CN114998960A - Expression recognition method based on positive and negative sample comparison learning - Google Patents
Expression recognition method based on positive and negative sample comparison learning Download PDFInfo
- Publication number
- CN114998960A CN114998960A CN202210595007.7A CN202210595007A CN114998960A CN 114998960 A CN114998960 A CN 114998960A CN 202210595007 A CN202210595007 A CN 202210595007A CN 114998960 A CN114998960 A CN 114998960A
- Authority
- CN
- China
- Prior art keywords
- sample
- samples
- positive
- negative
- representing
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 230000001815 facial effect Effects 0.000 claims abstract description 13
- 238000010801 machine learning Methods 0.000 claims abstract description 9
- 230000006870 function Effects 0.000 claims description 18
- 238000004364 calculation method Methods 0.000 claims description 11
- 238000013528 artificial neural network Methods 0.000 claims description 9
- 241000282414 Homo sapiens Species 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 2
- 230000008569 process Effects 0.000 abstract description 2
- 230000008921 facial expression Effects 0.000 description 3
- 230000015654 memory Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- LFQSCWFLJHTTHZ-UHFFFAOYSA-N Ethanol Chemical compound CCO LFQSCWFLJHTTHZ-UHFFFAOYSA-N 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000008451 emotion Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/174—Facial expression recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
Abstract
The invention discloses an expression recognition method based on positive and negative sample comparison learning, S1, collecting facial images; s2, inputting the facial image into a trained machine learning model to identify the expression in the facial image; s3, outputting the expression type of the facial image; in the process of machine learning model, a negative sample is introduced in the structural similarity contrast learning method, and the distance between the positive sample and the negative sample in a batch of samples is increased. Meanwhile, in a batch of samples, the imbalance between the quantities of the positive samples and the negative samples is considered, the negative samples are subjected to difficult sample discovery, so that the model increases and restricts the structural similarity of the positive samples, reduces and restricts the negative samples most similar to the positive samples, further exerts the effect of contrast learning, and improves the accuracy of expression recognition.
Description
Technical Field
The invention relates to the technical field of expression recognition, in particular to an expression recognition method based on positive and negative sample comparison learning.
Background
The facial expression recognition technology is used for recognizing the emotion of a person in an image through a computer, and has important practical application values including aspects of intelligent medical treatment, safe driving detection, online education, psychological coaching, entertainment interaction and the like. However, in view of the expression recognition accuracy in an actual scene, the accuracy of facial expression recognition at present has not reached the level that human beings can reach.
The expression recognition task in natural scenes is more challenging than that in controlled experimental scenes because the captured images in natural conditions contain more differences in environmental factors, such as lighting, resolution and more extensive pose changes. Secondly, because the acquisition conditions are not uniform, the expression recognition task in the natural scene shows a larger problem of large difference in similarity between classes, for example, the expression is also a 'surprised' expression, and the appearance shows a large difference because of the difference of the shooting angle, the illumination condition and the expression mode of the character. Also, images from different classes have similar appearances, showing large inter-class similarity. Another example is the category "happy", which is the same category for smiling and laughing due to different acquisition conditions, but their images have large differences in appearance, while for the two categories "sick" and "fear", the differences in appearance are small. Finally, the expression data set collected in the natural scene usually needs a labeling person to manually label, this action needs to consume a large amount of manpower, and more importantly, different people may make different judgments for the same image due to the influence of human subjectivity, so that the label of the expression has a large ambiguity.
The key of the human face expression recognition task in a natural scene is to extract features with expression significances, and a plurality of methods for learning and optimizing local features and a method based on metric learning have good effects. Although many methods have been previously proposed to address the problems of inter-class similarity, large intra-class variance and label ambiguity, they are all trained in a traditional supervised manner. In a natural scene, the problem of label ambiguity of the expression recognition task is not negligible, the original label is simply utilized, wrong information can be introduced, the ambiguous label can mislead the network model to learn the expression characteristics, and therefore supervised learning cannot completely dig out fine expression distinguishing characteristics.
Therefore, how to provide an expression recognition method based on positive and negative sample comparison learning, which can effectively improve the expression recognition accuracy rate, is a problem that needs to be solved urgently by those skilled in the art.
Disclosure of Invention
In view of this, the invention provides an expression recognition method based on positive and negative sample comparison learning, and aims to provide a more accurate expression recognition method.
In order to achieve the purpose, the invention adopts the following technical scheme:
an expression recognition method based on positive and negative sample comparison learning comprises the following steps:
s1, collecting a face image;
s2, inputting the facial image into a trained machine learning model to identify the expression in the facial image;
s3, outputting the expression type of the facial image;
the training method of the machine learning model in the S2 specifically includes:
s21, generating a plurality of new samples for each sample;
for a batch of samples
After strong data enhancement
After weak data enhancement
Wherein x is i As an image of a human face, y i Is x i A corresponding emoticon label; obtaining a current weak enhancement sample x ″ i And its strong enhancement sample x ″ i As a positive sample pair; will be associated with the current sample x ″ i The enhanced samples of all other samples in different classes are taken as negative sample pairs, and the negative sample pair set of the ith sample is expressed as
S22, inputting a deep neural network, and extracting the characteristics of each sample;
for weak enhancement sample x ″ i Sum strong enhancement sample x ″ i After feature extraction, respectively obtaining feature representations: u ═ U i |i＝1,2,……HW},V＝{v j 1,2, … … HW }, where u i ,v j As feature vector representations of the feed point and the destination point, respectively;
s23, calculating a loss function of the deep neural network;
defining the negative sample similarity set of the ith sample as
The aggregate size is K i ；
The total loss function is:
L＝L cls +βL hard
wherein β is a hyperparameter representing an equilibrium coefficient; l is cls Is a softmax classification loss function, L hard Adaptive weighting function for negative samples:
wherein N is the number of samples in the batch;
for the structure similarity transformation probability, gamma is the scaling factor, s i,j Representing the structural similarity between the ith sample and the jth enhanced sample; the sum of the probabilities for all samples is 1;
and S24, optimizing parameters of the deep neural network according to the loss function.
Preferably, in S23, the structural similarity S between the ith feature representation and the jth enhanced feature i,j The specific calculation method comprises the following steps:
s i,j ＝-s(U i ,V j )
where H and W are the height and width of convolution signatures U and V, respectively, c ij Representing the transmission cost, U, between the source supply-side node i to the destination node j i Convolution feature map, V, representing the ith sample j A convolution signature representing the jth sample,
the optimal transportation scheme is represented, and the optimal transportation scheme is obtained by calculating the following formula:
subject to f ij ≥0,i＝1,…,m,j＝1,…,k
wherein, the first and the second end of the pipe are connected with each other,
the local weights representing the source node and the destination node are calculated by the following formula:
wherein G is avg Representing a global average pooling operation.
Preferably, c ij The calculation method is as follows:
wherein u is i ，v j Are the feature vector representations of the feed point and the destination point.
Preferably, in S23, the relative magnitude p of the structural similarity between the ith sample and the kth enhanced sample i,k The specific calculation method comprises the following steps:
when p is i,k The larger the value, the greater the learning weight given to the corresponding partial sample.
Compared with the prior art, the invention discloses and provides an expression recognition method based on positive and negative sample comparison learning, and the method has the following beneficial effects:
1. according to the method, the EMD distance is acted on the feature map, the model can be guided to pay attention to the region related to the expression, the regions which are irrelevant to expression such as a noisy background are eliminated, and therefore the expression region is paid attention to effectively.
2. The method adopts the self-supervision idea to design the structural similarity constraint loss, and utilizes the contrast learning of positive and negative samples of the enhanced image to optimize the positive and negative samples and the classification loss under the condition of not depending on an original label, so as to learn the more generalized expression characteristics.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a schematic flow chart of an expression recognition method based on positive and negative sample comparison learning according to the present invention;
FIG. 2 is a schematic diagram of a training method of a machine learning model in an expression recognition method based on positive and negative sample comparison learning according to the present invention;
fig. 3 is a schematic diagram of adaptive weighting of negative samples in the expression recognition method based on positive and negative sample comparison learning according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention discloses an expression recognition method based on positive and negative sample comparison learning, which comprises the following steps:
s1, collecting a face image;
s2, inputting the facial image into a trained machine learning model to identify the expression in the facial image;
s3, outputting the expression type of the facial image;
the training method of the machine learning model in the S2 specifically comprises the following steps:
s21, generating a plurality of new samples for each sample;
for a batch of samples
After strong data enhancement
After weak data enhancement
Wherein x is i As an image of a human face, y i Is x i A corresponding emoticon label; obtaining a current weak enhancement sample x ″ i And its strongly enhanced sample x i As a positive sample pair; will be associated with the current sample x ″ i The enhanced samples of all other samples in different classes are taken as negative sample pairs, and the negative sample pair set of the ith sample is expressed as
Wherein images of two different viewing angles are obtained by weak enhancement and strong enhancement. The weak enhancement means that the image is zoomed in and flipped to a small degree, and the strong enhancement means that the data enhancement mode also comprises rotation and color transformation.
S22, inputting a deep neural network, and extracting the characteristics of each sample;
for weak enhancement sample x ″) i And a strongly enhanced sample x i After feature extraction, respectively obtaining feature representations: u ═ U i |i＝1,2,……HW},V＝{v j 1,2, … … HW }, where u i ,v j As feature vector representations of the feed point and the destination point, respectively;
s23, calculating a loss function of the deep neural network;
defining the negative sample similarity set of the ith sample as
The aggregate size is K i ；
The total loss function is:
L＝L cls +βL hard
wherein β is a hyperparameter representing an equilibrium coefficient; l is a radical of an alcohol cls Is a softmax classification loss function, L hard Adaptive weighting function for negative samples:
wherein N is the number of samples in the batch;
the probability is converted for the structural similarity, gamma is a scaling coefficient, and gamma influences the balance degree between the sample classification probabilities. The larger the value of gamma, the larger the difference in classification probability, and the smaller gamma, the closer the classification probability among samples. When the gamma value region is at zero, the probability distribution degenerates to a uniform distribution; s i,j Representing the structural similarity between the ith sample and the jth enhanced sample; the sum of the probabilities for all samples is 1; wherein s is i,j Representing the structural similarity between the ith sample and the jth enhanced sample.
And S24, optimizing parameters of the deep neural network according to the loss function.
In order to further implement the above technical solution, in S23, the structural similarity S between the ith feature representation and the jth enhanced feature i,j The specific calculation method comprises the following steps:
s i,j ＝-s(U i ,V j )
wherein H and W are each a convolutionHeight and width of feature maps U and V, c ij Representing the transmission cost, U, between the source supply-side node i to the destination node j i Convolution feature map, V, representing the ith sample j A convolution signature representing the jth sample,
the optimal transportation scheme is represented, and the optimal transportation scheme is obtained by calculating the following formula:
subject to f ij ≥0,i＝1,…,m,j＝1,…,k
wherein, the first and the second end of the pipe are connected with each other,
the local weights representing the source node and the destination node are calculated by the following formula:
wherein G is avg Representing a global average pooling operation.
To further implement the above solution, c ij The calculation method is as follows:
wherein u is i ，v j Are the feature vector representations of the feed point and the destination point.
To further implement the above technical solution, in S23, the relative size p of the structural similarity between the ith sample and the kth enhanced sample i,k The specific calculation method comprises the following steps:
when p is i,k The larger the value, the greater the learning weight given to the corresponding partial sample.
p i,k The relative magnitude of the structural similarity between the ith sample and the kth enhanced sample is shown, reflecting the degree of similarity of the two view angle features. p is a radical of i,k The larger the value, the more similar the sample pair, the more difficult the enhancement sample is to distinguish, and the model should give greater learning weight to this portion of the sample.
It should be noted that:
for sample x i The positive sample is defined as the enhancement sample x ″ i . The negative sample should maintain a larger structureSimilarity differences exist, samples originally labeled as the same class may exist in a batch, the samples may have higher structural similarity, and more error information is introduced when the enhanced samples of the same class are regarded as negative samples. For correctness, the invention considers the negative sample as the enhanced sample of other classes, and the set is defined as
I.e. enhancement samples of all other samples of a different class than the current sample. This guarantees to some extent the correctness of the negative examples.
The loss function is calculated as follows, taking into account the comparative learning of negative examples, and adding adaptive weights to the difficult negative examples.
The difficulty of the sample is a measure of the problem. And regarding each sample as a separate category, and converting each negative sample into a category probability through a softamx function according to the structural similarity score.
In this embodiment, β is a balance coefficient, which needs to be adjusted as a hyper-parameter, and β takes a value of 0.6 in this embodiment.
The deep neural network model of the present embodiment is run on an Nvid ia T itan 3090GPU server that mounts a Pytorch (v1.7) deep learning framework. This embodiment scales all images to 256 x 256 size for a ResNet-18 based backbone network. In the training stage, the input image is cut to 224 × 224 size at random position, and in the testing stage, the center position is cut to 224 × 224 size for testing. There is weak enhancement and strong enhancement in the data enhancement stage: when the image is weakly enhanced, the image is turned over with a probability of 0.5; when the enhancement is strong, color random dithering is performed with a probability of 0.2 in addition to the weak enhancement.
The video memory consumption in the structural similarity calculation process is considered to be too large, and particularly in the step of calculating the structural similarity between every two video memories in the batch, the video memory occupation ratio is large, so that the batch size is set to be 32. The model is optimized by using a random gradient descent method with momentum, the learning rate is 0.01, the momentum is 0.9, and the weight decay rate is 0.0001. The EMD distance calculation module needs to calculate the corresponding optimal transportation solution, which is calculated by the OpenCV library function.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. The device disclosed by the embodiment corresponds to the method disclosed by the embodiment, so that the description is simple, and the relevant points can be referred to the method part for description.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (4)
1. An expression recognition method based on positive and negative sample comparison learning is characterized by comprising the following steps:
s1, collecting a face image;
s2, inputting the facial image into a trained machine learning model to identify the expression in the facial image;
s3, outputting the expression type of the facial image;
the training method of the machine learning model in the S2 specifically includes:
s21, generating a plurality of new samples for each sample;
for a batch of samples
After strong data enhancement
After weak data enhancement
Wherein x is i As an image of a human face, y i Is x i A corresponding emoticon label; obtaining a current weak enhancement sample x ″ i And its strong enhancement sample x ″ i As a positive sample pair; will be associated with the current sample x ″ i The enhanced samples of all other samples in different classes are taken as negative sample pairs, and the set of negative sample pairs of the ith sample is expressed as
S22, inputting a deep neural network, and extracting the characteristics of each sample;
for weak enhancement sample x ″) i And a strongly enhanced sample x ″ i After feature extraction, respectively obtaining feature representations: u ═ U i |i＝1,2,……HW},V＝{v j 1,2, … … HW }, where u i ,v j As feature vector representations of a supply point and a destination point, respectively;
s23, calculating a loss function of the deep neural network;
defining the negative sample similarity set of the ith sample as
The aggregate size is K i ；
The total loss function is:
L＝L cls +βL hard
wherein β is a hyperparameter representing an equilibrium coefficient; l is cls Is a softmax classification loss function, L hard Adaptive weighting function for negative samples:
wherein N is the number of samples in the batch;
for the structure similarity transformation probability, gamma is the scaling factor, s i,j Representing the structural similarity between the ith sample and the jth enhanced sample; the sum of the probabilities for all samples is 1;
and S24, optimizing parameters of the deep neural network according to the loss function.
2. The expression recognition method based on positive-negative sample contrast learning of claim 1, wherein in S23, the structural similarity S between the ith feature representation and the jth enhanced feature i,j The specific calculation method comprises the following steps:
s i,j ＝-s(U i ,V j )
where H and W are the height and width of convolution signatures U and V, respectively, c ij Representing the transmission cost, U, between the source supply-side node i to the destination node j i Convolution feature map, V, representing the ith sample j A convolution signature graph representing the jth sample,
the optimal transportation scheme is represented, and the optimal transportation scheme is obtained by calculating the following formula:
subject to f ij ≥0,i＝1,…,m,j＝1,…,k
wherein the content of the first and second substances,
the local weights representing the source node and the destination node are calculated by the following formula:
wherein G avg Representing a global average pooling operation.
3. The expression recognition method based on positive and negative sample comparison learning as claimed in claim 2, wherein c is ij The calculation method is as follows:
wherein u i ，v j Are the feature vector representations of the feed point and the destination point.
4. The expression recognition method based on positive-negative sample contrast learning of claim 2, wherein in S23, the relative size p of the structural similarity between the ith sample and the kth enhanced sample i,k The specific calculation method comprises the following steps:
when p is i,k The larger the value, the greater the learning weight given to the corresponding partial sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210595007.7A CN114998960B (en) | 2022-05-28 | 2022-05-28 | Expression recognition method based on positive and negative sample contrast learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210595007.7A CN114998960B (en) | 2022-05-28 | 2022-05-28 | Expression recognition method based on positive and negative sample contrast learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114998960A true CN114998960A (en) | 2022-09-02 |
| CN114998960B CN114998960B (en) | 2024-03-26 |
Family
ID=83028670
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210595007.7A Active CN114998960B (en) | 2022-05-28 | 2022-05-28 | Expression recognition method based on positive and negative sample contrast learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114998960B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117079336A (en) * | 2023-10-16 | 2023-11-17 | 腾讯科技（深圳）有限公司 | Training method, device, equipment and storage medium for sample classification model |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110851645A (en) * | 2019-11-08 | 2020-02-28 | 吉林大学 | Image retrieval method based on similarity maintenance under depth metric learning |
| CN110866134A (en) * | 2019-11-08 | 2020-03-06 | 吉林大学 | Image retrieval-oriented distribution consistency keeping metric learning method |
| CN111723675A (en) * | 2020-05-26 | 2020-09-29 | 河海大学 | Remote sensing image scene classification method based on multiple similarity measurement deep learning |
| US20210042580A1 (en) * | 2018-10-10 | 2021-02-11 | Tencent Technology (Shenzhen) Company Limited | Model training method and apparatus for image recognition, network device, and storage medium |
| US20210390355A1 (en) * | 2020-06-13 | 2021-12-16 | Zhejiang University | Image classification method based on reliable weighted optimal transport (rwot) |
-
2022
- 2022-05-28 CN CN202210595007.7A patent/CN114998960B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210042580A1 (en) * | 2018-10-10 | 2021-02-11 | Tencent Technology (Shenzhen) Company Limited | Model training method and apparatus for image recognition, network device, and storage medium |
| CN110851645A (en) * | 2019-11-08 | 2020-02-28 | 吉林大学 | Image retrieval method based on similarity maintenance under depth metric learning |
| CN110866134A (en) * | 2019-11-08 | 2020-03-06 | 吉林大学 | Image retrieval-oriented distribution consistency keeping metric learning method |
| CN111723675A (en) * | 2020-05-26 | 2020-09-29 | 河海大学 | Remote sensing image scene classification method based on multiple similarity measurement deep learning |
| US20210390355A1 (en) * | 2020-06-13 | 2021-12-16 | Zhejiang University | Image classification method based on reliable weighted optimal transport (rwot) |
Non-Patent Citations (1)
| Title |
|---|
| 王博威;潘宗序;胡玉新;马闻;: "少量样本下基于孪生CNN的SAR目标识别", 雷达科学与技术, no. 06, 15 December 2019 (2019-12-15), pages 17 - 23 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117079336A (en) * | 2023-10-16 | 2023-11-17 | 腾讯科技（深圳）有限公司 | Training method, device, equipment and storage medium for sample classification model |
| CN117079336B (en) * | 2023-10-16 | 2023-12-22 | 腾讯科技（深圳）有限公司 | Training method, device, equipment and storage medium for sample classification model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114998960B (en) | 2024-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Mao et al. | Explain images with multimodal recurrent neural networks | |
| CN108984530A (en) | A kind of detection method and detection system of network sensitive content | |
| CN111460980B (en) | Multi-scale detection method for small-target pedestrian based on multi-semantic feature fusion | |
| CN109214001A (en) | A kind of semantic matching system of Chinese and method | |
| CN109242400A (en) | A kind of logistics express delivery odd numbers recognition methods based on convolution gating cycle neural network | |
| CN109710804B (en) | Teaching video image knowledge point dimension reduction analysis method | |
| CN110188708A (en) | A kind of facial expression recognizing method based on convolutional neural networks | |
| CN109344759A (en) | A kind of relatives' recognition methods based on angle loss neural network | |
| CN110263174B (en) | Topic category analysis method based on focus attention | |
| CN111414862A (en) | Expression recognition method based on neural network fusion key point angle change | |
| Yang et al. | Handwriting text recognition based on faster R-CNN | |
| CN110287952A (en) | A kind of recognition methods and system for tieing up sonagram piece character | |
| CN108345833A (en) | The recognition methods of mathematical formulae and system and computer equipment | |
| CN115457568B (en) | Historical document image noise reduction method and system based on generation countermeasure network | |
| CN114580362B (en) | System and method for generating return mark file | |
| CN112883931A (en) | Real-time true and false motion judgment method based on long and short term memory network | |
| CN114998960A (en) | Expression recognition method based on positive and negative sample comparison learning | |
| CN112434686B (en) | End-to-end misplaced text classification identifier for OCR (optical character) pictures | |
| CN116543269B (en) | Cross-domain small sample fine granularity image recognition method based on self-supervision and model thereof | |
| CN111931630B (en) | Dynamic expression recognition method based on facial feature point data enhancement | |
| CN116579348A (en) | False news detection method and system based on uncertain semantic fusion | |
| Oktaviani et al. | Optical character recognition for hangul character using artificial neural network | |
| CN115982652A (en) | Cross-modal emotion analysis method based on attention network | |
| CN115588220A (en) | Two-stage multi-scale self-adaptive low-resolution face recognition method and application | |
| CN115564988A (en) | Remote sensing image scene classification and semantic segmentation task method based on label smoothing |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |