracy rate.MIoU refers to the ratio between intersection and union of the correct erroneous pixel of consensus forecast.Not to current prevalence
These achievement datas are calculated with algorithm and then carry out Comparative result, it was demonstrated that the present invention obtains preferably in image, semantic segmentation field
As a result.
Test result is as follows:
Performance comparison of 1. present invention of table under different path conditions, multipath fusion can promote network known to
Performance
2. present invention of table is compared with other algorithms are in the performance under CityScapes data set
3. present invention of table is compared with other algorithms are in the performance under CamVid data set
It can be seen that accuracy rate and mIoU of the invention by the above correlation data and have compared with existing algorithm and significantly mention
It is high.
It is emphasized that embodiment of the present invention be it is illustrative, without being restrictive, therefore packet of the present invention
Include and be not limited to embodiment described in specific embodiment, it is all by those skilled in the art according to the technique and scheme of the present invention
The other embodiments obtained, also belong to the scope of protection of the invention.
Claims (5)
1. a kind of semantic segmentation method of the multipath Fusion Features based on deep learning, it is characterised in that the following steps are included:
Step 1, the depth of foundation feature that image is extracted using multipath Feature fusion；
The depth of foundation feature of extraction is passed through decoding end network by step 2, restores original image resolution information, and generate and divide
Cut result；
Step 3 trains network by target of cross entropy loss function, and accuracy rate and mIoU is used to evaluate network performance.
2. a kind of semantic segmentation method of multipath Fusion Features based on deep learning according to claim 1, special
Sign is: the depth of foundation feature includes global information and local message and is merged by the output of different convolutional layers
It arrives.
3. a kind of semantic segmentation method of multipath Fusion Features based on deep learning according to claim 1 or 2,
Be characterized in that: the concrete methods of realizing of the step 1 comprises the steps of:
(1) input picture is zoomed into uniform sizes 256 × 256, using VGG16 convolutional neural networks framework as the base of pre-training
Network is divided into 4 modules according to the size of output feature resolution by plinth convolutional neural networks, each module is in primitive network
On the basis of add two paths, the quantity of convolutional layer is 1 and 2 respectively；
(2) the image after scaling is input in modified VGG16 network structure, image passes through a series of convolution, batch normalizing
After change, Chi Hua, ReLU operation, each module exports a kind of feature, and resolution ratio is followed successively by 64 × 64,32 × 32,16 × 16 and 8
× 8 local feature.
4. a kind of semantic segmentation method of multipath Fusion Features based on deep learning according to claim 1 or 2,
Be characterized in that: the concrete methods of realizing of the step 2 comprises the steps of:
(1) 8 × 8 features are up-sampled into layer by convolution sum, 16 × 16 characteristic pattern is obtained, by this feature and coding side 16 × 16
Convolutional layer cascade, obtain more 16 × 16 characteristic patterns；
(2) 16 × 16 features are also passed through into the processing in (1), obtain 32 × 32 features；
(3) successively execute, until obtaining 128 × 128 characteristic patterns；
(4) 128 × 128 characteristic patterns are generated into 256 × 256 segmentation figure by up-sampling and process of convolution.
5. a kind of semantic segmentation method of multipath Fusion Features based on deep learning according to claim 1 or 2,
Be characterized in that: the concrete methods of realizing of the step 3 comprises the steps of:
(1) calculate prediction segmentation figure and the segmentation figure that has marked intersects entropy loss, utilizes back-propagation algorithm update weight.
(2) after the completion of network training, its estimated performance is measured using accuracy rate and mIoU.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810842201.4A CN109190626A (en) | 2018-07-27 | 2018-07-27 | A kind of semantic segmentation method of the multipath Fusion Features based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810842201.4A CN109190626A (en) | 2018-07-27 | 2018-07-27 | A kind of semantic segmentation method of the multipath Fusion Features based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109190626A true CN109190626A (en) | 2019-01-11 |
Family
ID=64937258
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810842201.4A Pending CN109190626A (en) | 2018-07-27 | 2018-07-27 | A kind of semantic segmentation method of the multipath Fusion Features based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109190626A (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109961434A (en) * | 2019-03-30 | 2019-07-02 | 西安电子科技大学 | Non-reference picture quality appraisement method towards the decaying of level semanteme |
| CN110020658A (en) * | 2019-03-28 | 2019-07-16 | 大连理工大学 | A kind of well-marked target detection method based on multitask deep learning |
| CN110059772A (en) * | 2019-05-14 | 2019-07-26 | 温州大学 | Remote sensing images semantic segmentation method based on migration VGG network |
| CN110517329A (en) * | 2019-08-12 | 2019-11-29 | 北京邮电大学 | A kind of deep learning method for compressing image based on semantic analysis |
| CN110796205A (en) * | 2019-11-04 | 2020-02-14 | 华南理工大学 | Feature mapping multiplexing codec architecture network and fast segmentation method |
| CN110992320A (en) * | 2019-11-22 | 2020-04-10 | 电子科技大学 | Medical image segmentation network based on double interleaving |
| CN111160351A (en) * | 2019-12-26 | 2020-05-15 | 厦门大学 | Fast high-resolution image segmentation method based on block recommendation network |
| CN111178211A (en) * | 2019-12-20 | 2020-05-19 | 北京迈格威科技有限公司 | Image segmentation method and device, electronic equipment and readable storage medium |
| CN111292317A (en) * | 2020-03-11 | 2020-06-16 | 四川大学华西医院 | Method for enhancing image local feature type multitask segmentation of in-situ cancer region in mammary duct |
| CN111553391A (en) * | 2020-04-09 | 2020-08-18 | 东南大学 | Feature fusion method in semantic segmentation technology |
| CN111666948A (en) * | 2020-05-27 | 2020-09-15 | 厦门大学 | Real-time high-performance semantic segmentation method and device based on multi-path aggregation |
| EP3712811A1 (en) * | 2019-03-20 | 2020-09-23 | NavInfo Europe B.V. | Real-time scene understanding system |
| CN111753892A (en) * | 2020-06-12 | 2020-10-09 | 清华大学 | Deep learning-based interpretation method of global visual field network system |
| CN112597882A (en) * | 2020-12-22 | 2021-04-02 | 自然资源部国土卫星遥感应用中心 | Remote sensing image snow detection method based on deep convolutional neural network |
| AU2020203026B2 (en) * | 2019-05-08 | 2021-06-24 | Raytheon Company | Calculating the precision of image annotations |
| CN113033570A (en) * | 2021-03-29 | 2021-06-25 | 同济大学 | Image semantic segmentation method for improving fusion of void volume and multilevel characteristic information |
| CN113139468A (en) * | 2021-04-24 | 2021-07-20 | 西安交通大学 | Video abstract generation method fusing local target features and global features |
| CN113597613A (en) * | 2019-03-22 | 2021-11-02 | 辉达公司 | Shape fusion for image analysis |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150319424A1 (en) * | 2014-04-30 | 2015-11-05 | Replay Technologies Inc. | System and method of multi-view reconstruction with user-selectable novel views |
| CN105676253A (en) * | 2016-01-15 | 2016-06-15 | 武汉光庭科技有限公司 | Longitudinal positioning system and method based on city road marking map in automatic driving |
| CN107085842A (en) * | 2017-04-01 | 2017-08-22 | 上海讯陌通讯技术有限公司 | The real-time antidote and system of self study multiway images fusion |
| CN107563381A (en) * | 2017-09-12 | 2018-01-09 | 国家新闻出版广电总局广播科学研究院 | The object detection method of multiple features fusion based on full convolutional network |
| CN107644426A (en) * | 2017-10-12 | 2018-01-30 | 中国科学技术大学 | Image, semantic dividing method based on pyramid pond encoding and decoding structure |
| CN107704866A (en) * | 2017-06-15 | 2018-02-16 | 清华大学 | Multitask Scene Semantics based on new neural network understand model and its application |
-
2018
- 2018-07-27 CN CN201810842201.4A patent/CN109190626A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150319424A1 (en) * | 2014-04-30 | 2015-11-05 | Replay Technologies Inc. | System and method of multi-view reconstruction with user-selectable novel views |
| CN105676253A (en) * | 2016-01-15 | 2016-06-15 | 武汉光庭科技有限公司 | Longitudinal positioning system and method based on city road marking map in automatic driving |
| CN107085842A (en) * | 2017-04-01 | 2017-08-22 | 上海讯陌通讯技术有限公司 | The real-time antidote and system of self study multiway images fusion |
| CN107704866A (en) * | 2017-06-15 | 2018-02-16 | 清华大学 | Multitask Scene Semantics based on new neural network understand model and its application |
| CN107563381A (en) * | 2017-09-12 | 2018-01-09 | 国家新闻出版广电总局广播科学研究院 | The object detection method of multiple features fusion based on full convolutional network |
| CN107644426A (en) * | 2017-10-12 | 2018-01-30 | 中国科学技术大学 | Image, semantic dividing method based on pyramid pond encoding and decoding structure |
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111723635B (en) * | 2019-03-20 | 2023-08-18 | 北京四维图新科技股份有限公司 | Real-time scene understanding system |
| US11210547B2 (en) * | 2019-03-20 | 2021-12-28 | NavInfo Europe B.V. | Real-time scene understanding system |
| CN111723635A (en) * | 2019-03-20 | 2020-09-29 | 北京四维图新科技股份有限公司 | Real-time scene understanding system |
| EP3712811A1 (en) * | 2019-03-20 | 2020-09-23 | NavInfo Europe B.V. | Real-time scene understanding system |
| CN113597613A (en) * | 2019-03-22 | 2021-11-02 | 辉达公司 | Shape fusion for image analysis |
| CN110020658A (en) * | 2019-03-28 | 2019-07-16 | 大连理工大学 | A kind of well-marked target detection method based on multitask deep learning |
| CN109961434B (en) * | 2019-03-30 | 2022-12-06 | 西安电子科技大学 | No-reference image quality evaluation method for hierarchical semantic attenuation |
| CN109961434A (en) * | 2019-03-30 | 2019-07-02 | 西安电子科技大学 | Non-reference picture quality appraisement method towards the decaying of level semanteme |
| US11315352B2 (en) | 2019-05-08 | 2022-04-26 | Raytheon Company | Calculating the precision of image annotations |
| AU2020203026B2 (en) * | 2019-05-08 | 2021-06-24 | Raytheon Company | Calculating the precision of image annotations |
| CN110059772B (en) * | 2019-05-14 | 2021-04-30 | 温州大学 | Remote sensing image semantic segmentation method based on multi-scale decoding network |
| CN110059772A (en) * | 2019-05-14 | 2019-07-26 | 温州大学 | Remote sensing images semantic segmentation method based on migration VGG network |
| CN110517329B (en) * | 2019-08-12 | 2021-05-14 | 北京邮电大学 | Deep learning image compression method based on semantic analysis |
| CN110517329A (en) * | 2019-08-12 | 2019-11-29 | 北京邮电大学 | A kind of deep learning method for compressing image based on semantic analysis |
| CN110796205B (en) * | 2019-11-04 | 2023-03-24 | 华南理工大学 | Feature mapping multiplexing codec architecture network and fast segmentation method |
| CN110796205A (en) * | 2019-11-04 | 2020-02-14 | 华南理工大学 | Feature mapping multiplexing codec architecture network and fast segmentation method |
| CN110992320B (en) * | 2019-11-22 | 2023-03-21 | 电子科技大学 | Medical image segmentation network based on double interleaving |
| CN110992320A (en) * | 2019-11-22 | 2020-04-10 | 电子科技大学 | Medical image segmentation network based on double interleaving |
| CN111178211B (en) * | 2019-12-20 | 2024-01-12 | 天津极豪科技有限公司 | Image segmentation method, device, electronic equipment and readable storage medium |
| CN111178211A (en) * | 2019-12-20 | 2020-05-19 | 北京迈格威科技有限公司 | Image segmentation method and device, electronic equipment and readable storage medium |
| CN111160351A (en) * | 2019-12-26 | 2020-05-15 | 厦门大学 | Fast high-resolution image segmentation method based on block recommendation network |
| CN111292317A (en) * | 2020-03-11 | 2020-06-16 | 四川大学华西医院 | Method for enhancing image local feature type multitask segmentation of in-situ cancer region in mammary duct |
| CN111292317B (en) * | 2020-03-11 | 2022-06-07 | 四川大学华西医院 | Method for enhancing image local feature type multitask segmentation of in-situ cancer region in mammary duct |
| CN111553391A (en) * | 2020-04-09 | 2020-08-18 | 东南大学 | Feature fusion method in semantic segmentation technology |
| CN111666948B (en) * | 2020-05-27 | 2023-05-30 | 厦门大学 | Real-time high-performance semantic segmentation method and device based on multipath aggregation |
| CN111666948A (en) * | 2020-05-27 | 2020-09-15 | 厦门大学 | Real-time high-performance semantic segmentation method and device based on multi-path aggregation |
| CN111753892B (en) * | 2020-06-12 | 2022-10-18 | 清华大学 | Deep learning-based interpretation method of global visual field network system |
| CN111753892A (en) * | 2020-06-12 | 2020-10-09 | 清华大学 | Deep learning-based interpretation method of global visual field network system |
| CN112597882A (en) * | 2020-12-22 | 2021-04-02 | 自然资源部国土卫星遥感应用中心 | Remote sensing image snow detection method based on deep convolutional neural network |
| CN113033570B (en) * | 2021-03-29 | 2022-11-11 | 同济大学 | Image semantic segmentation method for improving void convolution and multilevel characteristic information fusion |
| CN113033570A (en) * | 2021-03-29 | 2021-06-25 | 同济大学 | Image semantic segmentation method for improving fusion of void volume and multilevel characteristic information |
| CN113139468A (en) * | 2021-04-24 | 2021-07-20 | 西安交通大学 | Video abstract generation method fusing local target features and global features |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109190626A (en) | A kind of semantic segmentation method of the multipath Fusion Features based on deep learning | |
| CN109190752A (en) | The image, semantic dividing method of global characteristics and local feature based on deep learning | |
| CN110738207B (en) | Character detection method for fusing character area edge information in character image | |
| CN111553387B (en) | Personnel target detection method based on Yolov3 | |
| Wang et al. | RENet: Rectangular convolution pyramid and edge enhancement network for salient object detection of pavement cracks | |
| CN107590489A (en) | Object detection method based on concatenated convolutional neutral net | |
| CN111275688A (en) | Small target detection method based on context feature fusion screening of attention mechanism | |
| CN109902806A (en) | Method is determined based on the noise image object boundary frame of convolutional neural networks | |
| CN112241762B (en) | Fine-grained identification method for pest and disease damage image classification | |
| CN111898432B (en) | Pedestrian detection system and method based on improved YOLOv3 algorithm | |
| CN110569782A (en) | Target detection method based on deep learning | |
| CN112949408B (en) | Real-time identification method and system for target fish passing through fish channel | |
| CN112115795B (en) | Hyperspectral image classification method based on Triple GAN | |
| CN110458132A (en) | One kind is based on random length text recognition method end to end | |
| CN116206112A (en) | Remote sensing image semantic segmentation method based on multi-scale feature fusion and SAM | |
| Nayef et al. | Statistical grouping for segmenting symbols parts from line drawings, with application to symbol spotting | |
| Yin et al. | Road Damage Detection and Classification based on Multi-level Feature Pyramids. | |
| CN108197637A (en) | A kind of high pressure cabinet switch detection method based on deep learning | |
| CN103455798A (en) | Human detection method based on maximum geometric flow direction column diagram | |
| CN104408476A (en) | Deep sparse main component analysis-based polarimetric SAR image classification method | |
| CN113610088A (en) | Self-attention-based license plate character recognition method | |
| Chen et al. | Improved Faster R-CNN identification method for containers | |
| Wu et al. | Spoon surface defect detection based on improved yolo v3 | |
| Jiang | Deformable convolutional neural network for fine-grained image recognition | |
| CN110163106A (en) | Integral type is tatooed detection and recognition methods and system |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190111 |
|
| WD01 | Invention patent application deemed withdrawn after publication |