CN108806792A - Deep learning facial diagnosis system - Google Patents
Deep learning facial diagnosis system Download PDFInfo
- Publication number
- CN108806792A CN108806792A CN201711255031.1A CN201711255031A CN108806792A CN 108806792 A CN108806792 A CN 108806792A CN 201711255031 A CN201711255031 A CN 201711255031A CN 108806792 A CN108806792 A CN 108806792A
- Authority
- CN
- China
- Prior art keywords
- disease
- convolutional neural
- training
- neural networks
- deep learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
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
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention provides the facial diagnosis method and systems based on deep learning.The present invention trains deep learning network using the facial image largely with disease label, pass through successive ignition, after parameter update, this deep learning network can effectively extract the face feature including eye ear nose mouth eyebrow automatically, find out the inner link between face feature and each disease of human body, it is a kind of computer assisted automatic diagnosis method of non-intrusion type to make the effective detection and screening of disease to new subject's face image.This system can by persistent collection patient's face picture retraining undated parameter, to keep prediction judging result more accurate.It can effectively solve the problems, such as that backward area disease detection is difficult by such method and system, the people are enable conveniently and efficiently to carry out disease detection and the screening automatically of non-intrusion type by terminals such as portable computers, so that disease is treated in time, improves human life quality's level to improve.
Description
Technical field
The invention belongs to computer vision, machine learning and medical domain more particularly to a kind of logical based on deep learning
Cross the method and system that face diagnoses the illness.
Background technology
Before more than 2000 years, Chinese ancient book《The Yellow Emperor's Canon of Internal Medicine》Recording road, " vim and vigour are played all in the twelve regular channels, 365 tunnels
On in face walk sky key ".This shows the pathological change of the vital organs of the human body of people, can show relevant range on the face.In China, have
The doctor of experience can by the facial characteristics of observer grasp entire patient with part lesion situation, this diagnostic mode
It does " facial diagnosis ".The defect of facial diagnosis is that this diagnostic mode needs a large amount of experiences of doctor that could have relatively high accuracy rate.
With the development of science and technology, big data and deep learning technology are quickly grown in recent years.Deep learning technology is to utilize
Multi-level neural network can make computer learning understand the number such as complicated video/audio by the training of high-volume data
According to, and corresponding behavior can be made, even it is made not bad than the mankind sometimes.Deep learning network can extract complexity
Height, and the contrast characteristic of mankind's indigestion description.
2006, University of Toronto professor Geoffrey Hinton proposed the algorithm of deep learning.He and he
Student Ruslan Salakhutdinov exist《Science》On delivered an article, from this caused deep learning research heat
Tide.2012, the depth convolutional neural networks structure AlexNet that team of University of Toronto is proposed was regarded on a large scale in ImageNet
Feel in identification challenge match (ILSVRC) and has won champion.It is 2014, Google companies and Oxford University's visual geometric group each
From using their depth convolutional neural networks GoogleNet and VGGNet to achieve excellent achievement in ILSVRC, this two
Kind depth convolutional neural networks are more complicated than AlexNet in structure.2015, He Kaiming of Microsoft Research, Asia et al.
Champion ILSVRC has been won using residual error network ResNet, has all been obtained in the tasks such as image classification, target detection, semantic segmentation
The best result of current year.Residual error network has deeper depth, up to 152 layers.2016, integrated deep neural network existed
It makes an outstanding achievement in ILSVRC each tasks.
It can be seen that in the big data epoch now, the mankind can utilize the depth model analysis extraction sea of more sophisticated
The effective information contained in amount data, to more accurately predict unknown make.The essence of deep learning is exactly based on
The machine learning model with multiple hidden layers and a large amount of training data are built, learning characteristic is extracted, to improve classification
Or the accuracy of prediction.Numerous studies experiment shows the method learning characteristic using big data deep learning, than manual construction spy
Sign, can more efficiently give expression to the internal information of data.The present invention is exactly to be realized using the method for big data deep learning
The purpose to be diagnosed the illness by face.
Invention content
The purpose of the present invention is allowing the whole mankind conveniently and efficiently to carry out the detection and screening of disease, the life matter of the mankind is improved
Amount is horizontal, especially backward area.
To achieve the above object, I propose it is a kind of being based on big data deep learning technology, only need to by facial image,
Can non-intrusion type automatic detection and screening disease method and system.Its method includes the following steps：
S100：Suitable facial image sample of the acquisition with medical diagnosis on disease result label, establishing has medical diagnosis on disease mark
The face database of label.
S200：Using convolutional layer, pond layer, batch normalization and activation primitive ReLu are aided with residual error structure and Inception
Structure establishes depth convolutional neural networks.
S300：The picture that S100 is got is by pretreatment, and pretreated image is as training data, training depth
Convolutional neural networks model obtains network model parameter.
S400：Using trained depth convolutional neural networks model in S300, the feature of training data is extracted, is used
The feature of extraction trains random forest grader, obtains disaggregated model.
S500：Tested picture is inputted, the convolutional neural networks model extraction face picture to be sorted in S300 is utilized
Feature, the sorter model then obtained using S400 obtain the disease category belonging to face image to be detected.
The features of the present invention also characterized in that：
Step S100 is while acquiring the face-image of subject, the medical diagnosis on disease report provided at that time according to hospital,
Add medical diagnosis on disease result label.Negroid (black race), the Mongolian race (yellow), Gao Jia are established according to colour of skin difference
The database of rope ethnic group (white people).The picture of acquisition is pre-processed, by pretreated picture, have identical size and
Three color dimension inputs of RGB.
Step S200 has used the mixing module in conjunction with residual error structure and Inception structures in network struction.Network
Input data is handled using batch normalization (Batch Normalization) in structure, pond layer uses global average pond
Method.
Pretreated facial image is divided into training sample and test sample two parts in step S300, uses training sample
Training depth convolutional neural networks update weights using the training method and back-propagation algorithm for having supervision.And use figure
Processor (GPU) training depth convolutional neural networks.Finally test sample input depth convolutional neural networks are tested.It adopts
With the network such as VGG, AlexNet, ResNet of pre-training, Bilinear CNN, Part-based CNN etc. reuses fine tuning skill
This step function equally may be implemented in art.
Using trained depth convolutional neural networks model as feature extractor in step S400, with the spy of extraction
Sign training grader, obtains disaggregated model.Using Softmax, support vector machines (SVM), k nearest neighbor classification device (KNN) etc. is polynary
This step function can be achieved in sorter model.
The disease category that detection and screening may be implemented in step S500 has hyperthyroidism, thyroid gland machine
The low disease of energy, the Chinese is sick, acromegalia, renal failure, pathological jaundice of newborn, Down syndrome, thalassemia
Disease, diGeorge's syndrome, mitral stenosis.More and more kinds of Diseases will be added into.
The beneficial outcomes of the present invention are that the present invention can be by depth convolutional neural networks with medical diagnosis on disease result mark
Feature is automatically extracted under the data-driven of the training image of label.Depth convolutional neural networks can be identified to avoid traditional images and be calculated
The additional pre-treatment process of method.It is accurate to solve to deepen with network that residual error structure is applied in this depth convolutional neural networks
The problem of exactness declines.
Depth convolutional neural networks extraction feature can to avoid by color of light according to etc. influence, can extract it is more complicated and
The inenarrable feature of the mankind.Effective combination of depth convolutional neural networks and grader can not only extract abundant feature, but also
Suitable grader can be chosen to improve the accuracy of entire disaggregated model according to the characteristics of data collection.
Description of the drawings
Fig. 1 is the whole machine learning algorithm schematic diagram of the present invention；
Fig. 2 is the structural schematic diagram of the depth convolutional neural networks of the present invention；
Fig. 3 is the system structure diagram of the present invention.
Specific implementation mode
In order to describe the technical content, the structural feature, the achieved object and the effect of this invention in detail, below in conjunction with embodiment
And attached drawing is coordinated to be explained in detail.
Refering to fig. 1, in order to solve technical problem of the present invention, one aspect of the present invention is to provide one
Method of the kind based on deep learning, includes the following steps：
S100：Suitable facial image sample is acquired, the medical diagnosis on disease provided according to hospital when acquisition is reported, disease is added
Type label establishes the face database with medical diagnosis on disease result label.Database is divided into three categories, Negroid
(black race), the Mongolian race (yellow), Caucasian (white people).The picture of acquisition is pre-processed, by locating in advance
The picture of reason has three color dimension inputs of RGB.The following matconvnet codes based on Matlab platforms are pre-
The embodiment of processing step.
First the image normalization of reading is handled,
Img0=imresize (img0, net.meta.normalization.imageSize (1：2))；
Then subtracted image mean value,
Img0=img0-net.meta.normalization.averagelmage；
S200：Using convolutional layer, pond layer, batch normalization and activation primitive ReLu are aided with residual error structure and Inception
Structure establishes depth convolutional neural networks.
Since GPU has outstanding computation capability, it can be calculated for big data and very big help is provided, in this process
In, the training and deduction of depth convolutional neural networks are carried out using GPU.
Convolutional layer：Convolution is carried out to the image of input using different convolution kernels, a sheet by a sheet characteristic pattern can be obtained, each
Convolution kernel detects the special characteristic on all positions on input feature vector figure, and the weights realized on the same input feature vector figure are total
It enjoys.Each input feature vector pel element is x, and the weights that connection outputs and inputs characteristic pattern element are w, and output characteristic pattern element is
Y, offset parameter b, f are activation primitive, and formula is：
Activation primitive is using ReLu, formula：
F (x)=max (0, x)
Pond layer：It can remove unessential sample in characteristic pattern by down-sampling, training parameter be reduced, to reduce
Models fitting degree.Here we are using global average pond (global average pooling), to each characteristic pattern
Seek whole average value
Dropout：In the training process of deep learning network by according to certain probability by deep learning network
Neural unit is temporarily abandoned from network, can be effectively prevented over-fitting, here value 0.8.
Residual error structure：Residual error structure refers to that have two paths, a paths be input feature vector to each residual error module, another
Path is to carry out the residual error that multiple convolution operates to input feature vector, is finally added input feature vector with residual error：
H (x)=F (x)+x
X indicates that input feature vector, F (x) indicate that residual error feature, H (x) indicate residual error module output feature
Inception structures：The structure being made of parallel various sizes of convolutional layer (such as 1 × 1,3 × 3), increases god
The adaptability of width and network neural through network to scale.
So that neural network is deeper wider in conjunction with the Hybrid modules of residual error structure and Inception structures.
S300：The picture learning sample that S100 is got is obtained as training data, training convolutional neural networks model
Network model parameter.During training convolutional neural networks, more using the training method and back-propagation algorithm for having supervision
New weights.Process is as follows：
Calculate the error of sample output feature and theoretical label value：
By error with the reverse path of convolutional neural networks, it is transmitted to each layer, according to gradient descent algorithm, update power
Weight：
η is learning rate, and δ is error term, and right value update formula is：
Δwij=η δjxij
As convolution kernel weights are continuously updated, the output error of convolutional neural networks is steadily decreasing, and finally tends to be steady
It is fixed.Using the network such as VGG, AlexNet, ResNet of pre-training, Bilinear CNN, Part-based CNN etc. is reused
This step function equally may be implemented in fine tuning technology.
S400：Using trained convolutional neural networks model in S300, the feature of training data is extracted, with classification
Device is trained, and obtains disaggregated model.Support vector machines (SVM), the multivariate classifications device model such as k nearest neighbor classification device (KNN).Herein
It uses random forests algorithm for embodiment, generates multiple categorised decision trees with training set, these decision trees form random forest.?
When prediction, each tree can predict one as a result, it is final knot finally to take that class that classification results are most in all decision trees
It is Gini coefficient that fruit, which generates categorised decision tree and uses division index, and defined formula is：
It is p to have classification k, the probability that sample point belongs to kth class in classification problemk, sample size is | D |, | Ck| to belong to
Class CkNumber of samples.
S500：Refering to Fig. 3, subject inputs face picture in terminal, is transmitted to cloud computing platform and uses the volume in S300
Product neural network model extracts the feature of facial image to be sorted, and the sorter model then obtained using S400 is waited for
Disease category belonging to the face image of classification, returns to subject.The disease of detection and screening may be implemented in S500
Sick classification has hyperthyroidism, and hypothyroidism, the Chinese is sick, acromegalia, renal failure, newborn
Pathological jaundice, Down syndrome, thalassemia, diGeorge's syndrome, mitral stenosis.More and more kinds of Diseases will
It is added into the invention.
Example the above is only the implementation of the present invention is not intended to limit the scope of the invention, every to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (8)
1. deep learning facial diagnosis system, be it is a kind of being based on big data deep learning technology, by face feature to detection and
The system of screening disease.Method is it is characterized in that, it includes the following steps：
S100：The facial image sample for obtaining known medical diagnosis on disease result label, establishes the people with medical diagnosis on disease result label
Face image database.
S200：Utilize convolutional layer, pond layer, batch normalization and activation primitive ReLu combination residual error structures and Inception structures
Build depth convolutional neural networks.
S300：The picture that S100 is got is by pretreatment, and pretreated image is as training data, training depth convolution
Neural network model obtains network model parameter.
S400：Using trained depth convolutional neural networks model in S300, the feature of training data is extracted, with extraction
Feature train random forest grader, obtain disaggregated model.
S500：Tested picture is inputted, using the feature of the convolutional neural networks model extraction face picture to be sorted in S300,
Then the sorter model obtained using S400 obtains the disease category belonging to face image to be detected.
2. disease detection according to claim 1 and screening method, which is characterized in that the step S100 is in acquisition face
While portion's image, according to the diagnosis report of hospital, kinds of Diseases label is added.Negroid is established according to colour of skin difference
(black race), the Mongolian race (yellow), Caucasian (white people) database.The picture of acquisition is pre-processed,
By pretreated picture, there are the three color dimension inputs of identical size and RGB.
3. disease detection according to claim 1 and screening method, which is characterized in that the step S200 is in network struction
The middle mixing module used in conjunction with residual error structure and Inception structures.The method that pond layer uses global average pond.
4. disease detection according to claim 1 and screening method, which is characterized in that will pretreatment in the step S300
Facial image afterwards is divided into training sample and test sample two parts, trains depth convolutional neural networks with training sample, uses
There are the training method and back-propagation algorithm update weights of supervision.And use graphics processor (GPU) training depth convolution god
Through network.Finally test sample input depth convolutional neural networks are tested.Using the network of pre-training, fine tuning is reused
This step function equally may be implemented in technology.
5. disease detection according to claim 1 and screening method, which is characterized in that will have been instructed in the step S400
The depth convolutional neural networks model perfected trains grader as feature extractor, with the feature of extraction, obtains disaggregated model.
Using Softmax, support vector machines (SVM), this step work( can be achieved in its multivariate classification device model such as k nearest neighbor classification device (KNN)
Energy.
6. disease detection according to claim 1 and screening method, which is characterized in that can be in the step S500
Realize that the disease category of detection and screening has hyperthyroidism, hypothyroidism, the Chinese is sick, acromegaly
Disease, renal failure, pathological jaundice of newborn, Down syndrome, thalassemia, diGeorge's syndrome, mitral stenosis.
7. one kind being based on big data deep learning technology, the system by face feature to detect screening disease.System its
It is characterized in that having used the method for the present invention, system include：
Terminal：Man face image acquiring module, information sending module, information receiving module
Cloud computing platform：Image receiver module, the face database with medical diagnosis on disease result label, depth convolutional Neural
Network training module, depth convolutional neural networks classification prediction module, prediction result information sending module of classifying.
8. disease detection according to claim 7 and screening system, which is characterized in that user can pass through terminal device packet
Desktop computer is included, smart mobile phone, tablet computer, laptop, input facial image is to cloud computing platform, by claim 1
The depth convolutional neural networks automatically extract facial characteristics, and grader carries out disease detection.System will return to disease detection
Classification results are to terminal device.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2017103453755 | 2017-05-03 | ||
| CN201710345375 | 2017-05-03 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108806792A true CN108806792A (en) | 2018-11-13 |
| CN108806792B CN108806792B (en) | 2022-01-04 |
Family
ID=64095168
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711255031.1A Active CN108806792B (en) | 2017-05-03 | 2017-11-25 | Deep learning face diagnosis system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108806792B (en) |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109685968A (en) * | 2018-12-15 | 2019-04-26 | 西安建筑科技大学 | A kind of the identification model building and recognition methods of the banknote image defect based on convolutional neural networks |
| CN109741823A (en) * | 2018-11-23 | 2019-05-10 | 杭州电子科技大学 | A kind of pneumothorax aided diagnosis method based on deep learning |
| CN109766920A (en) * | 2018-12-18 | 2019-05-17 | 任飞翔 | Article characteristics Model Calculating Method and device based on deep learning |
| CN109785311A (en) * | 2019-01-14 | 2019-05-21 | 深圳和而泰数据资源与云技术有限公司 | A kind of methods for the diagnosis of diseases and relevant device |
| CN109920538A (en) * | 2019-03-07 | 2019-06-21 | 中南大学 | A kind of zero sample learning method based on data enhancing |
| CN110363075A (en) * | 2019-06-03 | 2019-10-22 | 陈丙涛 | Suspicious ill face detection system based on big data server |
| CN111104961A (en) * | 2019-10-31 | 2020-05-05 | 太原理工大学 | Method for classifying breast cancer based on improved MobileNet network |
| CN111341459A (en) * | 2020-02-28 | 2020-06-26 | 上海交通大学医学院附属上海儿童医学中心 | Training method of classified deep neural network model and genetic disease detection method |
| CN111353520A (en) * | 2019-07-03 | 2020-06-30 | 孙彤 | Method for identifying vermilion ink time sequence |
| CN111489325A (en) * | 2019-12-16 | 2020-08-04 | 杭州电子科技大学 | Instant identification device for hyperthyroidism disease and control method thereof |
| CN111803024A (en) * | 2020-08-27 | 2020-10-23 | 张桦 | System and method for measuring exophthalmos based on deep learning algorithm |
| CN111951950A (en) * | 2019-05-16 | 2020-11-17 | 中国科学院自动化研究所 | Three-dimensional data medical classification system, method and device based on deep learning |
| CN111951934A (en) * | 2020-08-20 | 2020-11-17 | 陈文立 | Novel acromegaly screening system and screening method thereof |
| CN112163512A (en) * | 2020-09-25 | 2021-01-01 | 杨铠郗 | Autism spectrum disorder face screening method based on machine learning |
| CN112687393A (en) * | 2020-12-29 | 2021-04-20 | 康奥生物科技(天津)股份有限公司 | Disease prediction system based on big data |
| CN113012823A (en) * | 2021-03-30 | 2021-06-22 | 海口讯婕城医疗科技有限公司 | Cloud platform system for electrocardiograph |
| CN113139597A (en) * | 2021-04-19 | 2021-07-20 | 中国人民解放军91054部队 | Statistical thought-based image distribution external detection method |
| CN113284613A (en) * | 2021-05-24 | 2021-08-20 | 暨南大学 | Face diagnosis system based on deep learning |
| CN113625094A (en) * | 2021-09-11 | 2021-11-09 | 沈阳工业大学 | Subway platform screen door fault diagnosis system |
| CN114502061A (en) * | 2018-12-04 | 2022-05-13 | 巴黎欧莱雅 | Image-based automatic skin diagnosis using deep learning |
| WO2023161956A1 (en) * | 2022-02-28 | 2023-08-31 | Saanvi Mehra | Device and method for assessing a subject for down syndrome |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102426652A (en) * | 2011-10-10 | 2012-04-25 | 北京工业大学 | Traditional Chinese medicine face color identifying and retrieving method based on image analysis |
| CN102973253A (en) * | 2012-10-31 | 2013-03-20 | 北京大学 | Method and system for monitoring human physiological indexes by using visual information |
| JP2014182424A (en) * | 2013-03-18 | 2014-09-29 | Seiko Epson Corp | Printing equipment, printing system, and printed matter |
| CN104699931A (en) * | 2013-12-09 | 2015-06-10 | 广州华久信息科技有限公司 | Neural network blood pressure prediction method and mobile phone based on human face |
| CN105590102A (en) * | 2015-12-30 | 2016-05-18 | 中通服公众信息产业股份有限公司 | Front car face identification method based on deep learning |
-
2017
- 2017-11-25 CN CN201711255031.1A patent/CN108806792B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102426652A (en) * | 2011-10-10 | 2012-04-25 | 北京工业大学 | Traditional Chinese medicine face color identifying and retrieving method based on image analysis |
| CN102973253A (en) * | 2012-10-31 | 2013-03-20 | 北京大学 | Method and system for monitoring human physiological indexes by using visual information |
| JP2014182424A (en) * | 2013-03-18 | 2014-09-29 | Seiko Epson Corp | Printing equipment, printing system, and printed matter |
| CN104699931A (en) * | 2013-12-09 | 2015-06-10 | 广州华久信息科技有限公司 | Neural network blood pressure prediction method and mobile phone based on human face |
| CN105590102A (en) * | 2015-12-30 | 2016-05-18 | 中通服公众信息产业股份有限公司 | Front car face identification method based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 梁玉梅: "基于面象特征的中医体质自动辨识***研究", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》 * |
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109741823A (en) * | 2018-11-23 | 2019-05-10 | 杭州电子科技大学 | A kind of pneumothorax aided diagnosis method based on deep learning |
| CN114502061A (en) * | 2018-12-04 | 2022-05-13 | 巴黎欧莱雅 | Image-based automatic skin diagnosis using deep learning |
| CN114502061B (en) * | 2018-12-04 | 2024-05-28 | 巴黎欧莱雅 | Image-based automatic skin diagnosis using deep learning |
| CN109685968A (en) * | 2018-12-15 | 2019-04-26 | 西安建筑科技大学 | A kind of the identification model building and recognition methods of the banknote image defect based on convolutional neural networks |
| CN109766920A (en) * | 2018-12-18 | 2019-05-17 | 任飞翔 | Article characteristics Model Calculating Method and device based on deep learning |
| CN109785311A (en) * | 2019-01-14 | 2019-05-21 | 深圳和而泰数据资源与云技术有限公司 | A kind of methods for the diagnosis of diseases and relevant device |
| CN109785311B (en) * | 2019-01-14 | 2021-06-04 | 深圳和而泰数据资源与云技术有限公司 | Disease diagnosis device, electronic equipment and storage medium |
| CN109920538A (en) * | 2019-03-07 | 2019-06-21 | 中南大学 | A kind of zero sample learning method based on data enhancing |
| CN109920538B (en) * | 2019-03-07 | 2022-11-25 | 中南大学 | Zero sample learning method based on data enhancement |
| CN111951950B (en) * | 2019-05-16 | 2023-11-24 | 中国科学院自动化研究所 | Three-dimensional data medical classification system based on deep learning |
| CN111951950A (en) * | 2019-05-16 | 2020-11-17 | 中国科学院自动化研究所 | Three-dimensional data medical classification system, method and device based on deep learning |
| CN110363075A (en) * | 2019-06-03 | 2019-10-22 | 陈丙涛 | Suspicious ill face detection system based on big data server |
| CN111353520A (en) * | 2019-07-03 | 2020-06-30 | 孙彤 | Method for identifying vermilion ink time sequence |
| CN111104961A (en) * | 2019-10-31 | 2020-05-05 | 太原理工大学 | Method for classifying breast cancer based on improved MobileNet network |
| CN111489325B (en) * | 2019-12-16 | 2023-07-28 | 杭州电子科技大学 | Instant identification device for hyperthyroidism disease and control method thereof |
| CN111489325A (en) * | 2019-12-16 | 2020-08-04 | 杭州电子科技大学 | Instant identification device for hyperthyroidism disease and control method thereof |
| CN111341459A (en) * | 2020-02-28 | 2020-06-26 | 上海交通大学医学院附属上海儿童医学中心 | Training method of classified deep neural network model and genetic disease detection method |
| CN111951934A (en) * | 2020-08-20 | 2020-11-17 | 陈文立 | Novel acromegaly screening system and screening method thereof |
| CN111803024B (en) * | 2020-08-27 | 2022-10-04 | 张桦 | System for measuring eyeball protrusion degree based on deep learning algorithm |
| CN111803024A (en) * | 2020-08-27 | 2020-10-23 | 张桦 | System and method for measuring exophthalmos based on deep learning algorithm |
| CN112163512A (en) * | 2020-09-25 | 2021-01-01 | 杨铠郗 | Autism spectrum disorder face screening method based on machine learning |
| CN112687393A (en) * | 2020-12-29 | 2021-04-20 | 康奥生物科技(天津)股份有限公司 | Disease prediction system based on big data |
| CN113012823A (en) * | 2021-03-30 | 2021-06-22 | 海口讯婕城医疗科技有限公司 | Cloud platform system for electrocardiograph |
| CN113139597A (en) * | 2021-04-19 | 2021-07-20 | 中国人民解放军91054部队 | Statistical thought-based image distribution external detection method |
| CN113139597B (en) * | 2021-04-19 | 2022-11-04 | 中国人民解放军91054部队 | Statistical thought-based image distribution external detection method |
| CN113284613A (en) * | 2021-05-24 | 2021-08-20 | 暨南大学 | Face diagnosis system based on deep learning |
| CN113625094A (en) * | 2021-09-11 | 2021-11-09 | 沈阳工业大学 | Subway platform screen door fault diagnosis system |
| WO2023161956A1 (en) * | 2022-02-28 | 2023-08-31 | Saanvi Mehra | Device and method for assessing a subject for down syndrome |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108806792B (en) | 2022-01-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108806792A (en) | Deep learning facial diagnosis system | |
| US10482603B1 (en) | Medical image segmentation using an integrated edge guidance module and object segmentation network | |
| Singh et al. | Image classification: a survey | |
| Cheng et al. | Facial expression recognition method based on improved VGG convolutional neural network | |
| CN110222140A (en) | A kind of cross-module state search method based on confrontation study and asymmetric Hash | |
| CN109559300A (en) | Image processing method, electronic equipment and computer readable storage medium | |
| CN106408562A (en) | Fundus image retinal vessel segmentation method and system based on deep learning | |
| CN109493342B (en) | Skin disease picture lesion type classification method based on deep learning | |
| CN109858429A (en) | A kind of identification of eye fundus image lesion degree and visualization system based on convolutional neural networks | |
| CN111611851B (en) | Model generation method, iris detection method and device | |
| Saleh et al. | Transfer learning‐based platform for detecting multi‐classification retinal disorders using optical coherence tomography images | |
| US20210406687A1 (en) | Method for predicting attribute of target object based on machine learning and related device | |
| CN113221655B (en) | Face spoofing detection method based on feature space constraint | |
| CN112419326B (en) | Image segmentation data processing method, device, equipment and storage medium | |
| CN111477337B (en) | Infectious disease early warning method, system and medium based on individual self-adaptive transmission network | |
| CN110443105A (en) | The immunofluorescence image kenel recognition methods of autoimmunity antibody | |
| CN107220598A (en) | Iris Texture Classification based on deep learning feature and Fisher Vector encoding models | |
| Fu et al. | Computerized tongue coating nature diagnosis using convolutional neural network | |
| CN114913923A (en) | Cell type identification method aiming at open sequencing data of single cell chromatin | |
| CN112949469A (en) | Image recognition method, system and equipment for face tampered image characteristic distribution | |
| CN110378353A (en) | A kind of tongue picture feature extracting method, system and computer readable storage medium | |
| CN105809200A (en) | Biologically-inspired image meaning information autonomous extraction method and device | |
| CN113011436A (en) | Traditional Chinese medicine tongue color and fur color collaborative classification method based on convolutional neural network | |
| Cao et al. | 3D convolutional neural networks fusion model for lung nodule detection onclinical CT scans | |
| Medi et al. | Skinaid: a gan-based automatic skin lesion monitoring method for iomt frameworks |
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 |