CN109376692A - Migration convolution neural network method towards facial expression recognition - Google Patents
Migration convolution neural network method towards facial expression recognition Download PDFInfo
- Publication number
- CN109376692A CN109376692A CN201811398259.0A CN201811398259A CN109376692A CN 109376692 A CN109376692 A CN 109376692A CN 201811398259 A CN201811398259 A CN 201811398259A CN 109376692 A CN109376692 A CN 109376692A
- Authority
- CN
- China
- Prior art keywords
- migration
- facial expression
- neural network
- convolution neural
- data set
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- 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
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses the migration convolution neural network methods towards facial expression recognition, including following procedure: S1, obtains Facial Expression Image data set, is divided into training set, verifying collection and test set；S2, by migration network and convolutional neural networks cascade building migration convolution neural network model；The input for migrating convolution network model is Facial Expression Image data, is exported as human face expression classification；S3 is trained migration convolution neural network model using training set, and optimizes trained migration convolution neural network model using verifying collection；S4, by the migration convolution neural network model after optimization, the accuracy rate for carrying out facial expression recognition to test set is tested.The present invention solves the problems, such as that small data set can not restrain and over-fitting on a large amount of networks using migration convolution neural network recognization human face expression.
Description
Technical field
The present invention relates to image identification technical fields, and in particular to a kind of migration convolutional Neural towards facial expression recognition
Network method.
Background technique
Facial expression recognition is the important component in human-computer interaction and affection computation research.With the hair of artificial intelligence
The maturation of exhibition and robot building system, field of human-computer interaction show huge market and application prospect.
Traditional facial expression recognition research method is based primarily upon geometrical characteristic, to the change in location such as the eyes, eyebrow, mouth of people spy
Sign carries out Expression Recognition.It requires that feature is manually set, the information content for extracting feature is quite limited to, and accuracy rate is difficult to reach application
It is required that.
With the development of high-performance server, it is widely applied to by the deep learning algorithm of representative of convolutional neural networks
In the fields such as computer vision, automatic Pilot, good effect is achieved.Expression recognition method based on convolutional neural networks is logical
Data-driven is crossed, building convolutional layer learns to extract abstract characteristic information from expression data library, finally using full articulamentum point
Class.However, depth convolutional network depends on data-driven, the static faces expression picture in many data sets is very little, can not
When obtaining large-scale dataset up to a million, direct training effect is not ideal enough, is easy to produce over-fitting, the extensive effect of identification compared with
Difference.
Summary of the invention
In order to overcome the shortcomings of in the prior art, the present invention provides a kind of, and the migration convolution towards facial expression recognition is refreshing
Through network method, using migration convolution neural network recognization human face expression, solving small data set can not receive on a large amount of networks
It holds back and the problem of over-fitting.
In order to solve the above-mentioned technical problems, the present invention provides a kind of migration convolution nerve net towards facial expression recognition
Network method, characterized in that the following steps are included:
S1 obtains Facial Expression Image data set, is divided into training set, verifying collection and test set；
S2, by migration network and convolutional neural networks cascade building migration convolution neural network model；Migrate convolution network model
Input be Facial Expression Image data, export as human face expression classification；
S3 is trained migration convolution neural network model using training set, and optimizes trained move using verifying collection
Move convolutional neural networks model；
S4, by the migration convolution neural network model after optimization, the accuracy rate for carrying out facial expression recognition to test set is tested.
Further, in S1, obtaining Facial Expression Image data set includes following procedure:
S11 obtains CK+ and FER2013 Facial Expression Image data set；
S12 extends CK+ data set；
S13 carries out image normalization to CK+ and FER2013 data set.
Further, migration network is Inception_v3 network.
Further, migration network includes 6 convolutional layers, 3 Inception layers, 2 pond layers and 1 full articulamentum,
Full articulamentum is removed in cascade.
Further, convolutional neural networks include 1 convolutional layer, 1 pond layer and a full articulamentum.
Further, human face expression classification includes anger, detests, is frightened, glad, sad, surprised and neutral.
The beneficial effect comprise that
1), the present invention has carried out image procossing to original data set, on the one hand extends the quantity of facial expression image data set, another party
Face converts the 2-D gray image of 48x48x1 to using opencv the three-dimension high-resolution image of 229x229x3, greatly reduces
Expression information improves the effect of Expression Recognition a possibility that migrating the loss in network indirectly.
2), the present invention combines transfer learning method and convolutional neural networks, and cascade constructs the migration convolution mind
Through network, solves the problems, such as that small data set can not restrain and over-fitting on a large amount of networks, be deep learning method in decimal
Method is provided according to the practical application in collection field.
3), the present invention devises the operation of convolution sum pondization after migrating network, plays and extracts expressive features information and pick
Except the effect of redundancy, the precision of recognition of face is improved.
4) mechanism to be decayed in the network that the present invention constructs using Relu activation primitive and learning rate, has further prevented mistake
The generation of fitting phenomenon, migration the extensive of neural network work well.
Detailed description of the invention
Fig. 1 is the flow chart of the method for the present invention；
Fig. 2 is the network structure for migrating convolutional neural networks.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.Following embodiment is only used for clearly illustrating the present invention
Technical solution, and not intended to limit the protection scope of the present invention.
A kind of migration convolution neural network method towards facial expression recognition of the invention, it is shown in Figure 1, including with
Lower process:
S1 obtains Facial Expression Image data set, is divided into training set, test set and verifying and collects.
Facial Expression Image data set, line data set of going forward side by side pretreatment, tool are obtained from existing Facial expression database
Body includes following procedure:
S11 obtains CK+ and FER2013 Facial Expression Image data set.
FER2013 data set is collected from data science contest kaggle, the expression library by training set, test set and verifying
Three parts are constituted, and wherein training set includes the gray level image of 28,709 48x48 altogether, and human face expression is divided into angry, detest, is feared
Fear, is glad, sad, is surprised and this 7 kinds neutral, and CK+ data set is expanded by Cohn Kanade expression library, image kind
Class and size are all the same, include totally 988 human face expressions.
S12 extends CK+ data set.
Image in the lesser CK+ data set of picture number is subjected to 5 random croppings and mirror surface symmetry operation, is realized
The effect that ten times of data set extension.
S13 carries out image normalization to CK+ and FER2013 data set.
Gray level image in the data set of CK+ and FER2013 is normalized, using under tensorflow frame
Opencv resize method by 42x42 the and FER2013 data set in the CK+ data set of cutting 48x48 two dimension ash
Degree figure is converted into the three-dimensional color image of 229x229x3.
Collect training set, test set and verifying is randomly divided into the ratio of 8:1:1 through step S1 treated image.
S2 migrates network and convolutional neural networks cascade building migration convolution neural network model；Migrate convolution nerve net
The input of network model is Facial Expression Image data, is exported as human face expression classification.
The migration network selected is Inception_v3 network trained on Image-Net data set, and the network is former
Begin to solve is more object classification problems on the Image-Net data set for having tens million of pictures.The network includes 6 convolution
Layer, 3 Inception layers, 2 pond layers and 1 full articulamentum remove full articulamentum in cascade, at this time input data ruler
Very little is 299x299x3, Output Size 8x8x2048.
It, will the trained Inception_v3 network migration on Image-Net data set using the thought of transfer learning
To facial expression recognition task, is cascaded with designed convolutional neural networks and constitute migration convolutional neural networks.
Wherein convolutional neural networks carry out the extraction of information including one layer of convolutional layer and pond layer and screen out and one layer
Full articulamentum classifies to expression, and input data is having a size of 8x8x2048, Output Size 7x1.The convolution of the convolutional layer
Core is dimensioned to 3x3, and step-length 1, being filled with is 0；The convolution kernel of the pond layer is dimensioned to 2x2, and step-length is
2, being filled with is 0；The convolution kernel size of articulamentum is 1, and dimension is set as 7 dimensions.
S3, preset migration convolutional neural networks are trained using training set, and optimize training using verifying collection
Good migration convolution neural network model.
The migration convolutional neural networks are trained using training set, then pass through the study situation and verifying of training
Performance on collection is adjusted the hyper parameter in network, obtains optimal network model；
Collect training set, test set and verifying is randomly divided into the ratio of 8:1:1 through step S1 treated image.It will train again
The method of collection data batch processing is trained in batches in the migration convolutional network described in step S2, the setting of batch processing primary quantity
It is 64, iterates 10000 times.Wherein, when training every time in batches, it is assumed that a figure in the batch data sample of input is
(X, Y), X are that the three-dimensional matrice of input picture indicates, Y is one of known seven kinds of expression labels, and seven kinds of expressions are anger, detest
It dislikes, is frightened, glad, sad, surprised and neutral.The dimension of X is 229x299x3, and the dimension of Y is 7x1.X is fed for parameter first
Fixed transfer learning network portion, the eigenmatrix of 8x8x2048 of the output comprising dimensional images feature, the eigenmatrix
Be re-used as the input of convolutional neural networks part, finally export 7x1 size predicted vector, network query function go out the vector with it is described
Loss function between label Y, and the weight of convolutional neural networks part is adjusted reversely to optimize by stochastic gradient descent method
Network parameter.
The network completed according to training showing come the hyper parameter in further regulating networks on verifying collection, it is main to adjust
The regularization coefficients such as the number of iterations, batch processing amount, the attenuation rate of loss function, activation primitive, learning rate and learning rate.
In the near-optimization network structure that training is completed, the number of iterations is 30000 times, and batch processing amount is 256, loss function
For Softmax, activation primitive is Relu function, and learning rate is set as 0.001, and the attenuation rate of learning rate is 0.01.
S4, facial expression recognition test: by the optimal network model, the Facial Expression Image of test set is carried out
Accuracy rate test.
It is using Facial Expression Image in the test set of the CK+ handled well and FER2013 as input sample, the sample is defeated
Enter and carry out facial expression recognition in the migration convolutional neural networks of trained near-optimization, the results show that the migration convolution
Accuracy rate of the neural network on CK+ data set is 99.6%, and the accuracy rate on FER2013 data set is 87.5%, is higher than it
He compares discrimination of seven kinds of methods on FER2013 as shown in table 1 at recognition methods.
The accuracy rate of 1 existing method of table and the method for the present invention on FER2013 data set
| Method | Discrimination/% |
| SVM | 64.78 |
| Gabor wavelet transformation | 67.04% |
| DBN | 69.77 |
| N1 | 76.48 |
| N2 | 73.92 |
| DLCNN | 80.77 |
| The method of the present invention | 87.50 |
The above is only a preferred embodiment of the present invention, it is noted that those skilled in the art are come
It says, without departing from the technical principles of the invention, several improvements and modifications can also be made, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (6)
1. the migration convolution neural network method towards facial expression recognition, characterized in that including following procedure:
S1 obtains Facial Expression Image data set, is divided into training set, verifying collection and test set；
S2, by migration network and convolutional neural networks cascade building migration convolution neural network model；Migrate convolution network model
Input be Facial Expression Image data, export as human face expression classification；
S3 is trained migration convolution neural network model using training set, and optimizes trained move using verifying collection
Move convolutional neural networks model；
S4, by the migration convolution neural network model after optimization, the accuracy rate for carrying out facial expression recognition to test set is tested.
2. the migration convolution neural network method according to claim 1 towards facial expression recognition, characterized in that S1
In, obtaining Facial Expression Image data set includes following procedure:
S11 obtains CK+ and FER2013 Facial Expression Image data set；
S12 extends CK+ data set；
S13 carries out image normalization to CK+ and FER2013 data set.
3. the migration convolution neural network method according to claim 1 towards facial expression recognition, characterized in that S2
In, migration network is Inception_v3 network.
4. the migration convolution neural network method according to claim 1 towards facial expression recognition, characterized in that migration
Network includes 6 convolutional layers, 3 Inception layers, 2 pond layers and 1 full articulamentum, removes full articulamentum in cascade.
5. the migration convolution neural network method according to claim 1 towards facial expression recognition, characterized in that convolution
Neural network includes 1 convolutional layer, 1 pond layer and a full articulamentum.
6. the migration convolution neural network method according to claim 1 towards facial expression recognition, characterized in that face
Expression classification includes anger, detests, is frightened, glad, sad, surprised and neutral.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811398259.0A CN109376692A (en) | 2018-11-22 | 2018-11-22 | Migration convolution neural network method towards facial expression recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811398259.0A CN109376692A (en) | 2018-11-22 | 2018-11-22 | Migration convolution neural network method towards facial expression recognition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109376692A true CN109376692A (en) | 2019-02-22 |
Family
ID=65382790
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811398259.0A Pending CN109376692A (en) | 2018-11-22 | 2018-11-22 | Migration convolution neural network method towards facial expression recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109376692A (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110163098A (en) * | 2019-04-17 | 2019-08-23 | 西北大学 | Based on the facial expression recognition model construction of depth of seam division network and recognition methods |
| CN110348350A (en) * | 2019-07-01 | 2019-10-18 | 电子科技大学 | A kind of driver status detection method based on facial expression |
| CN110399821A (en) * | 2019-07-17 | 2019-11-01 | 上海师范大学 | Customer satisfaction acquisition methods based on facial expression recognition |
| CN110427846A (en) * | 2019-07-19 | 2019-11-08 | 西安工业大学 | It is a kind of using convolutional neural networks to the face identification method of uneven small sample |
| CN110569742A (en) * | 2019-08-19 | 2019-12-13 | 昆山琪奥智能科技有限公司 | Micro-expression analysis and study judging system |
| CN110930297A (en) * | 2019-11-20 | 2020-03-27 | 咪咕动漫有限公司 | Method and device for migrating styles of face images, electronic equipment and storage medium |
| CN111652171A (en) * | 2020-06-09 | 2020-09-11 | 电子科技大学 | Construction method of facial expression recognition model based on double branch network |
| CN111667559A (en) * | 2020-05-27 | 2020-09-15 | 西北工业大学 | Polymorphic human face emotion generation method based on deep migration network |
| CN111814713A (en) * | 2020-07-15 | 2020-10-23 | 陕西科技大学 | Expression recognition method based on BN parameter transfer learning |
| CN112052772A (en) * | 2020-08-31 | 2020-12-08 | 福建捷宇电脑科技有限公司 | Face shielding detection algorithm |
| CN112132092A (en) * | 2020-09-30 | 2020-12-25 | 四川弘和通讯有限公司 | Fire extinguisher and fire blanket identification method based on convolutional neural network |
| CN112200894A (en) * | 2020-12-07 | 2021-01-08 | 江苏原力数字科技股份有限公司 | Automatic digital human facial expression animation migration method based on deep learning framework |
| CN112307923A (en) * | 2020-10-30 | 2021-02-02 | 北京中科深智科技有限公司 | Partitioned expression migration method and system |
| CN112560687A (en) * | 2020-04-03 | 2021-03-26 | 上海应用技术大学 | Face recognition method |
| CN113011314A (en) * | 2021-03-16 | 2021-06-22 | 华南理工大学 | Facial expression recognition method based on frequency domain features and product neural network |
| CN113205449A (en) * | 2021-05-21 | 2021-08-03 | 珠海金山网络游戏科技有限公司 | Expression migration model training method and device and expression migration method and device |
| CN113255543A (en) * | 2021-06-02 | 2021-08-13 | 西安电子科技大学 | Facial expression recognition method based on graph convolution network |
| CN113762205A (en) * | 2021-09-17 | 2021-12-07 | 深圳市爱协生科技有限公司 | Human face image operation trace detection method, computer equipment and readable storage medium |
| CN114581736A (en) * | 2022-03-15 | 2022-06-03 | 首都师范大学 | Method and system for building emotion picture library |
| CN113505740B (en) * | 2021-07-27 | 2023-10-10 | 北京工商大学 | Face recognition method based on transfer learning and convolutional neural network |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280426A (en) * | 2018-01-23 | 2018-07-13 | 深圳极视角科技有限公司 | Half-light source expression recognition method based on transfer learning and device |
| CN110738071A (en) * | 2018-07-18 | 2020-01-31 | 浙江中正智能科技有限公司 | face algorithm model training method based on deep learning and transfer learning |
-
2018
- 2018-11-22 CN CN201811398259.0A patent/CN109376692A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280426A (en) * | 2018-01-23 | 2018-07-13 | 深圳极视角科技有限公司 | Half-light source expression recognition method based on transfer learning and device |
| CN110738071A (en) * | 2018-07-18 | 2020-01-31 | 浙江中正智能科技有限公司 | face algorithm model training method based on deep learning and transfer learning |
Non-Patent Citations (1)
| Title |
|---|
| 翟懿奎: "面向人脸表情识别的迁移卷积神经网络研究", 《信号处理》 * |
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110163098A (en) * | 2019-04-17 | 2019-08-23 | 西北大学 | Based on the facial expression recognition model construction of depth of seam division network and recognition methods |
| CN110348350A (en) * | 2019-07-01 | 2019-10-18 | 电子科技大学 | A kind of driver status detection method based on facial expression |
| CN110348350B (en) * | 2019-07-01 | 2022-03-25 | 电子科技大学 | Driver state detection method based on facial expressions |
| CN110399821A (en) * | 2019-07-17 | 2019-11-01 | 上海师范大学 | Customer satisfaction acquisition methods based on facial expression recognition |
| CN110427846A (en) * | 2019-07-19 | 2019-11-08 | 西安工业大学 | It is a kind of using convolutional neural networks to the face identification method of uneven small sample |
| CN110427846B (en) * | 2019-07-19 | 2022-12-06 | 西安工业大学 | Face recognition method for small unbalanced samples by using convolutional neural network |
| CN110569742A (en) * | 2019-08-19 | 2019-12-13 | 昆山琪奥智能科技有限公司 | Micro-expression analysis and study judging system |
| CN110930297A (en) * | 2019-11-20 | 2020-03-27 | 咪咕动漫有限公司 | Method and device for migrating styles of face images, electronic equipment and storage medium |
| CN110930297B (en) * | 2019-11-20 | 2023-08-18 | 咪咕动漫有限公司 | Style migration method and device for face image, electronic equipment and storage medium |
| CN112560687A (en) * | 2020-04-03 | 2021-03-26 | 上海应用技术大学 | Face recognition method |
| CN112560687B (en) * | 2020-04-03 | 2024-01-30 | 上海应用技术大学 | Face recognition method |
| CN111667559A (en) * | 2020-05-27 | 2020-09-15 | 西北工业大学 | Polymorphic human face emotion generation method based on deep migration network |
| CN111652171A (en) * | 2020-06-09 | 2020-09-11 | 电子科技大学 | Construction method of facial expression recognition model based on double branch network |
| CN111652171B (en) * | 2020-06-09 | 2022-08-05 | 电子科技大学 | Construction method of facial expression recognition model based on double branch network |
| CN111814713A (en) * | 2020-07-15 | 2020-10-23 | 陕西科技大学 | Expression recognition method based on BN parameter transfer learning |
| CN112052772A (en) * | 2020-08-31 | 2020-12-08 | 福建捷宇电脑科技有限公司 | Face shielding detection algorithm |
| CN112132092A (en) * | 2020-09-30 | 2020-12-25 | 四川弘和通讯有限公司 | Fire extinguisher and fire blanket identification method based on convolutional neural network |
| CN112307923A (en) * | 2020-10-30 | 2021-02-02 | 北京中科深智科技有限公司 | Partitioned expression migration method and system |
| CN112200894B (en) * | 2020-12-07 | 2021-03-09 | 江苏原力数字科技股份有限公司 | Automatic digital human facial expression animation migration method based on deep learning framework |
| CN112200894A (en) * | 2020-12-07 | 2021-01-08 | 江苏原力数字科技股份有限公司 | Automatic digital human facial expression animation migration method based on deep learning framework |
| CN113011314A (en) * | 2021-03-16 | 2021-06-22 | 华南理工大学 | Facial expression recognition method based on frequency domain features and product neural network |
| CN113011314B (en) * | 2021-03-16 | 2023-07-18 | 华南理工大学 | Facial expression recognition method based on frequency domain characteristics and product neural network |
| CN113205449A (en) * | 2021-05-21 | 2021-08-03 | 珠海金山网络游戏科技有限公司 | Expression migration model training method and device and expression migration method and device |
| CN113255543B (en) * | 2021-06-02 | 2023-04-07 | 西安电子科技大学 | Facial expression recognition method based on graph convolution network |
| CN113255543A (en) * | 2021-06-02 | 2021-08-13 | 西安电子科技大学 | Facial expression recognition method based on graph convolution network |
| CN113505740B (en) * | 2021-07-27 | 2023-10-10 | 北京工商大学 | Face recognition method based on transfer learning and convolutional neural network |
| CN113762205A (en) * | 2021-09-17 | 2021-12-07 | 深圳市爱协生科技有限公司 | Human face image operation trace detection method, computer equipment and readable storage medium |
| CN114581736A (en) * | 2022-03-15 | 2022-06-03 | 首都师范大学 | Method and system for building emotion picture library |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109376692A (en) | Migration convolution neural network method towards facial expression recognition | |
| CN108615010B (en) | Facial expression recognition method based on parallel convolution neural network feature map fusion | |
| CN110532900B (en) | Facial expression recognition method based on U-Net and LS-CNN | |
| CN109711413A (en) | Image, semantic dividing method based on deep learning | |
| CN107506722A (en) | One kind is based on depth sparse convolution neutral net face emotion identification method | |
| CN112464865A (en) | Facial expression recognition method based on pixel and geometric mixed features | |
| CN109299716A (en) | Training method, image partition method, device, equipment and the medium of neural network | |
| CN107154043A (en) | A kind of Lung neoplasm false positive sample suppressing method based on 3DCNN | |
| Tiwari et al. | Image classification using deep neural network | |
| CN107145893A (en) | A kind of image recognition algorithm and system based on convolution depth network | |
| CN109190643A (en) | Based on the recognition methods of convolutional neural networks Chinese medicine and electronic equipment | |
| CN105205449A (en) | Sign language recognition method based on deep learning | |
| CN107424145A (en) | The dividing method of nuclear magnetic resonance image based on three-dimensional full convolutional neural networks | |
| CN111652273B (en) | Deep learning-based RGB-D image classification method | |
| CN115966010A (en) | Expression recognition method based on attention and multi-scale feature fusion | |
| CN110909680A (en) | Facial expression recognition method and device, electronic equipment and storage medium | |
| CN109325513A (en) | A kind of image classification network training method based on magnanimity list class single image | |
| CN110096991A (en) | A kind of sign Language Recognition Method based on convolutional neural networks | |
| CN111401156A (en) | Image identification method based on Gabor convolution neural network | |
| Wang et al. | Improvement of MNIST image recognition based on CNN | |
| CN114170657A (en) | Facial emotion recognition method integrating attention mechanism and high-order feature representation | |
| Tabernik et al. | Towards deep compositional networks | |
| CN116883364A (en) | Apple leaf disease identification method based on CNN and Transformer | |
| CN118247821A (en) | Personnel emotion recognition method based on mixed attention mechanism and multi-scale feature fusion | |
| Manzoor et al. | Ancient coin classification based on recent trends of deep learning. |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190222 |
|
| RJ01 | Rejection of invention patent application after publication |