CN113269243A - Association identification method and device based on generative countermeasure network - Google Patents
Association identification method and device based on generative countermeasure network Download PDFInfo
- Publication number
- CN113269243A CN113269243A CN202110541068.0A CN202110541068A CN113269243A CN 113269243 A CN113269243 A CN 113269243A CN 202110541068 A CN202110541068 A CN 202110541068A CN 113269243 A CN113269243 A CN 113269243A
- Authority
- CN
- China
- Prior art keywords
- training
- network
- detection signal
- generative
- real
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2132—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on discrimination criteria, e.g. discriminant analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06N3/048—Activation functions
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
A correlation identification method and equipment based on a generative countermeasure network are provided, the correlation identification method comprises the following steps: the method comprises the steps of taking the category of an object to be identified as a target of network training, enabling a set of random speckle sequences to act on objects with the same type and different forms of the object to be identified, carrying out sampling for multiple times to form a barrel detection signal array of the object, enabling different objects to correspond to different barrel detection signal arrays, taking the barrel detection signal array as a sample of the network training, and training by using a generative confrontation network; after training is finished, detecting a target object by using a random speckle sequence used in the training process, collecting corresponding barrel detection signals, and forming a corresponding barrel detection signal array after multiple times of sampling is finished; and inputting the bucket detection signal array into the trained generative confrontation network, and outputting the target object class. The invention realizes the identification without depending on the target image information and can greatly shorten the time for acquiring and processing data.
Description
Technical Field
The invention belongs to the field of target identification, and relates to a correlation identification method and device based on a generative countermeasure network.
Background
Target identification is always a key and difficult problem of great concern in the fields of national economy and military wars. Currently, identification techniques based on target image information depend heavily on the quality of the target image information acquisition, as well as the quality of the processing and analysis of the acquired images. Meanwhile, the complexity of the system is increased invisibly in the mode, the recognition speed is influenced, and the practical process and the application scene of the target recognition technology are restricted. Related imaging, also known as quantum imaging or ghost imaging, is a novel imaging technique developed on the basis of quantum entanglement. In 1995, it was first completed by the Chinese scientists Shih inkstone and Pittman at the university of Marylan, USA. Compared with the traditional imaging technology, the correlated imaging has the characteristics of lens-free imaging, strong disturbance resistance, non-localization and the like, and has good application prospects in the aspects of remote sensing imaging, weak light detection, medical imaging, security inspection, penetrating scattering medium imaging and the like. Therefore, there is a need to implement a scheme that can apply correlation imaging to target recognition.
Disclosure of Invention
The invention aims to solve the problem of target identification depending on target image information in the prior art, and provides a correlation identification method and device based on a generative confrontation network.
In order to achieve the purpose, the invention has the following technical scheme:
an association identification method based on a generative countermeasure network comprises the following steps:
s1, taking the category of the object to be identified as the target of network training, enabling a set of random speckle sequences to act on the object with the same type and different form of the object to be identified, sampling for multiple times to form a barrel detection signal array of the object, enabling different objects to correspond to different barrel detection signal arrays, taking the barrel detection signal array as a sample of network training, and training by using a generative confrontation network;
s2, after the training is finished, detecting the target object by using the random speckle sequence used in the training process of the step S1, collecting corresponding barrel detection signals, and forming a corresponding barrel detection signal array after the sampling for many times is finished;
and S3, inputting the bucket detection signal array of the step S2 into the trained generative confrontation network, and outputting the target object type.
As a preferred embodiment of the correlation identification method of the present invention, the step S1 of acting a set of random speckle sequences on objects of the same type and different shapes of the object to be identified includes the following steps: a set of random speckles of known sequence is used to illuminate different classes of objects in turn, and the total intensity of the light reflected by the objects is collected, i.e. a bucket detection signal array.
As a preferable solution of the association identification method of the present invention, the training of step S1 by using the generative confrontation network includes the following steps:
1) sampling real data, obtaining a corresponding label y, transmitting the label y to a discriminator, and updating parameters according to an output result;
2) generating random noise, and transmitting the label y in the step 1) and the random noise into a generator to generate false data;
3) transmitting the false data generated in the step 2) and the label y into a discriminator;
4) the generator adjusts the parameters according to the output of the discriminator;
5) through successive cycles until the generator and arbiter reach nash equilibrium.
As a preferred embodiment of the association identification method of the present invention, step S3 trains the generative confrontation network according to the following steps:
1) preparing a training set and a testing set: according to a computed ghost imaging mechanism, a random speckle sequence is used for illuminating a target for multiple times, and an acquired barrel-shaped signal array is divided into a training set and a testing set;
2) establishing a network: based on the TensorFlow-gpu1.13, keras2.1.5 version, Pycharm was used to construct the following three models: generating a model, a discrimination model and a confrontation training model;
3) training a network: and circularly inputting the images in the training set and the corresponding categories thereof, and simultaneously training the generator and the discriminator model.
As a preferred scheme of the correlation identification method of the invention, each class of bucket detection signal array has a class label c of a corresponding target, and all the class labels c-P are usedCAnd the normally distributed random number z is used as the input of the generator, and the output is XfakeG (c, z); the input of the discriminator is a real barrel detection data sample and corresponding class labels c-P thereofCTagged data X generated by the sum generatorfakeThe output of which is the true-false probability distribution X of the generated data samplerealD (c, x); the objective function of the generative countermeasure network includes two parts: log-likelihood L of real bucket detection samplesSAnd log-likelihood L of real bucket detection class labelsCThe distribution is represented by the following formula:
Ls＝E[logP(S＝real|Xreal)]+E[logP(S＝fake|Xfake)]
Lc＝E[logP(C＝c|Xreal)]+[ElogP(C＝c|Xfake)]。
as a preferred solution of the correlation identification method of the present invention, the generator aims to find a value such that LS+LCTaking the maximum value, the discriminator targeting the found value such that LS-LCThe maximum value is taken.
As a preferable scheme of the association identification method of the present invention, the output probability distribution is 0.5 when the discriminator cannot distinguish whether the input data is real data; the discriminator learns the characteristic distribution of the real sample data in the process of confronting the generator and judges whether the input sample is from the real sample or the generated sample and the corresponding type of the input sample.
The invention also provides a correlation identification system based on the generative countermeasure network, which comprises:
the generation type confrontation network training module is used for taking the category of an object to be identified as a target of network training, enabling a set of random speckle sequences to act on objects with the same type and different forms of the object to be identified, performing sampling for multiple times to form barrel detection signal arrays of the object, enabling different objects to correspond to different barrel detection signal arrays, taking the barrel detection signal arrays as samples of network training, and training by using a generation type confrontation network;
the barrel detection signal array generation module is used for detecting a target object by using a random speckle sequence used in the generation type confrontation network training process after the generation type confrontation network training is finished, collecting corresponding barrel detection signals and forming a corresponding barrel detection signal array after the sampling is finished for multiple times;
and the target object type judging module is used for inputting the bucket detection signal array of the target object into the trained generative confrontation network and outputting the target object type.
The invention also provides a terminal device, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to realize the steps of the association identification method based on the generative countermeasure network.
The invention also provides a computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for identifying associations based on generative confrontation networks.
Compared with the prior art, the invention has the following beneficial effects: the method realizes the identification without depending on target image information, and designs a method for identifying the target through a barrel detection signal array in the correlated imaging based on the system architecture of the correlated imaging and by utilizing the characteristic of a generating type countermeasure network to input random noise. The invention can greatly shorten the time of data acquisition and processing, and the invention recognizes by the echo signal of the target after the training is finished, therefore, the invention does not need to carry out image recovery by a large amount of data, does not need to carry out optimization and analysis processing on the image, and tests show that ten samples are recognized only for 0.081 seconds.
Drawings
FIG. 1 is a block diagram of a recognition method for a target object using handwritten letters according to an embodiment of the present invention;
FIG. 2 is a flowchart of a method for identifying associations based on a generative countermeasure network according to an embodiment of the present invention;
FIG. 3 is an exemplary graph of experimental results of an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
Referring to fig. 1, the invention uses handwritten pictures corresponding to letters ABCDEFGHIJ and numbers 0123456789 as target objects, defines category labels corresponding to the target ABCDEFGHIJ and 0123456789 as 0123456789, and inputs test data and outputs the test data as corresponding category labels after completing GAN network training.
Referring to fig. 2, the association identification method based on the generative countermeasure network of the present invention mainly includes the following steps:
s1, taking the category of the object to be identified as a target for network training, and enabling a set of random speckle sequences to act on objects with the same type and different forms as the object to be identified, wherein the method comprises the following steps: a set of random speckles of known sequence is used to illuminate different classes of objects in turn, and the total intensity of the light reflected by the objects is collected, i.e. a bucket detection signal array.
The method comprises the following steps of forming a barrel detection signal array of an object through multiple sampling, enabling different objects to correspond to different barrel detection signal arrays, using the barrel detection signal array as a sample for network training, and training by using a generative countermeasure network, wherein the method comprises the following steps:
1) sampling real data, obtaining a corresponding label y, transmitting the label y to a discriminator, and updating parameters according to an output result;
2) generating random noise, and transmitting the label y in the step 1) and the random noise into a generator to generate false data;
3) transmitting the false data generated in the step 2) and the label y into a discriminator;
4) the generator adjusts the parameters according to the output of the discriminator;
5) through successive cycles until the generator and arbiter reach nash equilibrium.
S2, after the training is finished, detecting the target object by using the random speckle sequence used in the training process of the step S1, collecting corresponding barrel detection signals, and forming a corresponding barrel detection signal array after the sampling for many times is finished;
and S3, inputting the bucket detection signal array of the step S2 into the trained generative confrontation network, and outputting the target object type.
Training the generative confrontation network according to the following steps:
1) preparing a training set and a testing set: according to a computed ghost imaging mechanism, a random speckle sequence is used for illuminating a target for multiple times, and an acquired barrel-shaped signal array is divided into a training set and a testing set;
2) establishing a network: based on the TensorFlow-gpu1.13, keras2.1.5 version, Pycharm was used to construct the following three models: generating a model, a discrimination model and a confrontation training model;
3) training a network: and circularly inputting the images in the training set and the corresponding categories thereof, and simultaneously training the generator and the discriminator model.
The game competition mode adopted by the generative countermeasure network enables the real data to be approximated without modeling in advance by sampling. This approach, while simple and effective, has the disadvantage that it is too free and the output of the generator cannot be controlled. Therefore, a generation countermeasure network with constraint conditions is provided, specifically, a class label is added to the obtained bucket detection signal array, for example, the bucket detection signal array of the target A corresponds to a class 0 label, for example, the bucket detection signal array of the target B corresponds to a class 1 label, and so on, and the bucket detection signal array is divided into a training set and a test set with labels. In the training process, training set data is sampled, a corresponding label y is obtained and is transmitted to the discriminator, and according to an output result, the generator learns data distribution updating parameters of a bucket detection signal array corresponding to the label, so that the purpose of constraint is achieved. The association identification method based on the generative countermeasure network guides the data generation process by adding constraint conditions in both the discriminator and the generator. This simple improvement has proven to be very effective and widely practiced.
Conditional generation countermeasure network in training processEach type of bucket detection array has a class label c, c-P corresponding to the targetCAnd the normally distributed random number z is used as the input of the generator, and the output is XfakeG (c, z). The input of the discriminator is a real barrel detection data sample and corresponding class labels c-P thereofCTagged data X generated by the sum generatorfakeThe output of which is the true-false probability distribution X of the generated data sampletureD (c, x). The objective function includes two parts: log-likelihood L of real bucket detection samplesSAnd log-likelihood L of real bucket detection class labelsCThe distribution can be expressed as:
Ls＝E[logP(S＝real|Xreal)]+E[logP(S＝fake|Xfake)]
Lc＝E[logP(C＝c|Xreal)]+[ElogP(C＝c|Xfake)]
the output of the discriminator consists of two parts: some of the outputs are true or false of the training sample image, corresponding to LSIt normalizes the output of the upper layer using sigmoid function. The other part outputs the class of the training sample image corresponding to LCWhich normalizes the output of the upper layer using the softmax function.
And (3) taking the cross entropy of the training sample output and the mark as a training loss function, comparing the sample output and the mark, controlling the change of the learning rate by using an optimizer Adam, and finally defining the operation of reducing the loss to carry out the confrontation training. During training, the generator and the discriminator game with each other, and the discriminator aims to find out a proper value to ensure that LS+LCGet the maximum value and the goal of the generator is to find the appropriate value for LS-LCAnd obtaining the maximum value, and finally enabling the network to reach a Nash equilibrium state, wherein the output probability distribution is 0.5 when the discriminator cannot distinguish whether the input data is real data. By introducing the bucket detection data class label as a condition variable, the defect that the initially generated countermeasure network is free and uncontrollable in the training process is overcome, and finally the generator can be controlled to generate the bucket detection data of the corresponding class through the bucket detection data class label. The discriminator can continuously learn the true sample in the process of confrontationThe data feature distribution judges whether the input sample is from a real sample or a generated sample and the corresponding type of the sample.
Referring to fig. 3, the present invention uses a set of random speckles of known sequence and targets of the type 10 "0123456789", the size of the speckles and targets being 28 × 28. After sampling for many times, an array of barrel detection signals is obtained and divided into a training set and a testing set, and training is carried out according to the method provided by the invention. Building a three-part model: generating a model, discriminating the model and a generator and discriminator confrontation training model. And reading pictures of the file data of the training set and the corresponding categories thereof in a loop, and simultaneously carrying out countermeasure training on the generated network and the judgment network. And performing a confrontation training operation once every batch of data is read, and storing the model after the training is finished. And loading the trained network model during testing, reading the pictures in the test set file, executing a discriminator, and judging the category of the test pictures to achieve the aim of identification. Through testing, the ten-time recognition rate of a single target is 100%.
The invention also provides an association identification system based on the generative countermeasure network, which comprises:
the generation type confrontation network training module is used for taking the category of an object to be identified as a target of network training, enabling a set of random speckle sequences to act on objects with the same type and different forms of the object to be identified, performing sampling for multiple times to form barrel detection signal arrays of the object, enabling different objects to correspond to different barrel detection signal arrays, taking the barrel detection signal arrays as samples of network training, and training by using a generation type confrontation network;
the barrel detection signal array generation module is used for detecting a target object by using a random speckle sequence used in the generation type confrontation network training process after the generation type confrontation network training is finished, collecting corresponding barrel detection signals and forming a corresponding barrel detection signal array after the sampling is finished for multiple times;
and the target object type judging module is used for inputting the bucket detection signal array of the target object into the trained generative confrontation network and outputting the target object type.
The invention also provides a terminal device, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor implements the steps of the association identification method based on the generative countermeasure network when executing the computer program.
The invention also proposes a computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method for identifying associations based on a generative confrontation network.
The computer program may be divided into one or more modules/units, which are stored in the memory and executed by the processor to perform the association identification method of the present invention.
The terminal can be a desktop computer, a notebook, a palm computer, a cloud server and other computing equipment, and can also be a processor and a memory. The processor may be a Central Processing Unit (CPU), other general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The memory may be used to store computer programs and/or modules, and the processor may implement various functions of the association identification system of the present invention by executing or executing the computer programs and/or modules stored in the memory, as well as by invoking data stored in the memory.
The above-mentioned embodiments are only preferred embodiments of the present invention, and are not intended to limit the technical solution of the present invention, and it should be understood by those skilled in the art that the technical solution can be modified and replaced by a plurality of simple modifications and replacements without departing from the spirit and principle of the present invention, and the modifications and replacements also fall into the protection scope covered by the claims.
Claims (10)
1. An association identification method based on a generative countermeasure network is characterized by comprising the following steps:
s1, taking the category of the object to be identified as the target of network training, enabling a set of random speckle sequences to act on the object with the same type and different form of the object to be identified, sampling for multiple times to form a barrel detection signal array of the object, enabling different objects to correspond to different barrel detection signal arrays, taking the barrel detection signal array as a sample of network training, and training by using a generative confrontation network;
s2, after the training is finished, detecting the target object by using the random speckle sequence used in the training process of the step S1, collecting corresponding barrel detection signals, and forming a corresponding barrel detection signal array after the sampling for many times is finished;
and S3, inputting the bucket detection signal array of the step S2 into the trained generative confrontation network, and outputting the target object type.
2. The association identification method based on the generative countermeasure network as claimed in claim 1, wherein the step S1 of applying a set of random speckle sequences to objects of the same type and different morphology of the object to be identified comprises the following steps: a set of random speckles of known sequence is used to illuminate different classes of objects in turn, and the total intensity of the light reflected by the objects is collected, i.e. a bucket detection signal array.
3. The association identification method based on the generative countermeasure network as claimed in claim 1, wherein:
the training of step S1 using the generative confrontation network includes the following steps:
1) sampling real data, obtaining a corresponding label y, transmitting the label y to a discriminator, and updating parameters according to an output result;
2) generating random noise, and transmitting the label y in the step 1) and the random noise into a generator to generate false data;
3) transmitting the false data generated in the step 2) and the label y into a discriminator;
4) the generator adjusts the parameters according to the output of the discriminator;
5) through successive cycles until the generator and arbiter reach nash equilibrium.
4. The association identification method based on the generative countermeasure network as claimed in claim 1, wherein:
step S3 trains the generative confrontation network according to the following steps:
1) preparing a training set and a testing set: according to a computed ghost imaging mechanism, a random speckle sequence is used for illuminating a target for multiple times, and an acquired barrel-shaped signal array is divided into a training set and a testing set;
2) establishing a network: based on the TensorFlow-gpu1.13, keras2.1.5 version, Pycharm was used to construct the following three models: generating a model, a discrimination model and a confrontation training model;
3) training a network: and circularly inputting the images in the training set and the corresponding categories thereof, and simultaneously training the generator and the discriminator model.
5. The association identification method based on the generative countermeasure network as claimed in claim 4, wherein: the bucket detection signal array of each category has a category label c of a corresponding target, and all the category labels c-PCAnd the normally distributed random number z is used as the input of the generator, and the output is XfakeG (c, z); the input of the discriminator is a real barrel detection data sample and corresponding class labels c-P thereofCTagged data X generated by the sum generatorfakeThe output of which is the true-false probability distribution X of the generated data samplerealD (c, x); the objective function of the generative countermeasure network includes two parts: log-likelihood L of real bucket detection samplesSAnd log-likelihood L of real bucket detection class labelsCThe distribution is represented by the following formula:
Ls＝E[log P(S＝real|Xreal)]+E[log P(S＝fake|Xfake)]
Lc＝E[log P(C＝c|Xreal)]+[E log P(C＝c|Xfake)]。
6. the association identification method based on the generative countermeasure network as claimed in claim 5, wherein: the generator is targeted to find the value such that LS+LCTaking the maximum value, the discriminator targeting the found value such that LS-LCThe maximum value is taken.
7. The association identification method based on the generative countermeasure network as claimed in claim 4, wherein: the output probability distribution is 0.5 when the discriminator cannot distinguish whether the input data is real data; the discriminator learns the characteristic distribution of the real sample data in the process of confronting the generator and judges whether the input sample is from the real sample or the generated sample and the corresponding type of the input sample.
8. An association identification system based on a generative confrontation network, comprising:
the generation type confrontation network training module is used for taking the category of an object to be identified as a target of network training, enabling a set of random speckle sequences to act on objects with the same type and different forms of the object to be identified, performing sampling for multiple times to form barrel detection signal arrays of the object, enabling different objects to correspond to different barrel detection signal arrays, taking the barrel detection signal arrays as samples of network training, and training by using a generation type confrontation network;
the barrel detection signal array generation module is used for detecting a target object by using a random speckle sequence used in the generation type confrontation network training process after the generation type confrontation network training is finished, collecting corresponding barrel detection signals and forming a corresponding barrel detection signal array after the sampling is finished for multiple times;
and the target object type judging module is used for inputting the bucket detection signal array of the target object into the trained generative confrontation network and outputting the target object type.
9. A terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: the processor, when executing the computer program, implements the steps of the association identification method based on the generative countermeasure network according to any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that: the computer program, when being executed by a processor, implements the steps of the method for identifying associations based on a generative countermeasure network according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110541068.0A CN113269243A (en) | 2021-05-18 | 2021-05-18 | Association identification method and device based on generative countermeasure network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110541068.0A CN113269243A (en) | 2021-05-18 | 2021-05-18 | Association identification method and device based on generative countermeasure network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113269243A true CN113269243A (en) | 2021-08-17 |
Family
ID=77231701
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110541068.0A Pending CN113269243A (en) | 2021-05-18 | 2021-05-18 | Association identification method and device based on generative countermeasure network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113269243A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114359667A (en) * | 2021-12-30 | 2022-04-15 | 西安交通大学 | Strength coherent identification method and equipment based on generating type countermeasure network |
| CN114494055A (en) * | 2022-01-19 | 2022-05-13 | 西安交通大学 | Ghost imaging method, system, equipment and storage medium based on circulating neural network |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111027439A (en) * | 2019-12-03 | 2020-04-17 | 西北工业大学 | SAR target recognition method for generating countermeasure network based on auxiliary classification |
| US20210073630A1 (en) * | 2019-09-10 | 2021-03-11 | Robert Bosch Gmbh | Training a class-conditional generative adversarial network |
-
2021
- 2021-05-18 CN CN202110541068.0A patent/CN113269243A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210073630A1 (en) * | 2019-09-10 | 2021-03-11 | Robert Bosch Gmbh | Training a class-conditional generative adversarial network |
| CN111027439A (en) * | 2019-12-03 | 2020-04-17 | 西北工业大学 | SAR target recognition method for generating countermeasure network based on auxiliary classification |
Non-Patent Citations (2)
| Title |
|---|
| YUCHEN HE, YIBING CHEN: "《Generative-Adversarial-Networks-based Ghost Recognition》", 《ARXIV》, pages 1 - 4 * |
| 易珂;: "基于深度网络ACGAN的图像识别", 电子技术与软件工程, no. 18 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114359667A (en) * | 2021-12-30 | 2022-04-15 | 西安交通大学 | Strength coherent identification method and equipment based on generating type countermeasure network |
| CN114359667B (en) * | 2021-12-30 | 2024-01-30 | 西安交通大学 | Intensity coherent identification method and equipment based on generation type countermeasure network |
| CN114494055A (en) * | 2022-01-19 | 2022-05-13 | 西安交通大学 | Ghost imaging method, system, equipment and storage medium based on circulating neural network |
| CN114494055B (en) * | 2022-01-19 | 2024-02-06 | 西安交通大学 | Ghost imaging method, ghost imaging system, ghost imaging equipment and ghost imaging storage medium based on cyclic neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bendale et al. | Towards open set deep networks | |
| CN109583342B (en) | Human face living body detection method based on transfer learning | |
| Russakovsky et al. | Detecting avocados to zucchinis: what have we done, and where are we going? | |
| CN111709311B (en) | Pedestrian re-identification method based on multi-scale convolution feature fusion | |
| Wang et al. | Two-stage method based on triplet margin loss for pig face recognition | |
| CN110363220B (en) | Behavior class detection method and device, electronic equipment and computer readable medium | |
| Prasetyo et al. | A comparison of yolo and mask r-cnn for segmenting head and tail of fish | |
| CN113269243A (en) | Association identification method and device based on generative countermeasure network | |
| CN103593648B (en) | Face recognition method for open environment | |
| Talas et al. | CamoGAN: evolving optimum camouflage with Generative Adversarial Networks | |
| CN113435355A (en) | Multi-target cow identity identification method and system | |
| Chin et al. | Dorsal hand vein authentication system using artificial neural network | |
| CN111965639A (en) | Radar signal analysis method based on bidirectional long-and-short-term memory neural network | |
| Ha et al. | Fine-grained interactive attention learning for semi-supervised white blood cell classification | |
| Vaidya et al. | Leveraging YOLOv7 for plant disease detection | |
| Wang et al. | Dog breed classification based on deep learning | |
| CN112949469A (en) | Image recognition method, system and equipment for face tampered image characteristic distribution | |
| Soysal et al. | An introduction to zero-shot learning: An essential review | |
| Zhu et al. | A novel simple visual tracking algorithm based on hashing and deep learning | |
| Gong et al. | An analysis of plant diseases identification based on deep learning methods | |
| Li et al. | A study on IR target recognition approach in aerial jamming environment based on Bayesian probabilistic model | |
| Masum et al. | YOLOv5 vs. YOLOv8 in Marine Fisheries: Balancing Class Detection and Instance Count | |
| CN111797690A (en) | Optical fiber perimeter intrusion identification method and device based on wavelet neural network grating array | |
| Wang et al. | Design of Intelligent Detection System for Smoking Based on Improved YOLOv4. | |
| CN114581819A (en) | Video behavior identification method 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 |