CN108021863A - Electronic device, the character classification by age method based on image and storage medium - Google Patents

Electronic device, the character classification by age method based on image and storage medium Download PDF

Info

Publication number
CN108021863A
CN108021863A CN201711059224.XA CN201711059224A CN108021863A CN 108021863 A CN108021863 A CN 108021863A CN 201711059224 A CN201711059224 A CN 201711059224A CN 108021863 A CN108021863 A CN 108021863A
Authority
CN
China
Prior art keywords
age
facial image
fixed point
pixel
image
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
Application number
CN201711059224.XA
Other languages
Chinese (zh)
Inventor
戴磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Technology Shenzhen Co Ltd
Original Assignee
Ping An Technology Shenzhen Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ping An Technology Shenzhen Co Ltd filed Critical Ping An Technology Shenzhen Co Ltd
Priority to CN201711059224.XA priority Critical patent/CN108021863A/en
Publication of CN108021863A publication Critical patent/CN108021863A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00221Acquiring or recognising human faces, facial parts, facial sketches, facial expressions
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00221Acquiring or recognising human faces, facial parts, facial sketches, facial expressions
    • G06K9/00268Feature extraction; Face representation
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00221Acquiring or recognising human faces, facial parts, facial sketches, facial expressions
    • G06K9/00288Classification, e.g. identification
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/20Image acquisition
    • G06K9/34Segmentation of touching or overlapping patterns in the image field
    • G06K9/342Cutting or merging image elements, e.g. region growing, watershed, clustering-based techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00221Acquiring or recognising human faces, facial parts, facial sketches, facial expressions
    • G06K2009/00322Acquiring or recognising human faces, facial parts, facial sketches, facial expressions estimating age from face image; using age information for improving recognition

Abstract

Character classification by age method and storage medium the invention discloses a kind of electronic device, based on image, including:After first facial image at age to be identified is received, first facial image is cut using default tailoring rule, obtains the second facial image of default quantity;It is utilized respectively predetermined facial image identification model and age identification is carried out to the second facial image, generates the corresponding age characteristics vector of each second facial image;Each age characteristics vector is subjected to processing of averaging, obtains average age feature vector;Character classification by age analysis is carried out using predetermined character classification by age function pair average age feature vector, obtains the corresponding average age type of the second facial image, which is the corresponding age type of the first facial image.In this way, can both be effectively prevented from the problem of human face recognition model training process is computationally intensive, efficiency is low, the accuracy to character classification by age according to facial image can also be lifted.

Description

Electronic device, the character classification by age method based on image and storage medium
Technical field
The present invention relates to field of face identification, more particularly to a kind of electronic device, the character classification by age method based on image and Storage medium.
Background technology
In recent years, with the development of face recognition technology, identification demand of the user to facial image is also higher and higher, example Such as, in the network promotion age bracket of the corresponding people of the image is identified by identifying facial image, and according to the year identified Age section collects the welcome degree of the user of the user preferences of different age group and different age group to product etc..
At present, common face recognition technology, it is necessary to obtain more personal before character classification by age is carried out according to facial image The sample set that face image is formed carries out the classification of multiple-age bracket, and will meet the sample of current age section as positive sample collection, its The sample of his age bracket is trained as negative sample collection, generates the facial image identification that character classification by age is carried out according to facial image Model.And the usually corresponding negative sample number of a certain age bracket is the decades of times of positive sample number, positive sample number and negative sample number are uneven Weighing apparatus, causes the facial image identification model that training generates there are certain error, if by introducing some virtual positive sample numbers Or the division rule of change positive sample and negative sample comes further to the facial image identification model progress for character classification by age Training, to lift the recognition accuracy of facial image identification model, then can cause the training process calculation amount of human face recognition model Very greatly, the problem of efficiency is low.
The content of the invention
In view of this, the present invention proposes a kind of character classification by age method based on image, can be lifted according to facial image pair The accuracy of character classification by age, and the problem of human face recognition model training process is computationally intensive, efficiency is low can be effectively prevented from.
First, to achieve the above object, the present invention proposes a kind of electronic device, the electronic device include memory and The processor being connected with the memory, the processor are used to perform the age based on image point stored in the memory Class system, the character classification by age system based on image realize following steps when being performed by the processor:
Step is cut, after first facial image at age to be identified is received, using default tailoring rule by described the One facial image is cut, and obtains the second facial image of default quantity;
Age characteristics vector generation step, is utilized respectively predetermined facial image identification model to second face Image carries out age identification, generates the corresponding age characteristics vector of each second facial image, the predetermined people Face image identification model includes the generation network of age characteristics vector;
Average age feature vector generation step, carries out processing of averaging by each age characteristics vector, is put down Equal age characteristics vector;
Character classification by age step, the age is carried out using average age feature vector described in predetermined character classification by age function pair Classification analysis, obtains the corresponding average age type of second facial image, and the average age type is described first The corresponding age type of facial image.
Preferably, the facial image identification model includes the output with the generation net mate of age characteristics vector Layer, the output layer include the character classification by age function.
Preferably, the default tailoring rule includes:
Detect face scope and age characteristics that first facial image includes;
The corresponding pixel region of face is determined according to the face scope of detection, finds out the corresponding starting pixels of the pixel region And end pixel;
Fixed point pixel using the starting pixels as first default size crop box, since the starting pixels, along The abscissa direction of the starting pixels, the horizontal fixed point pixel of a default size crop box is determined every the first number of pixels, Laterally the pixel region is cut once using default size crop box at fixed point pixel at each, is obtained at least one described Second facial image, second facial image include at least one age characteristics.
Preferably, the default tailoring rule further includes:Since the starting pixels, along the vertical seat of the starting pixels Direction is marked, the longitudinal direction fixed point pixel of a default size crop box is determined every the second number of pixels;
Selection longitudinal direction pinpoints pixel one by one, and after one longitudinal direction fixed point pixel of selection, this is longitudinally pinpointed to pixel as pre- If the fixed point pixel of size crop box, since this longitudinally fixed point pixel, the abscissa direction of pixel is longitudinally pinpointed along this, often The horizontal fixed point pixel of a default size crop box is determined every the first number of pixels, is laterally utilized at each at fixed point pixel Default size crop box cuts once the pixel region, obtains at least one second facial image, second face Image includes at least one age characteristics.
Preferably, the predetermined facial image identification model includes convolutional neural networks model, and the age is special The generation network of sign vector is convolutional neural networks, and the age characteristics vector generation step includes being utilized respectively the convolution god The age characteristics included through network to second facial image is analyzed, and it is corresponding to generate each second facial image The age characteristics vector.
In addition, to achieve the above object, the present invention also provides a kind of character classification by age method based on image, this method includes Following steps:
A, after first facial image at age to be identified is received, using default tailoring rule by the first face figure As being cut, the second facial image of default quantity is obtained;
B, it is utilized respectively predetermined facial image identification model and age identification is carried out to second facial image, it is raw Into the corresponding age characteristics vector of each second facial image, the predetermined facial image identification model includes institute State the generation network of age characteristics vector;
C, each age characteristics vector is subjected to processing of averaging, obtains average age feature vector;
D, character classification by age analysis is carried out using average age feature vector described in predetermined character classification by age function pair, obtained To the corresponding average age type of second facial image, the average age type is that first facial image corresponds to Age type.
Preferably, the facial image identification model includes the output with the generation net mate of age characteristics vector Layer, the output layer include the character classification by age function.
Preferably, the default tailoring rule includes:Detect face scope and the year that first facial image includes Age feature;
The corresponding pixel region of face is determined according to the face scope of detection, finds out the corresponding starting pixels of the pixel region And end pixel;
Fixed point pixel using the starting pixels as first default size crop box, since the starting pixels, along The abscissa direction of the starting pixels, the horizontal fixed point pixel of a default size crop box is determined every the first number of pixels, Laterally the pixel region is cut once using default size crop box at fixed point pixel at each, is obtained at least one described Second facial image, second facial image include at least one age characteristics.
Preferably, the default tailoring rule further includes:
Since the starting pixels, along the ordinate direction of the starting pixels, one is determined every the second data pixels The longitudinal direction fixed point pixel of default size crop box;
Selection longitudinal direction pinpoints pixel one by one, and after one longitudinal direction fixed point pixel of selection, this is longitudinally pinpointed to pixel as pre- If the fixed point pixel of size crop box, since this longitudinally fixed point pixel, the abscissa direction of pixel is longitudinally pinpointed along this, often The horizontal fixed point pixel of a default size crop box is determined every the first number of pixels, is laterally utilized at each at fixed point pixel Default size crop box cuts once the pixel region, obtains at least one second facial image, second face Image includes at least one age characteristics.
Further, to achieve the above object, the present invention also provides a kind of computer-readable recording medium, the computer Readable storage medium storing program for executing is stored with the character classification by age system based on image, and the character classification by age system based on image can be by least one A processor performs, so that the step of at least one processor performs character classification by age method based on image described above.
Compared to the prior art, electronic device proposed by the invention, the character classification by age method based on image and storage are situated between Matter, first after first facial image at age to be identified is received, is cut out first face using default tailoring rule Cut, obtain the second facial image of default quantity;Then predetermined facial image identification model is utilized respectively to the second people Face image carries out age identification, generates the corresponding age characteristics vector of each second facial image;Then by each age characteristics Vector carries out processing of averaging, and obtains average vector;Finally carried out using predetermined character classification by age function pair average vector Character classification by age is analyzed, and obtains the corresponding average age type of the second facial image, which is the first facial image Corresponding age type.In this way, can both be effectively prevented from that human face recognition model training process is computationally intensive, efficiency is low asks Topic, can also lift the accuracy to character classification by age according to facial image.
Brief description of the drawings
Fig. 1 is the schematic diagram of the one optional hardware structure of electronic device of the present invention;
Fig. 2 is the Program modual graph of the character classification by age program preferred embodiment of the invention based on image;
Fig. 3 is the implementation process diagram of the character classification by age method preferred embodiment of the invention based on image.
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.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to the accompanying drawings and embodiments, it is right The present invention is further elaborated.It should be appreciated that specific embodiment described herein is only to explain the present invention, not For limiting the present invention.Based on the embodiments of the present invention, those of ordinary skill in the art are not before creative work is made All other embodiments obtained are put, belong to the scope of protection of the invention.
It should be noted that the description for being related to " first ", " second " etc. in the present invention is only used for description purpose, and cannot It is interpreted as indicating or implies its relative importance or imply the quantity of the technical characteristic indicated by indicating.Thus, define " the One ", at least one this feature can be expressed or be implicitly included to the feature of " second ".In addition, the skill between each embodiment Art scheme can be combined with each other, but must can be implemented as basis with those of ordinary skill in the art, when technical solution It will be understood that the combination of this technical solution is not present with reference to there is conflicting or can not realize when, also not in application claims Protection domain within.
As shown in fig.1, it is the schematic diagram of the 10 1 optional hardware structure of electronic device of the present invention.In the present embodiment, Electronic device 10 may include, but be not limited only to, can be in communication with each other by communication bus 14 connection memory 11, processor 12 and Display 13.It is pointed out that Fig. 1 illustrate only the electronic device 10 with component 11-13, it should be understood that simultaneously All components shown realistic are not applied, the more or less component of the implementation that can be substituted.
Wherein, memory 11 includes at least a type of computer-readable recording medium, computer-readable recording medium Including flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memories etc.), random access storage device (RAM), quiet State random access storage device (SRAM), read-only storage (ROM), electrically erasable programmable read-only memory (EEPROM), can compile Journey read-only storage (PROM), magnetic storage, disk, CD etc..In certain embodiments, memory 11 can be electronics dress Put 10 internal storage unit, such as the hard disk or memory of electronic device 10.In further embodiments, memory 11 can also It is the plug-in type hard disk being equipped with the External memory equipment of electronic device 10, such as electronic device 10, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card (Flash Card) etc..Certainly, store Device 11 can also both include the internal storage unit of electronic device 10 or including its External memory equipment.In the present embodiment, storage Device 11 is installed on the operating system and types of applications software of electronic device 10 commonly used in storage, for example, for storing based on figure Character classification by age program of picture etc..Export or will export each in addition, memory 11 can be also used for temporarily storing Class data.
Processor 12 can be in certain embodiments central processing unit (Central Processing Unit, CPU), Controller, microcontroller, microprocessor or other data processing chips.Processor 12 is commonly used in control electronic device 10 Overall operation.In the present embodiment, processor 12 is used to perform the program code stored in memory 11 or processing data, example Such as, for performing character classification by age program based on image stored in memory 11 etc..
Display 13 can be in certain embodiments light-emitting diode display, liquid crystal display, touch-control liquid crystal display and OLED (Organic Light-Emitting Diode, Organic Light Emitting Diode) touches device etc..Display 13 is used to be shown in The process of processing information and for showing visual user interface in electronic device 10, for example, being based on image for showing Character classification by age interface etc..
Communication bus 14 is used for realization the communication connection between modules
So far, oneself is through describing the hardware configuration and function of electronic device 10 proposed by the invention in detail.Need to illustrate , the character classification by age program based on image being stored in memory 11 is performed by processor 12, to realize that the present invention is each The step of character classification by age method based on image of embodiment.
In one embodiment, when the character classification by age program based on image is performed by processor 12, following steps are realized:
Step is cut, it is using default tailoring rule that this is the first if there is first facial image at age to be identified Face image is cut, and obtains the second facial image of default quantity;
Age characteristics vector generation step, is utilized respectively predetermined facial image identification model to the second facial image Age identification is carried out, generates the corresponding age characteristics vector of each second facial image, predetermined facial image identifies mould Type includes the generation network of age characteristics vector;
Average age feature vector generation step, carries out processing of averaging by each age characteristics vector, obtains average year Age feature vector;
Character classification by age step, character classification by age is carried out using predetermined character classification by age function pair average age feature vector Analysis, obtains the corresponding average age type of the second facial image, wherein, average age type is first facial image Corresponding age type.
Specifically, in the present embodiment, default tailoring rule includes:
Detect face scope and age characteristics that the first facial image includes;Face pair is determined according to the face scope of detection The pixel region answered, for example, the pixel region that the quadrangle of the minimum area comprising face determines is the corresponding pixel region of face Domain, finds out the corresponding starting pixels of the pixel region, for example, the pixel in the most upper left corner of the pixel region is the starting picture Element, and end pixel, for example, the pixel of the last cell of the pixel region is the end pixel;Using the starting pixels as First default size crop box, for example, the crop box of 60*60 pixels, fixed point pixel, for example, fixed point pixel refer to it is pre- If the vertex in the most upper left corner of size crop box, since the starting pixels, along the abscissa direction of the starting pixels, every First quantity, for example, 50, pixel determines the horizontal fixed point pixel of a default size crop box, for example, laterally fixed point pixel Refer to the vertex in the most upper left corner of default size crop box, default size crop box is laterally utilized at fixed point pixel at each The pixel region is cut once, obtains at least one second facial image, wherein, the second facial image is comprising at least one Age characteristics.Explanation is needed further exist for, age characteristics includes the distance between the colour of skin, face mask, each wheel of face Wide and each profile of face size, expression, color spot and wrinkle etc..
Alternatively, in another embodiment, default tailoring rule further includes:
Since the starting pixels, along the ordinate direction of the starting pixels, every the second quantity, for example, 60, as Element determines the longitudinal direction fixed point pixel of a default size crop box;Selection longitudinal direction fixed point pixel one by one, determines selecting a longitudinal direction After point pixel, using this, longitudinally fixed point pixel is as default size crop box, for example, the crop box of 60*60 pixels, fixed point picture Element, since this longitudinally fixed point pixel, the abscissa direction of pixel is longitudinally pinpointed along this, every the first quantity, for example, 50 A, pixel determines the horizontal fixed point pixel of a default size crop box, at each laterally at fixed point pixel using default big Small crop box cuts once the pixel region, obtains at least one second facial image, wherein, the second facial image includes At least one age characteristics.
It should be noted that presetting the corresponding size of size crop box needs to know than predetermined human face recognition model The size of other picture is big, and in one embodiment, presetting the corresponding size of size crop box needs to know than predetermined face The size big 15% to 25% of the other identifiable picture of model, and when cutting every time, what the second facial image for being cropped to included Age characteristics difference is more notable better, and the number of cutting is not easy excessively, also unsuitable very few, if cutting number is excessive, can increase Add calculation amount, the inefficiency for causing the age to be predicted, if cutting number is very few, can cause to put forward age forecasting accuracy Rise unobvious.Experimental data shows that the general number that cuts is advisable for 6 to 12 times.
Further, in the present embodiment, predetermined facial image identification model includes convolutional neural networks model, The generation network of age characteristics vector is convolutional neural networks, and age characteristics vector generation step includes being utilized respectively convolutional Neural The age characteristics that network includes the second facial image is analyzed, and it is special to generate each second facial image corresponding age Sign vector.
Next, each age characteristics vector is carried out processing of averaging, average age feature vector is obtained.
Then, using predetermined character classification by age function, for example, in the present embodiment, predetermined classification function For Softmax classification functions, character classification by age analysis is carried out to average age feature vector, it is corresponding flat to obtain the second facial image Equal age type, wherein, average age type is the corresponding age type of the first facial image, it is necessary to illustrate, in this reality Applying age type in example includes infancy, infancy, adolescence, midlife, senescence phase etc..
By above-described embodiment, electronic device proposed by the present invention, is receiving the first face figure at age to be identified As after, first facial image is cut using default tailoring rule, obtains the second facial image of default quantity; It is utilized respectively predetermined facial image identification model and age identification is carried out to the second facial image, generates each second face The corresponding age characteristics vector of image, predetermined facial image identification model include the generation network of age characteristics vector; Each age characteristics vector is subjected to processing of averaging, obtains average age feature vector;Utilize predetermined character classification by age Function pair average age feature vector carries out character classification by age analysis, obtains the corresponding average age type of the second facial image, puts down Equal age type is the corresponding age type of first facial image.Both human face recognition model can be effectively prevented to train The problem of journey is computationally intensive, efficiency is low, can also lift the accuracy to character classification by age according to facial image.
Further, in one embodiment, realized according to the character classification by age program based on image according to its each several part Function is different, can be divided into one or more virtual program modules.
As shown in fig.2, the program module signal of the character classification by age program preferred embodiment based on image for the present invention Figure.As shown in Figure 2, in the present embodiment, the function difference realized according to the character classification by age program each several part based on image, base It is divided into the character classification by age program of image and cuts module 201, age characteristics vector generation module 202, average age feature Vector generation module 203 and character classification by age module 204, the functions or operations that wherein module 202-204 is realized with above It is similar, no longer it is described in detail herein, exemplarily, such as wherein:
Module 201 is cut, if for there is first facial image at age to be identified, will using default tailoring rule First facial image is cut, and obtains the second facial image of default quantity;
Age characteristics vector generation module 202, for being utilized respectively predetermined facial image identification model to second Facial image carries out age identification, generates the corresponding age characteristics vector of each second facial image, wherein, predetermined people Face image identification model includes the generation network of age characteristics vector;
Average age feature vector generation module 203, for each age characteristics vector to be carried out processing of averaging, obtains Average age feature vector;
Character classification by age module 204, for being carried out using predetermined character classification by age function pair average age feature vector Character classification by age is analyzed, and obtains the corresponding average age type of the second facial image, then the average age type obtained is first The corresponding age type of facial image.
In addition, the present invention also provides a kind of character classification by age method based on image.As shown in fig.3, the base for the present invention In the implementing procedure figure of the character classification by age method preferred embodiment of image.This method can be performed by a device, which can With by software and/or hardware realization.
From the figure 3, it may be seen that this is in the present embodiment, the character classification by age method based on image includes step S301 to step S304。
Step S301, it is using default tailoring rule that this is the first if there is first facial image at age to be identified Face image is cut, and obtains the second facial image of default quantity;
Step S302, is utilized respectively predetermined facial image identification model and carries out age knowledge to the second facial image Not, the corresponding age characteristics vector of each second facial image is generated, wherein, predetermined facial image identification model includes The generation network of age characteristics vector;
Step S303, carries out processing of averaging by each age characteristics vector, obtains average age feature vector;
Step S304, character classification by age point is carried out using predetermined character classification by age function pair average age feature vector Analysis, obtains the corresponding average age type of the second facial image, then the average age type obtained is the first face figure As corresponding age type.
Specifically, in the present embodiment, default tailoring rule includes:
Detect face scope and age characteristics that the first facial image includes;Face pair is determined according to the face scope of detection The pixel region answered, for example, the pixel region that the quadrangle of the minimum area comprising face determines is the corresponding pixel region of face Domain, finds out the corresponding starting pixels of the pixel region, for example, the pixel in the most upper left corner of the pixel region is the starting picture Element, and end pixel, for example, the pixel of the last cell of the pixel region is the end pixel;Using the starting pixels as First default size crop box, for example, the crop box of 60*60 pixels, fixed point pixel, for example, fixed point pixel refer to it is pre- If the vertex in the most upper left corner of size crop box, since the starting pixels, along the abscissa direction of the starting pixels, every First quantity, for example, 50, pixel determines the horizontal fixed point pixel of a default size crop box, for example, laterally fixed point pixel Refer to the vertex in the most upper left corner of default size crop box, default size crop box is laterally utilized at fixed point pixel at each The pixel region is cut once, obtains at least one second facial image, wherein, the second facial image is comprising at least one Age characteristics.Explanation is needed further exist for, age characteristics includes the distance between the colour of skin, face mask, each wheel of face Wide and each profile of face size, expression, color spot and wrinkle etc..
Alternatively, in another embodiment, default tailoring rule further includes:
Since the starting pixels, along the ordinate direction of the starting pixels, every the second quantity, for example, 60, as Element determines the longitudinal direction fixed point pixel of a default size crop box;Selection longitudinal direction fixed point pixel one by one, determines selecting a longitudinal direction After point pixel, using this, longitudinally fixed point pixel is as default size crop box, for example, the crop box of 60*60 pixels, fixed point picture Element, since this longitudinally fixed point pixel, the abscissa direction of pixel is longitudinally pinpointed along this, every the first quantity, for example, 50 A, pixel determines the horizontal fixed point pixel of a default size crop box, at each laterally at fixed point pixel using default big Small crop box cuts once the pixel region, obtains at least one second facial image, wherein, the second facial image includes At least one age characteristics.
It should be noted that presetting the corresponding size of size crop box needs to know than predetermined human face recognition model The size of other picture is big, and in one embodiment, presetting the corresponding size of size crop box needs to know than predetermined face The size big 15% to 25% of the other identifiable picture of model, and when cutting every time, what the second facial image for being cropped to included Age characteristics difference is more notable better, and the number of cutting is not easy excessively, also unsuitable very few, if cutting number is excessive, can increase Add calculation amount, the inefficiency for causing the age to be predicted, if cutting number is very few, can cause to put forward age forecasting accuracy Rise unobvious.Experimental data shows that the general number that cuts is advisable for 6 to 12 times.
Further, in the present embodiment, predetermined facial image identification model includes convolutional neural networks model, The generation network of age characteristics vector is convolutional neural networks, and age characteristics vector generation step includes being utilized respectively convolutional Neural The age characteristics that network includes the second facial image is analyzed, and it is special to generate each second facial image corresponding age Sign vector.
Next, each age characteristics vector is carried out processing of averaging, average age feature vector is obtained.
Then, using predetermined character classification by age function, for example, in the present embodiment, predetermined classification function For Softmax classification functions, character classification by age analysis is carried out to average age feature vector, it is corresponding flat to obtain the second facial image Equal age type, wherein, average age type is the corresponding age type of the first facial image, it is necessary to illustrate, in this reality Applying age type in example includes infancy, infancy, adolescence, midlife, senescence phase etc..
By above-described embodiment, the character classification by age method proposed by the present invention based on image, is receiving year to be identified After first facial image in age, first facial image is cut using default tailoring rule, obtains default quantity The second facial image;It is utilized respectively predetermined facial image identification model and age identification is carried out to the second facial image, The corresponding age characteristics vector of each second facial image is generated, predetermined facial image identification model includes age characteristics The generation network of vector;Each age characteristics vector is subjected to processing of averaging, obtains average age feature vector;Using advance Definite character classification by age function pair average age feature vector carries out character classification by age analysis, and it is corresponding flat to obtain the second facial image Equal age type, average age type is the corresponding age type of first facial image.Both face can be effectively prevented from The problem of identification model training process is computationally intensive, efficiency is low, can also be lifted according to facial image to the accurate of character classification by age Degree.
In addition, the embodiment of the present invention also proposes a kind of computer-readable recording medium, on the computer-readable recording medium The character classification by age program based on image is stored with, is somebody's turn to do and following step is realized when the character classification by age program based on image is executed by processor Suddenly:
Step is cut, will be the first using default tailoring rule after first facial image at age to be identified is received Face image is cut, and obtains the second facial image of default quantity;
Age characteristics vector generation step, is utilized respectively predetermined facial image identification model to the second facial image Age identification is carried out, generates the corresponding age characteristics vector of each second facial image, wherein, predetermined facial image is known Other model includes the generation network of age characteristics vector;
Average age feature vector generation step, carries out processing of averaging by each age characteristics vector, obtains average year Age feature vector;
Character classification by age step, character classification by age is carried out using predetermined character classification by age function pair average age feature vector Analysis, obtains the corresponding average age type of the second facial image, wherein, average age type is corresponding for the first facial image Age type.
The computer-readable recording medium embodiment of the present invention age with above-mentioned electronic device and based on image point Each embodiment of class method is essentially identical, does not make tired state herein.
By above-described embodiment, the character classification by age method proposed by the present invention based on image can be both effectively prevented from The problem of human face recognition model training process is computationally intensive, efficiency is low, can also be lifted according to facial image to character classification by age Accuracy.
The embodiments of the present invention are for illustration only, do not represent the quality of embodiment.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side Method can add the mode of required general hardware platform to realize by software, naturally it is also possible to by hardware, but in many cases The former is more preferably embodiment.Based on such understanding, technical scheme substantially in other words does the prior art Going out the part of contribution can be embodied in the form of software product, which is stored in a storage medium In (such as ROM/RAM, magnetic disc, CD), including some instructions are used so that a station terminal equipment (can be mobile phone, computer, takes Be engaged in device, air conditioner, or network equipment etc.) perform method described in each embodiment of the present invention.
It these are only the preferred embodiment of the present invention, be not intended to limit the scope of the invention, it is every to utilize this hair The equivalent structure or equivalent flow shift that bright specification and accompanying drawing content are made, is directly or indirectly used in other relevant skills Art field, is included within the scope of the present invention.

Claims (10)

1. a kind of electronic device, it is characterised in that the electronic device includes memory and the processing being connected with the memory Device, the processor be used to performing the character classification by age program based on image that is stored in the memory, described based on image Character classification by age program realizes following steps when being performed by the processor:
Step is cut, will be described the first using default tailoring rule after first facial image at age to be identified is received Face image is cut, and obtains the second facial image of default quantity;
Age characteristics vector generation step, is utilized respectively predetermined facial image identification model to second facial image Age identification is carried out, generates the corresponding age characteristics vector of each second facial image, the predetermined face figure As identification model includes the generation network of age characteristics vector;
Average age feature vector generation step, carries out processing of averaging by each age characteristics vector, obtains average year Age feature vector;
Character classification by age step, character classification by age is carried out using average age feature vector described in predetermined character classification by age function pair Analysis, obtains the corresponding average age type of second facial image, the average age type is first face The corresponding age type of image.
2. electronic device as claimed in claim 1, it is characterised in that the facial image identification model includes and the age The output layer of the generation net mate of feature vector, the output layer include the character classification by age function.
3. electronic device as claimed in claim 2, it is characterised in that the default tailoring rule includes:
Detect face scope and age characteristics that first facial image includes;
The corresponding pixel region of face is determined according to the face scope of detection, finds out the corresponding starting pixels of the pixel region and knot Beam pixel;
Fixed point pixel using the starting pixels as first default size crop box, since the starting pixels, rises along this The abscissa direction of beginning pixel, determines the horizontal fixed point pixel of a default size crop box, every every the first number of pixels One laterally cuts once the pixel region using default size crop box at fixed point pixel, obtains at least one described second Facial image, second facial image include at least one age characteristics.
4. electronic device as claimed in claim 3, it is characterised in that the default tailoring rule further includes:
Since the starting pixels, along the ordinate direction of the starting pixels, determine that one is preset every the second number of pixels The longitudinal direction fixed point pixel of size crop box;
Selection longitudinal direction fixed point pixel one by one, after a longitudinal direction fixed point pixel is selected, using this, longitudinally fixed point pixel is big as presetting The fixed point pixel of small crop box, since this longitudinally fixed point pixel, the abscissa direction of pixel, Mei Ge are longitudinally pinpointed along this One number of pixels determines the horizontal fixed point pixel of a default size crop box, at each laterally at fixed point pixel using default Size crop box cuts once the pixel region, obtains at least one second facial image, second facial image Include at least one age characteristics.
5. electronic device as claimed in claim 4, it is characterised in that the predetermined facial image identification model includes Convolutional neural networks model, the generation network of the age characteristics vector is convolutional neural networks, and the age characteristics vector is raw Analyzed into step including being utilized respectively the age characteristics that the convolutional neural networks include second facial image, it is raw Into the corresponding age characteristics vector of each second facial image.
A kind of 6. character classification by age method based on image, it is characterised in that described method includes following steps:
A, after first facial image at age to be identified is received, using default tailoring rule by first facial image into Row is cut, and obtains the second facial image of default quantity;
B, it is utilized respectively predetermined facial image identification model and age identification is carried out to second facial image, generation is each The corresponding age characteristics vector of a second facial image, the predetermined facial image identification model include the year The generation network of age feature vector;
C, each age characteristics vector is subjected to processing of averaging, obtains average age feature vector;
D, character classification by age analysis is carried out using average age feature vector described in predetermined character classification by age function pair, obtains institute The corresponding average age type of the second facial image is stated, the average age type is first facial image corresponding year Age type.
7. the character classification by age method based on image as claimed in claim 6, it is characterised in that the facial image identification model Include the output layer of the generation net mate with age characteristics vector, the output layer includes the character classification by age function.
8. the character classification by age method based on image as claimed in claim 7, it is characterised in that the default tailoring rule bag Include:
Detect face scope and age characteristics that first facial image includes;
The corresponding pixel region of face is determined according to the face scope of detection, finds out the corresponding starting pixels of the pixel region and knot Beam pixel;
Fixed point pixel using the starting pixels as first default size crop box, since the starting pixels, rises along this The abscissa direction of beginning pixel, determines the horizontal fixed point pixel of a default size crop box, every every the first number of pixels One laterally cuts once the pixel region using default size crop box at fixed point pixel, obtains at least one described second Facial image, second facial image include at least one age characteristics.
9. the character classification by age method based on image as claimed in claim 8, it is characterised in that the default tailoring rule is also Including:
Since the starting pixels, along the ordinate direction of the starting pixels, determine that one is preset every the second data pixels The longitudinal direction fixed point pixel of size crop box;
Selection longitudinal direction fixed point pixel one by one, after a longitudinal direction fixed point pixel is selected, using this, longitudinally fixed point pixel is big as presetting The fixed point pixel of small crop box, since this longitudinally fixed point pixel, the abscissa direction of pixel, Mei Ge are longitudinally pinpointed along this One number of pixels determines the horizontal fixed point pixel of a default size crop box, at each laterally at fixed point pixel using default Size crop box cuts once the pixel region, obtains at least one second facial image, second facial image Include at least one age characteristics.
10. a kind of computer-readable recording medium, the computer-readable recording medium storage has the character classification by age based on image Program, the character classification by age program based on image can be performed by least one processor, so that at least one processor The step of performing the character classification by age method based on image as any one of claim 6-9.
CN201711059224.XA 2017-11-01 2017-11-01 Electronic device, the character classification by age method based on image and storage medium Pending CN108021863A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711059224.XA CN108021863A (en) 2017-11-01 2017-11-01 Electronic device, the character classification by age method based on image and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201711059224.XA CN108021863A (en) 2017-11-01 2017-11-01 Electronic device, the character classification by age method based on image and storage medium
PCT/CN2018/076182 WO2019085338A1 (en) 2017-11-01 2018-02-10 Electronic apparatus, image-based age classification method and system, and storage medium

Publications (1)

Publication Number Publication Date
CN108021863A true CN108021863A (en) 2018-05-11

Family

ID=62079656

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711059224.XA Pending CN108021863A (en) 2017-11-01 2017-11-01 Electronic device, the character classification by age method based on image and storage medium

Country Status (2)

Country Link
CN (1) CN108021863A (en)
WO (1) WO2019085338A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111626303A (en) * 2020-05-29 2020-09-04 南京甄视智能科技有限公司 Sex and age identification method, sex and age identification device, storage medium and server

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7912246B1 (en) * 2002-10-28 2011-03-22 Videomining Corporation Method and system for determining the age category of people based on facial images
EP3074928A4 (en) * 2013-11-25 2017-11-15 Ehsan Fazl Ersi System and method for face recognition
CN105303149B (en) * 2014-05-29 2019-11-05 腾讯科技(深圳)有限公司 The methods of exhibiting and device of character image
CN105718869B (en) * 2016-01-15 2019-07-02 网易(杭州)网络有限公司 The method and apparatus of face face value in a kind of assessment picture
CN106485235B (en) * 2016-10-24 2019-05-03 厦门美图之家科技有限公司 A kind of convolutional neural networks generation method, age recognition methods and relevant apparatus

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111626303A (en) * 2020-05-29 2020-09-04 南京甄视智能科技有限公司 Sex and age identification method, sex and age identification device, storage medium and server
CN111626303B (en) * 2020-05-29 2021-04-13 南京甄视智能科技有限公司 Sex and age identification method, sex and age identification device, storage medium and server

Also Published As

Publication number Publication date
WO2019085338A1 (en) 2019-05-09

Similar Documents

Publication Publication Date Title
US10657652B2 (en) Image matting using deep learning
US10984295B2 (en) Font recognition using text localization
US9349076B1 (en) Template-based target object detection in an image
US10699166B2 (en) Font attributes for font recognition and similarity
CN106097353B (en) Method for segmenting objects and device, computing device based on the fusion of multi-level regional area
US10580179B2 (en) Method and apparatus for processing video image and electronic device
US10977523B2 (en) Methods and apparatuses for identifying object category, and electronic devices
US20140119642A1 (en) Segmenting Human Hairs and Faces
CN103839250B (en) The method and apparatus processing for face-image
US7457432B2 (en) Specified object detection apparatus
CN104616021B (en) Traffic sign image processing method and device
EP2902941A1 (en) System and method for visually distinguishing faces in a digital image
US10635890B2 (en) Facial recognition method and apparatus, electronic device, and storage medium
US9418304B2 (en) System and method for recognizing text information in object
CN107798299B (en) Bill information identification method, electronic device and readable storage medium
US10210418B2 (en) Object detection system and object detection method
KR20190122907A (en) Fine-grained categorization
JP2014215852A (en) Image process device, program and image process method
US20100103197A1 (en) Method for adjusting font size on screen
CN107194398B (en) Vehicle damages recognition methods and the system at position
US10817741B2 (en) Word segmentation system, method and device
CN108229299A (en) The recognition methods of certificate and device, electronic equipment, computer storage media
US9824304B2 (en) Determination of font similarity
CN106709475B (en) Obstacle recognition method and device, computer equipment and readable storage medium
CN107016387B (en) Method and device for identifying label

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