CN110414298A - A kind of more attribute joint recognition methods of monkey face - Google Patents
A kind of more attribute joint recognition methods of monkey face Download PDFInfo
- Publication number
- CN110414298A CN110414298A CN201810398339.XA CN201810398339A CN110414298A CN 110414298 A CN110414298 A CN 110414298A CN 201810398339 A CN201810398339 A CN 201810398339A CN 110414298 A CN110414298 A CN 110414298A
- Authority
- CN
- China
- Prior art keywords
- monkey
- face
- training
- monkey face
- attribute
- 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
- 241000510093 Quadrula metanevra Species 0.000 title claims abstract description 90
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 61
- 241000282693 Cercopithecidae Species 0.000 claims abstract description 15
- 241000282675 Lagothrix Species 0.000 claims abstract description 5
- 238000013507 mapping Methods 0.000 claims abstract description 5
- 238000012360 testing method Methods 0.000 claims description 15
- 238000002360 preparation method Methods 0.000 claims description 4
- 238000004321 preservation Methods 0.000 claims description 4
- 230000008676 import Effects 0.000 claims description 3
- 230000003252 repetitive effect Effects 0.000 claims description 3
- 238000011160 research Methods 0.000 description 10
- 238000013527 convolutional neural network Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 241000282577 Pan troglodytes Species 0.000 description 3
- 241001494479 Pecora Species 0.000 description 3
- 241001481816 Eulemur mongoz Species 0.000 description 2
- 241000282575 Gorilla Species 0.000 description 2
- 241001465754 Metazoa Species 0.000 description 2
- 208000002193 Pain Diseases 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000006378 damage Effects 0.000 description 2
- 230000001815 facial effect Effects 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000013526 transfer learning Methods 0.000 description 2
- 206010016059 Facial pain Diseases 0.000 description 1
- 241000282326 Felis catus Species 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000008921 facial expression Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 208000028173 post-traumatic stress disease Diseases 0.000 description 1
- 230000008569 process Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a kind of more attribute joint recognition methods of monkey face, comprising: building monkey face sample database: acquisition monkey video and picture information are handled, and generate monkey face sample database based on Matlab, and all samples all have the handmarking of Lagothrix property classification;It generates face character and identifies pre-training model: establishing the ResNet50 of face character identification pre-training model according to the study of the disturbance of identical mapping;The more attributes of training monkey face combine identification model: being based on MXNet platform, on the basis of face character identifies pre-training model, monkey face sample database are recycled to carry out tuning training, obtain the more attributes of monkey face and combine identification model.The present invention can realize while identify the target of two attributes according to the common trait of age and gender attribute.
Description
Technical field
The present invention relates to computer vision field, in particular to a kind of more attribute joint recognition methods of monkey face.
Background technique
More attribute joint identifications (Multi-Task) refer to that training can be simultaneously using the common feature between attribute
Realize the model of multiple Attribute Recognitions.In terms of face, with the development that single attribute identifies, the identification of more attributes joint gradually at
For the trend of Attribute Recognition, this joint identification is better than single attribute identification in terms of accuracy rate and efficiency, it may have wider
Wealthy application space.The depth network of currently used more attribute joint identifications is VGG network, for combining the face category of identification
Property is usually age, gender, nationality, nationality etc..The researchers such as Abdulnab will obtain in CNN network application to binary feature
The characteristic present of particular community, to learn these features using Multi-Task and make more Attribute Recognitions.Liu et al. researcher
Then cascaded two CNN networks: LNet and ANet identifies the facial attribute under natural conditions.Hwang devises a feature
Shared network implementations identification mission.
In terms of zooscopy, computer vision technique has also been had been applied in the attributes research of individual, is made that perhaps
Mostly corresponding research achievement.Cambridge University researcher develops a set of sheep human facial expression recognition system using machine learning algorithm
System, can assess the pain index of sheep automatically;Some researchers propose that the face of a kind of entitled " mongoose lemur face ID system " is known
Other system, has reached higher accuracy rate;Other researchers carry out chimpanzee using convolutional neural networks technology (CNN)
The research of image level, is successfully realized the identification of the attributes such as the identity to chimpanzee, age, age cohort, gender.Equally,
In the research to gorilla, researcher extracts feature using the method that Gabor characteristic and local reserved mapping combine, and utilizes
Rarefaction representation face recognition algorithms are classified to realize.The successful application of these technologies is the management of biological reserve bio-diversity
Provide new channel.
There are mainly two types of the methods of traditional individual: one is by capture and labelling technique;It is another then rely on research
Cognition of the personnel for individual shape.However, traditional research method is but faced with many predicaments.
Capture and labelling technique are referred to using artificial method label animal (as worn ear tag or electronic chaplet).Although
This way can meet the needs of individual research to a certain extent, but there is also this simultaneously with high costs, equipment damage or
The risk of aging bring data acquisition inaccuracy.Also, during capture and label, captured animal may cause
It gets a fright, causes some physiological damages even moral injury, influence life, procreation after it.And another traditional
Body research method --- knowledge of the researcher itself for individual profile variation --- be then easy by internal or observer it
Between error influence, and bring observation Data Data integration aspect difficulty.
Likewise, application of the computer vision technique in zooscopy still has problems.The face pain of sheep
Pain identification more attention is the variation of the form of ear, and facial research is relatively fewer;What the identification of cat face used is still to pass
The algorithm of the machine learning of system, it is relatively low in accuracy rate;Mongoose lemur ID identifying system lays particular emphasis on identification rather than attributive analysis;
The attributive analysis of gorilla and chimpanzee is to carry out single attribute analysis using CNN network, lacks more attribute Conjoint Analysis.
Summary of the invention
The main object of the present invention is to propose a kind of more attribute joint recognition methods of monkey face, it is intended to overcome problem above.
To achieve the above object, the more attribute joint recognition methods of a kind of monkey face proposed by the present invention, include the following steps:
S10 constructs monkey face sample database: acquisition monkey video and picture information are handled, and generate monkey based on Matlab
Face sample database, and all samples all have the handmarking of Lagothrix property classification;
S20 generates face character and identifies pre-training model: establishing face character according to the study of the disturbance of identical mapping and knows
Other pre-training model ResNet50;
The more attributes of S30 training monkey face combine identification model: being based on MXNet platform, identify pre-training model in face character
On the basis of, it recycles monkey face sample database to carry out tuning training, obtains the more attributes of monkey face and combine identification model.
Preferably, the step of handling in the S10 the monkey video and picture information of acquisition include:
S101 reads primary video data every 5 frames and finds there are the frame of monkey face in read frame, use rectangle
Frame intercepts monkey face image, its name, frame index, coordinate information are recorded its recording documents;
S102 saves intercepted monkey face image and its recording documents.
Preferably, for the monkey face image intercepted in the S102 with the preservation of jpg format, pixel size is 100 × 100.
Preferably, the coordinate information include the coordinate of interception image position, interception image rectangle frame length and width.
Preferably, the S30 is specifically included:
The preparation of S301 training set and test set: total sample at age and gender joint recognition training will be used for by 5:1's
Ratio is divided into training set and test set, and the recording documents of determining training set and test set are path and the category of log data set
The .lst format file of inventory converts .rec format file for the .lst format file of recording documents by dos operating system;
S302 imports the resnet-50-0000.params file of face character identification pre-training model, and its flatten
Layer addition below is used for the fc1 layer that the age identifies and its is softmax1 layers corresponding, and in flatten layers of parallel addition below
For gender identification fc2 layer and its is softmax2 layers corresponding, and be combined into one with softmax2 layers for softmax1 layers
A final softmax layer, the output for final recognition result;
S303 sets the tuning training parameter of the more attribute joint identification models of monkey face according to monkey face sample database;
Repetitive exercise, the more attributes of acquisition monkey face combine identification models to S304 for several times.
Preferably, the .lst format file of the recording documents is write according to the form column of ID, category, path and filename.
Preferably, age and gender are distinguished in a manner of classification 1, classification 2 and are write out by the category in the category inventory.
Preferably for the data at age and gender double-attribute, category is write out according to classification 1,2 juxtaposition of classification.
Technical solution of the present invention is by establishing to monkey video and the progress monkey face interception of monkey photo is shot under wild environment
Monkey face sample database.The database is for training the more attribute joint identification models of monkey face.Then, more attributes in the present invention
Joint recognizer passes through transfer learning face character identification model: ResNet50, carries out tuning training using monkey face sample, obtains
It has arrived and has combined identification model for more attributes at monkey face age and gender attribute identification.The model can be according to age and gender category
Property common trait realize simultaneously identification two attributes target.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
The structure shown according to these attached drawings obtains other attached drawings.
Fig. 1 is monkey face sample database establishment process schematic diagram;
Fig. 2 is that the more attributes of monkey face based on ResNet50 combine identification model structure chart;
Fig. 3 is the monkey face image of four age levels, and (a) childhood refers to that (b) youth refers to 6-10 years old 0-5 years old,
(c) middle age refers to that (d) old age refers to 16-20 years old 11-15 years old;
Fig. 4 is the monkey face image of female and male,
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiment is only a part of the embodiments of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its
His embodiment, shall fall within the protection scope of the present invention.
It is to be appreciated that if relating to directionality instruction (such as up, down, left, right, before and after ...) in the embodiment of the present invention,
Then directionality instruction be only used for explain under a certain particular pose (as shown in the picture) between each component relative positional relationship,
Motion conditions etc., if the particular pose changes, directionality instruction is also correspondingly changed correspondingly.
As shown in Figs 1-4, the more attribute joint recognition methods of a kind of monkey face proposed by the present invention, include the following steps:
S10 constructs monkey face sample database: acquisition monkey video and picture information are handled, and generate monkey based on Matlab
Face sample database, and all samples all have the handmarking of Lagothrix property classification;
S20 generates face character and identifies pre-training model: establishing face character according to the study of the disturbance of identical mapping and knows
Other pre-training model ResNet50;
The more attributes of S30 training monkey face combine identification model: being based on MXNet platform, identify pre-training model in face character
On the basis of, it recycles monkey face sample database to carry out tuning training, obtains the more attributes of monkey face and combine identification model.
Preferably, the step of handling in the S10 the monkey video and picture information of acquisition include:
S101 reads primary video data every 5 frames and finds there are the frame of monkey face in read frame, use rectangle
Frame intercepts monkey face image, its name, frame index, coordinate information are recorded its recording documents;
S102 saves intercepted monkey face image and its recording documents.
Preferably, for the monkey face image intercepted in the S102 with the preservation of jpg format, pixel size is 100 × 100.
Preferably, the coordinate information include the coordinate of interception image position, interception image rectangle frame length and width.
Preferably, the S30 is specifically included:
The preparation of S301 training set and test set: total sample at age and gender joint recognition training will be used for by 5:1's
Ratio is divided into training set and test set, and the recording documents of determining training set and test set are path and the category of log data set
The .lst format file of inventory converts .rec format file for the .lst format file of recording documents by dos operating system;
S302 imports the resnet-50-0000.params file of face character identification pre-training model, and its flatten
Layer addition below is used for the fc1 layer that the age identifies and its is softmax1 layers corresponding, and in flatten layers of parallel addition below
For gender identification fc2 layer and its is softmax2 layers corresponding, and be combined into one with softmax2 layers for softmax1 layers
A final softmax layer, the output for final recognition result;
S303 sets the tuning training parameter of the more attribute joint identification models of monkey face according to monkey face sample database;
Repetitive exercise, the more attributes of acquisition monkey face combine identification models to S304 for several times.
Preferably, the .lst format file of the recording documents is write according to the form column of ID, category, path and filename.
Preferably, age and gender are distinguished in a manner of classification 1, classification 2 and are write out by the category in the category inventory.
Preferably for the data at age and gender double-attribute, category is write out according to classification 1,2 juxtaposition of classification.
In embodiments of the present invention, the format that the present invention acquires monkey video is MP4 format, and the pixel of picture information figure is equal
It is 1080 × 1920, monkey face sample database is to be generated based on Matlab using video data and picture information.For video
Data, such as Fig. 1 (a): firstly, reading primary video data every 5 frames.In the frame of reading, finds there are the frame of monkey face, use
Rectangle frame intercepts monkey face image, and records in recording documents: name, frame index, coordinate information.Save the monkey face image of interception
And recording documents.The monkey face of interception is unified for 100 pixels × 100 pixels with the preservation of jpg format, pixel size.
For picture information, as shown in Fig. 1 (b), what is taken is same method.Photo is read, monkey face is therefrom intercepted
Sample, and generate recording documents.Finally, obtaining the sample for carrying out the more attribute joint identification model training of monkey face has 2823
.Wherein, it is opened for training set 2351, test set 472 is opened, and ratio is about 5:1.According to such pro rate training set and survey
The reason of examination collection, there is two: one, it is ensured that the quantity of training set sample, to guarantee that the more attribute joint identification models of monkey face can be use up
Required feature may be sufficiently extracted, realization accurately identifies;Second, eliminating as much as the contingency of test result, exclude certain
Special picture, such as especially fuzzy image, the influence to test result obtain consistent universal test result.
All samples for carrying out the more attribute joint identification model training of monkey face all take handmarking's attribute classification
Mode.The handmarking of Lagothrix property classification refers to taking manual sort's measure to the monkey face sample got, and generates note
Record document.I.e. each sample possesses the category corresponding to two kinds of different attributes.For age attribute, sample can be divided into four
Class, i.e. childhood (0-5 years old), young (6-10 years old), middle aged (11-15 years old) and old age (16-20 years old).It, can be with for gender attribute
It is divided into two classes, female and male.That is, participating in the sample of training for each, two categories, a class can be all matched
The type of mark mark age attribute, the type of another mark gender attribute.Likewise, the recording mode of recording documents is also to press
According to: sample names, age categories, gender classification record line by line.
Total number of samples for age and gender joint recognition training are as follows: 2823, it is divided into training set 2351 and opens, test set
472, ratio is about 5:1.The recording documents of training set and test set refer to describing path and the category inventory of data set
.lst file.The format of this file is write according to the form column of ID, category, path and filename.For age and property
The data of other double-attribute, category are just write out according to mode as classification 1, classification 2 side by side.After finishing writing .lst document, ordered in DOS
Enable the corresponding .rec file of lower generation.Pre-training model is ResNet50, that is, needs to prepare the resnet- downloaded in advance
50-0000.params file.
The training of more attribute joint identification (Multi-Task) models of monkey face is carried out based on MXNet platform.Monkey face is more
Attribute joint identification model is not directly to be trained using monkey face sample, but in face character identification model ResNet50
On the basis of, tuning training is carried out using monkey face sample, to obtain the more attribute joint identification models of monkey face.
Specific practice is exactly: at the beginning of training, importing resnet-50-0000.params file, modification
Flatten layers.The fc1 layer that the age identifies is used in flatten layers of addition below and its is softmax1 layers corresponding;Together
When, the fc2 layer that gender identifies is used in flatten layers of parallel addition below and its is softmax2 layers corresponding.Finally will
Softmax1 layers and softmax2 layers are combined into a final softmax layer, the output for final recognition result.Its network
Structure is as shown in Figure 2.
After setting training parameter according to setting table 1, so that it may start having trained for the more attribute joint identification models of monkey face.
It will eventually get the more attribute joint identification models of monkey face of needs.
Table 1: the more attributes of monkey face based on ResNet50 combine identification model tuning training parameter
Parameter | Tuning training |
The number of iterations | 1000 |
Number of tasks | 2 |
Learning rate | 0.001 |
Technical solution of the present invention is by establishing to monkey video and the progress monkey face interception of monkey photo is shot under wild environment
Monkey face sample database.The database is for training the more attribute joint identification models of monkey face.Then, more attributes in the present invention
Joint recognizer passes through transfer learning face character identification model: ResNet50, carries out tuning training using monkey face sample, obtains
It has arrived and has combined identification model for more attributes at monkey face age and gender attribute identification.The model can be according to age and gender category
Property common trait realize simultaneously identification two attributes target.
Relative to pervious zooscopy technology, the present invention is had the advantage that
1, of monkey face in image or image data can quickly be known by not needing any priori knowledge and practical contact
Body attribute information improves the efficiency and accuracy rate of zooscopy, reduces research cost.
2, it supports more attributes to identify simultaneously, improves the efficiency of identification.
Recognition accuracy the present invention is based on the more attribute joint identification models of ResNet50 monkey face at the age is 39.65%,
The accuracy rate of gender identification is 84.38%.
Basis of the invention is the more attribute joint identification models of the monkey face arrived using the training of ResNet50 tuning, the mould
Type can identify simultaneously age and gender attribute from monkey face image.Therefore, any more for monkey face based on proposed by the present invention
The depth network model of Attribute Recognition is included within the present invention, for example uses AlexNet, other minds such as GoogLeNet
Through network structure, take directly trained method or using different network models as pre-training model, such as
InceptionV3 model etc..
Claims (8)
1. a kind of more attribute joint recognition methods of monkey face, which comprises the steps of:
S10 constructs monkey face sample database: acquisition monkey video and picture information are handled, and generate monkey face based on Matlab
Sample database, and all samples all have the handmarking of Lagothrix property classification;
S20 generates face character and identifies pre-training model: it is pre- to establish face character identification according to the study of the disturbance of identical mapping
Training pattern ResNet50;
The more attributes of S30 training monkey face combine identification model: MXNet platform are based on, in the base of face character identification pre-training model
On plinth, monkey face sample database is recycled to carry out tuning training, obtains the more attributes of monkey face and combine identification model.
2. the more attribute joint recognition methods of monkey face as described in claim 1, which is characterized in that the monkey of acquisition in the S10
The step of sub-video and picture information are handled include:
S101 reads primary video data every 5 frames, in read frame, finds there are the frame of monkey face, is cut using rectangle frame
Monkey face image is taken, its name, frame index, coordinate information are recorded into its recording documents;
S102 saves intercepted monkey face image and its recording documents.
3. the more attribute joint recognition methods of monkey face as claimed in claim 2, which is characterized in that the monkey intercepted in the S102
For face image with the preservation of jpg format, pixel size is 100 × 100.
4. the more attribute joint recognition methods of monkey face as claimed in claim 2, which is characterized in that the coordinate information includes interception
The length and width of the coordinate of picture position, the rectangle frame of interception image.
5. the more attribute joint recognition methods of monkey face as described in claim 1, which is characterized in that the S30 is specifically included:
The preparation of S301 training set and test set: total sample at age and gender joint recognition training will be used in the ratio of 5:1
It is divided into training set and test set, and the recording documents of determining training set and test set are path and the category inventory of log data set
.lst format file .rec format file is converted for the .lst format file of recording documents by dos operating system;
S302 imports the resnet-50-0000.params file of face character identification pre-training model, and its flatten layers
Addition below is used for the fc1 layer that the age identifies and its is softmax1 layers corresponding, and the parallel addition use behind flatten layers
In gender identification fc2 layer and its is softmax2 layers corresponding, and be combined into one with softmax2 layers for softmax1 layers
Final softmax layer, the output for final recognition result;
S303 sets the tuning training parameter of the more attribute joint identification models of monkey face according to monkey face sample database;
Repetitive exercise, the more attributes of acquisition monkey face combine identification models to S304 for several times.
6. the more attribute joint recognition methods of monkey face as claimed in claim 5, which is characterized in that the .lst lattice of the recording documents
Formula document is write according to the form column of ID, category, path and filename.
7. the more attribute joint recognition methods of monkey face as claimed in claim 5, which is characterized in that the category in the category inventory
Age and gender are distinguished in a manner of classification 1, classification 2 and write out.
8. the more attribute joint recognition methods of monkey face as claimed in claim 5, which is characterized in that for age and gender double-attribute
Data, category writes out according to classification 1,2 juxtaposition of classification.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810398339.XA CN110414298B (en) | 2018-04-28 | 2018-04-28 | Monkey face multi-attribute joint identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810398339.XA CN110414298B (en) | 2018-04-28 | 2018-04-28 | Monkey face multi-attribute joint identification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110414298A true CN110414298A (en) | 2019-11-05 |
CN110414298B CN110414298B (en) | 2023-07-07 |
Family
ID=68356979
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810398339.XA Active CN110414298B (en) | 2018-04-28 | 2018-04-28 | Monkey face multi-attribute joint identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110414298B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110781866A (en) * | 2019-11-08 | 2020-02-11 | 成都大熊猫繁育研究基地 | Panda face image gender identification method and device based on deep learning |
CN111079624A (en) * | 2019-12-11 | 2020-04-28 | 北京金山云网络技术有限公司 | Method, device, electronic equipment and medium for collecting sample information |
CN112036520A (en) * | 2020-11-06 | 2020-12-04 | 四川师范大学 | Panda age identification method and device based on deep learning and storage medium |
CN116057549A (en) * | 2020-08-28 | 2023-05-02 | 爱你康控股株式会社 | Premium calculation system and premium calculation method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107103281A (en) * | 2017-03-10 | 2017-08-29 | 中山大学 | Face identification method based on aggregation Damage degree metric learning |
CN107247947A (en) * | 2017-07-07 | 2017-10-13 | 北京智慧眼科技股份有限公司 | Face character recognition methods and device |
CN107330451A (en) * | 2017-06-16 | 2017-11-07 | 西交利物浦大学 | Clothes attribute retrieval method based on depth convolutional neural networks |
-
2018
- 2018-04-28 CN CN201810398339.XA patent/CN110414298B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107103281A (en) * | 2017-03-10 | 2017-08-29 | 中山大学 | Face identification method based on aggregation Damage degree metric learning |
CN107330451A (en) * | 2017-06-16 | 2017-11-07 | 西交利物浦大学 | Clothes attribute retrieval method based on depth convolutional neural networks |
CN107247947A (en) * | 2017-07-07 | 2017-10-13 | 北京智慧眼科技股份有限公司 | Face character recognition methods and device |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110781866A (en) * | 2019-11-08 | 2020-02-11 | 成都大熊猫繁育研究基地 | Panda face image gender identification method and device based on deep learning |
CN111079624A (en) * | 2019-12-11 | 2020-04-28 | 北京金山云网络技术有限公司 | Method, device, electronic equipment and medium for collecting sample information |
CN111079624B (en) * | 2019-12-11 | 2023-09-01 | 北京金山云网络技术有限公司 | Sample information acquisition method and device, electronic equipment and medium |
CN116057549A (en) * | 2020-08-28 | 2023-05-02 | 爱你康控股株式会社 | Premium calculation system and premium calculation method |
CN112036520A (en) * | 2020-11-06 | 2020-12-04 | 四川师范大学 | Panda age identification method and device based on deep learning and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110414298B (en) | 2023-07-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110414298A (en) | A kind of more attribute joint recognition methods of monkey face | |
WO2020010785A1 (en) | Classroom teaching cognitive load measuring system | |
CN110647875B (en) | Method for segmenting and identifying model structure of blood cells and blood cell identification method | |
CN112581438B (en) | Slice image recognition method and device, storage medium and electronic equipment | |
CN107153844A (en) | The accessory system being improved to flowers identifying system and the method being improved | |
CN110543912B (en) | Method for automatically acquiring cardiac cycle video in fetal key section ultrasonic video | |
CN110135231A (en) | Animal face recognition methods, device, computer equipment and storage medium | |
CN109934182A (en) | Object behavior analysis method, device, electronic equipment and computer storage medium | |
CN110348293A (en) | A kind of commodity recognition method and system | |
CN111382727A (en) | Deep learning-based dog face identification method | |
CN113822907B (en) | Image processing method and device | |
CN117237351B (en) | Ultrasonic image analysis method and related device | |
CN116703837B (en) | MRI image-based rotator cuff injury intelligent identification method and device | |
CN115937661B (en) | 3D scene understanding method, system, electronic equipment and storage medium | |
CN112132137A (en) | FCN-SPP-Focal Net-based method for identifying correct direction of abstract picture image | |
Liang et al. | SPRNet: Automatic fetal standard plane recognition network for ultrasound images | |
CN111652837A (en) | AI-based thyroid nodule left and right lobe positioning and ultrasonic report error correction method | |
CN110188709A (en) | The detection method and detection system of oil drum in remote sensing image based on deep learning | |
CN111079617A (en) | Poultry identification method and device, readable storage medium and electronic equipment | |
CN113947780B (en) | Sika face recognition method based on improved convolutional neural network | |
He et al. | Body condition scoring network based on improved YOLOX | |
CN113673422A (en) | Pet type identification method and identification system | |
CN111464743A (en) | Photographic composition matching method and system | |
CN110633754A (en) | Intelligent medical record character recognition method based on neural network | |
CN115309941B (en) | AI-based intelligent tag retrieval method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |