CN113837046A - Small sample remote sensing image scene classification method based on iterative feature distribution learning - Google Patents
Small sample remote sensing image scene classification method based on iterative feature distribution learning Download PDFInfo
- Publication number
- CN113837046A CN113837046A CN202111089674.XA CN202111089674A CN113837046A CN 113837046 A CN113837046 A CN 113837046A CN 202111089674 A CN202111089674 A CN 202111089674A CN 113837046 A CN113837046 A CN 113837046A
- Authority
- CN
- China
- Prior art keywords
- feature
- learning
- matrix
- small sample
- classification
- 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
- 238000000034 method Methods 0.000 title claims abstract description 47
- 239000011159 matrix material Substances 0.000 claims abstract description 44
- 238000012549 training Methods 0.000 claims abstract description 8
- 238000004364 calculation method Methods 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 239000000571 coke Substances 0.000 claims description 2
- 238000013461 design Methods 0.000 abstract description 4
- 238000012512 characterization method Methods 0.000 abstract description 3
- 230000007246 mechanism Effects 0.000 abstract description 3
- 238000013135 deep learning Methods 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a small sample remote sensing image scene classification method based on iterative feature distribution learning, which comprises the following steps of: step one, setting an iterative feature distribution learning network framework; generating a similarity matrix between the image samples; step three, training the small sample classifier by using guiding knowledge based on the classification prediction probability and the similarity matrix; and step four, correcting the characteristic distribution by using the prediction probability distribution matrix and the attention weight, and then iteratively updating the whole network. The invention has simple structure and reasonable design, firstly generates the similarity matrix to obtain the characteristic correlation among samples, secondly trains the classifier under the guidance of historical classification prediction, finally corrects the characteristics by adopting an attention mechanism, and re-inputs the corrected characteristics into the network to realize the iterative update of the whole network. The method can further improve the characterization capability of the category through iterative feature distribution learning, thereby improving the accuracy of small sample classification.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a small sample remote sensing image scene classification method based on iterative feature distribution learning.
Background
With the development of artificial intelligence technology, deep learning can achieve satisfactory results in some scenarios on computer vision tasks by means of deep and complex network models, huge training data support and strong hardware support. However, for a small amount of (single) sample learning tasks, deep learning cannot improve learning ability from the aspects of complex network models and a large amount of training data. Therefore, small sample learning techniques that learn data patterns from a small number of (single) samples are currently the direction of intense research in the field of deep learning.
Small sample learning aims at achieving the ability to identify unknown classes with a small number of training samples, which is similar to the process of the human brain to link and reason about unknown things based on a priori knowledge. The core of small sample learning is to classify unlabeled samples correctly in the case of few labeled samples.
The small sample remote sensing image scene classification technology based on deep learning can realize high-precision classification of label-free images under the condition of limited labeling data. Therefore, the small sample remote sensing image scene classification technology has a huge application prospect in the applications of earth environment monitoring, ground target classification, rare animal classification and the like.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a small sample remote sensing image scene classification method based on iterative feature distribution learning, which is simple in structure and reasonable in design, aiming at the problems that when the number of samples is insufficient, a depth network model is easy to over-fit and the classification performance is poor. The overall design idea is as follows: firstly generating a similarity matrix to obtain characteristic correlation among samples, secondly training a classifier under the guidance of historical classification prediction, finally correcting the characteristics by adopting an attention mechanism, and inputting the corrected characteristics into the network again to realize iterative update of the whole network. Through iterative feature distribution learning, the characterization capability of the category can be further improved, and therefore the accuracy of small sample classification is improved.
In order to solve the technical problems, the invention adopts the technical scheme that: a small sample remote sensing image scene classification method based on iterative feature distribution learning is characterized in that: the method comprises the following steps:
step one, setting an iterative feature distribution learning network framework:
step 101, setting a network framework, including a feature extractor F and a relation measurement model FRClassifier fLAttention model fAFeature learning model fC；
Step 102, setting the learning rate and the iteration times of a network;
step two, generating a similarity matrix between the image samples:
step 201, inputting an image set X composed of a support set and a query set into a feature extractor F to obtain a feature Fk(X)；
Step 202, feature Fk(X) expanding in a first dimension to obtain a feature Fi k(X)；
Step 203, based on the feature Fi k(X) exchanging the first dimension and the second dimension to obtain the characteristics
Step 204, calculating a relation matrix
And similarity matrix
Step three, training the small sample classifier by using guiding knowledge based on the classification prediction probability and the similarity matrix:
step 301, feature Fk(X) inputting the data into a classifier to obtain a classification prediction probability
Step 302, predict probability of classification
Similarity matrix with next iteration
Stacking in a second dimension as a guide knowledge;
step 303, input the stacking result into the classifier, using the focus loss LFLOptimizing the classifier;
step four, correcting the characteristic distribution by using the prediction probability distribution matrix and the attention weight, and then iteratively updating the whole network:
step 401, predict probability of classification
Expanding on the first dimension to obtain a probability matrix Pi k(X)；
Step 402, based on the probability matrix Pi k(X) exchanging the first dimension and the second dimension to obtain a matrix
Step 403, calculating a prediction probability distribution matrix
And attention weight
Thereby obtaining corrected features Fk+1(X)；
And step 404, starting from step 202 in the step two, repeating the process, and obtaining a final classification result after the specified iteration times.
The small sample remote sensing image scene classification method based on iterative feature distribution learning is characterized in that:the relationship matrix in step 204
The calculation formula of (2) is as follows:
similarity matrix
The calculation formula of (2) is as follows:
where k denotes the kth iteration, FR[·]Representing a relational metric function.
The small sample remote sensing image scene classification method based on iterative feature distribution learning is characterized in that: classification of prediction probabilities in step 301
The calculation formula of (2) is as follows:
where Cat (-) denotes stacking on channels, OneHot (-) denotes unique Hot code, fL[·]Representing a classifier function, y is the true label corresponding to the support set sample, PinitRepresenting the initialized query set sample probability scores.
The small sample remote sensing image scene classification method based on iterative feature distribution learning is characterized in that: loss of coke L in step 303FLThe calculation formula of (2) is as follows:
wherein
The method comprises the steps of classifying and predicting probability, y is a sample real label, alpha represents a parameter for adjusting loss weight, and gamma represents an adjusting factor for adjusting the descending rate of the sample weight.
A base as described aboveThe small sample remote sensing image scene classification method based on iterative feature distribution learning is characterized by comprising the following steps of: probability distribution matrix in step 403
The calculation formula of (2) is as follows:
attention weight
The calculation formula of (2) is as follows:
correction feature Fk+1The calculation formula of (X) is:
wherein f isA[·]Representing the attention model function, fC{. denotes a feature learning model function.
The small sample remote sensing image scene classification method based on iterative feature distribution learning is characterized in that: relationship metric model FRThe network structure of (a) is as follows: input layer → first convolution layer → second convolution layer → third convolution layer; classifier fLThe network structure of (a) is as follows: input layer → first fully connected layer → second fully connected layer; attention model fAThe network structure of (a) is as follows: input layer → first convolution layer → second convolution layer → third convolution layer; feature learning model fCThe network structure of (a) is as follows: input layer → first convolutional layer → second convolutional layer.
The small sample remote sensing image scene classification method based on iterative feature distribution learning is characterized in that: the convolutional layers are all based on one basic convolution unit of a neural network VGG, ResNet, GoogleNet or AlexNet.
Compared with the prior art, the invention has the following advantages:
1. in the method, a characteristic distribution learning mode is adopted, the characteristic association among the samples is firstly obtained, then the classifier is optimized, and then the sample characteristic is corrected by adopting an attention mechanism, so that the effective information of the image target can be strengthened, the characterization capability of the category is enhanced, and the classification accuracy of the small samples is further improved.
2. The invention has simple structure and reasonable design, adopts the model architecture of loop iteration, and continuously adjusts the model network parameters better by optimizing the characteristics, thereby effectively improving the convergence speed of the model.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
The method of the present invention will be described in further detail below with reference to the accompanying drawings and embodiments of the invention.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Spatially relative terms, such as "above … …," "above … …," "above … …," "above," and the like, may be used herein for ease of description to describe one device or feature's spatial relationship to another device or feature as illustrated in the figures. It will be understood that the spatially relative terms are intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the figures. For example, if a device in the figures is turned over, devices described as "above" or "on" other devices or configurations would then be oriented "below" or "under" the other devices or configurations. Thus, the exemplary term "above … …" can include both an orientation of "above … …" and "below … …". The device may be otherwise variously oriented (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein interpreted accordingly.
As shown in fig. 1, the method for classifying scenes of small sample remote sensing images based on iterative feature distribution learning of the present invention includes the following steps:
step one, setting an iterative feature distribution learning network framework:
step 101, setting a network framework, including a feature extractor F and a relation measurement model FRClassifier fLAttention model fAFeature learning model fC；
Step 102, setting the learning rate of the network to be 0.001 and the iteration number to be 1000;
in the small sample remote sensing image scene classification method based on iterative feature distribution learning, it needs to be explained that a relation measurement model FRNetwork structure ofSequentially comprises the following steps: input layer → first convolution layer → second convolution layer → third convolution layer; classifier fLThe network structure of (a) is as follows: input layer → first fully connected layer → second fully connected layer; attention model fAThe network structure of (a) is as follows: input layer → first convolution layer → second convolution layer → third convolution layer; feature learning model fCThe network structure of (a) is as follows: input layer → first convolutional layer → second convolutional layer.
In the method for classifying the small sample remote sensing image scene based on iterative feature distribution learning, it should be noted that the convolution layers are all basic convolution units based on one of neural networks VGG, ResNet, GoogleNet or AlexNet.
Step two, generating a similarity matrix between the image samples:
step 201, inputting an image set X composed of a support set and a query set into a feature extractor F to obtain a feature Fk(X); for each batch of tasks, the dimension of the image set X is (10, 3, 84, 84), and after the feature extractor F, the data dimension becomes (10, 128, 5, 5), i.e., Fk(X) dimension (10, 128, 5, 5);
step 202, feature Fk(X) expanding in a first dimension to obtain a feature Fi k(X)，Fi k(X) dimension (1, 10, 128, 5, 5);
step 203, based on the feature Fi k(X) exchanging the first dimension and the second dimension to obtain the characteristics
Dimension (10, 1, 128, 5, 5);
step 204, calculating a relation matrix
And similarity matrix
Relationship matrix
The calculation formula of (2) is as follows:
similarity matrix
The calculation formula of (2) is as follows:
where k denotes the kth iteration, FR[·]Representing a relational metric function.
Step three, training the small sample classifier by using guiding knowledge based on the classification prediction probability and the similarity matrix:
step 301, feature Fk(X) inputting the data into a classifier to obtain a classification prediction probability
The calculation formula is as follows:
where Cat (-) denotes stacking on channels, OneHot (-) denotes unique Hot code, fL[·]Representing a classifier function, y is the true label corresponding to the support set sample, PinitRepresenting an initialized query set sample probability score;
step 302, predict probability of classification
Similarity matrix with next iteration
Stacking in a second dimension as a guide knowledge;
step 303, input the stacking result into the classifier, using the focus loss LFLOptimization of the classifier, LFLThe calculation formula of (2) is as follows:
wherein
The method comprises the steps of classifying and predicting probability, y is a sample real label, alpha represents a parameter for adjusting loss weight, and gamma represents an adjusting factor for adjusting the descending rate of the sample weight.
Step four, correcting the characteristic distribution by using the prediction probability distribution matrix and the attention weight, and then iteratively updating the whole network:
step 401, predict probability of classification
Expanding on the first dimension to obtain a probability matrix Pi k(X)，
Is (10, 5), Pi kThe dimension of (X) is (1, 10, 5);
step 402, based on the probability matrix Pi k(X) exchanging the first dimension and the second dimension to obtain a matrix
Is (10, 1, 5);
step 403, calculating a prediction probability distribution matrix
And attention weight
Thereby obtaining corrected features Fk+1(X), probability distribution matrix
The calculation formula of (2) is as follows:
attention weight
The calculation formula of (2) is as follows:
correction feature Fk+1The calculation formula of (X) is:
wherein f isA[·]Representing the attention model function, fC{. represents a feature learning model function;
and step 404, starting from step 202 in the step two, repeating the process, and obtaining a final classification result after the specified iteration times.
The technical effects of the invention are explained by simulation experiments as follows:
1. simulation conditions and content
The experimental data of the invention is a remote sensing image data set NWPU-RESISC45 collected by northwest industrial university, the image size is 256 multiplied by 256, 45 types of different ground features, and the total number is 31500 images. Table one is the comparison of the accuracy of the existing small sample classification technique. In simulation experiments, the invention and the comparison method are both implemented in Python3.7 by programming.
2. Analysis of simulation results
Table-small sample classification accuracy comparison
As can be seen from the table I, compared with the existing model based on the measurement method and the model based on the gradient method, the method provided by the invention has higher classification precision, and the excellent effect of the method provided by the invention on classifying the remote sensing images of small samples is proved. According to the 5-way 1-shot and 5-way 5-shot experimental results of the small sample classification standard, the method disclosed by the invention can effectively improve the classification effect of the small sample remote sensing image.
The above embodiments are only examples of the present invention, and are not intended to limit the present invention, and all simple modifications, changes and equivalent structural changes made to the above embodiments according to the technical spirit of the present invention still fall within the protection scope of the technical solution of the present invention.
Claims (7)
1. A small sample remote sensing image scene classification method based on iterative feature distribution learning is characterized in that: the method comprises the following steps:
step one, setting an iterative feature distribution learning network framework:
step 101, setting a network framework, including a feature extractor F and a relation measurement model FRClassifier fLAttention model fAFeature learning model fC；
Step 102, setting the learning rate and the iteration times of a network;
step two, generating a similarity matrix between the image samples:
step 201, inputting an image set X composed of a support set and a query set into a feature extractor F to obtain a feature Fk(X)；
Step 202, feature Fk(X) expanding in a first dimension to obtain a feature Fi k(X)；
Step 203, based on the feature Fi k(X) exchanging the first dimension and the second dimension to obtain the characteristics
Step 204, calculating a relation matrix
And similarity matrix
Step three, training the small sample classifier by using guiding knowledge based on the classification prediction probability and the similarity matrix:
step 301, feature Fk(X) inputting the data into a classifier to obtain a classification prediction probability
Step 302, predict probability of classification
Similarity matrix with next iteration
Stacking in a second dimension as a guide knowledge;
step 303, input the stacking result into the classifier, using the focus loss LFLOptimizing the classifier;
step four, correcting the characteristic distribution by using the prediction probability distribution matrix and the attention weight, and then iteratively updating the whole network:
step 401, predict probability of classification
Expanding on the first dimension to obtain a probability matrix Pi k(X)；
Step 402, based on the probability matrix Pi k(X) exchanging the first dimension and the second dimension to obtain a matrix
Step 403, calculating a prediction probability distribution matrix
And attention weight
Thereby obtaining corrected features Fk+1(X)；
And step 404, starting from step 202 in the step two, repeating the process, and obtaining a final classification result after the specified iteration times.
2. The method for classifying scenes of the small sample remote sensing images based on the iterative feature distribution learning as claimed in claim 1, wherein: the relationship matrix in step 204
The calculation formula of (2) is as follows:
similarity matrix
The calculation formula of (2) is as follows:
where k denotes the kth iteration, FR[·]Representing a relational metric function.
3. The method for classifying scenes of the small sample remote sensing images based on the iterative feature distribution learning as claimed in claim 1, wherein: classification of prediction probabilities in step 301
The calculation formula of (2) is as follows:
where Cat (-) denotes stacking on channels, OneHot (-) denotes unique Hot code, fL[·]Representing a classifier function, y is the true label corresponding to the support set sample, PinitRepresenting the initialized query set sample probability scores.
4. The method for classifying scenes of the small sample remote sensing images based on the iterative feature distribution learning as claimed in claim 1, wherein: loss of coke L in step 303FLThe calculation formula of (2) is as follows:
wherein
The method comprises the steps of classifying and predicting probability, y is a sample real label, alpha represents a parameter for adjusting loss weight, and gamma represents an adjusting factor for adjusting the descending rate of the sample weight.
5. The method for classifying scenes of the small sample remote sensing images based on the iterative feature distribution learning as claimed in claim 1, wherein: probability distribution matrix in step 403
The calculation formula of (2) is as follows:
attention weight
The calculation formula of (2) is as follows:
correction feature Fk+1The calculation formula of (X) is:
wherein f isA[·]Representing the attention model function, fC{. denotes a feature learning model function.
6. The method for classifying scenes of small-sample remote sensing images based on iterative feature distribution learning as claimed in claim 1, wherein the method is characterized in that: relationship metric model FRThe network structure of (a) is as follows: input layer → first convolution layer → second convolution layer → third convolution layer; classifier fLThe network structure of (a) is as follows: input layer → first fully connected layer → second fully connected layer; attention model fAThe network structure of (a) is as follows: input layer → first convolution layer → second convolution layer → third convolution layer; feature learning model fCThe network structure of (a) is as follows: input layer → first convolutional layer → second convolutional layer.
7. The small sample remote sensing image scene classification method based on iterative feature distribution learning according to claim 6, characterized in that: the convolutional layers are all based on one basic convolution unit of a neural network VGG, ResNet, GoogleNet or AlexNet.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111089674.XA CN113837046A (en) | 2021-09-17 | 2021-09-17 | Small sample remote sensing image scene classification method based on iterative feature distribution learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111089674.XA CN113837046A (en) | 2021-09-17 | 2021-09-17 | Small sample remote sensing image scene classification method based on iterative feature distribution learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113837046A true CN113837046A (en) | 2021-12-24 |
Family
ID=78959652
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111089674.XA Pending CN113837046A (en) | 2021-09-17 | 2021-09-17 | Small sample remote sensing image scene classification method based on iterative feature distribution learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113837046A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114120048A (en) * | 2022-01-26 | 2022-03-01 | 中兴通讯股份有限公司 | Image processing method, electronic device and computer storage medium |
| WO2023226227A1 (en) * | 2022-05-27 | 2023-11-30 | 福建龙氟新材料有限公司 | Automatic batching system for preparing electronic-grade hydrofluoric acid and batching method therefor |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110097173A (en) * | 2019-04-04 | 2019-08-06 | 天津大学 | A kind of convolutional neural networks topological structure improving network performance and generalization ability |
| CN113222011A (en) * | 2021-05-10 | 2021-08-06 | 西北工业大学 | Small sample remote sensing image classification method based on prototype correction |
| US20210264150A1 (en) * | 2020-02-26 | 2021-08-26 | Central South University | Urban remote sensing image scene classification method in consideration of spatial relationships |
-
2021
- 2021-09-17 CN CN202111089674.XA patent/CN113837046A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110097173A (en) * | 2019-04-04 | 2019-08-06 | 天津大学 | A kind of convolutional neural networks topological structure improving network performance and generalization ability |
| US20210264150A1 (en) * | 2020-02-26 | 2021-08-26 | Central South University | Urban remote sensing image scene classification method in consideration of spatial relationships |
| CN113222011A (en) * | 2021-05-10 | 2021-08-06 | 西北工业大学 | Small sample remote sensing image classification method based on prototype correction |
Non-Patent Citations (2)
| Title |
|---|
| QINGJIE ZENG ET AL.: "IDLN: Iterative Distribution Learning Network for Few-Shot Remote Sensing Image Scene Classification", IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, pages 1 - 4 * |
| 耿杰: "基于深度学习的SAR遥感图像分类方法研究", 中国优秀硕士学位论文全文数据库 工程科技II辑 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114120048A (en) * | 2022-01-26 | 2022-03-01 | 中兴通讯股份有限公司 | Image processing method, electronic device and computer storage medium |
| WO2023226227A1 (en) * | 2022-05-27 | 2023-11-30 | 福建龙氟新材料有限公司 | Automatic batching system for preparing electronic-grade hydrofluoric acid and batching method therefor |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111191732B (en) | Target detection method based on full-automatic learning | |
| CN109190524B (en) | Human body action recognition method based on generation of confrontation network | |
| CN114241282A (en) | Knowledge distillation-based edge equipment scene identification method and device | |
| CN110070074B (en) | Method for constructing pedestrian detection model | |
| CN109671102B (en) | Comprehensive target tracking method based on depth feature fusion convolutional neural network | |
| CN111160474A (en) | Image identification method based on deep course learning | |
| CN103425996B (en) | A kind of large-scale image recognition methods of parallel distributed | |
| CN111079847B (en) | Remote sensing image automatic labeling method based on deep learning | |
| CN112132014B (en) | Target re-identification method and system based on non-supervised pyramid similarity learning | |
| CN111639679A (en) | Small sample learning method based on multi-scale metric learning | |
| CN113222011B (en) | Small sample remote sensing image classification method based on prototype correction | |
| CN110942472B (en) | Nuclear correlation filtering tracking method based on feature fusion and self-adaptive blocking | |
| CN110516095A (en) | Weakly supervised depth Hash social activity image search method and system based on semanteme migration | |
| CN105760821A (en) | Classification and aggregation sparse representation face identification method based on nuclear space | |
| CN113128620B (en) | Semi-supervised domain self-adaptive picture classification method based on hierarchical relationship | |
| CN113837046A (en) | Small sample remote sensing image scene classification method based on iterative feature distribution learning | |
| CN106156805A (en) | A kind of classifier training method of sample label missing data | |
| CN110287985B (en) | Depth neural network image identification method based on variable topology structure with variation particle swarm optimization | |
| CN110110128B (en) | Fast supervised discrete hash image retrieval system for distributed architecture | |
| CN112200262B (en) | Small sample classification training method and device supporting multitasking and cross-tasking | |
| CN104680167B (en) | Auroral oval location determining method based on deep learning | |
| CN116597244A (en) | Small sample target detection method based on meta-learning method | |
| CN111652177A (en) | Signal feature extraction method based on deep learning | |
| CN114187506B (en) | Remote sensing image scene classification method of viewpoint-aware dynamic routing capsule network | |
| CN117671673B (en) | Small sample cervical cell classification method based on self-adaptive tensor subspace |
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: 20211224 |
|
| WD01 | Invention patent application deemed withdrawn after publication |