CN110210393A - The detection method and device of facial image - Google Patents
The detection method and device of facial image Download PDFInfo
- Publication number
- CN110210393A CN110210393A CN201910472084.1A CN201910472084A CN110210393A CN 110210393 A CN110210393 A CN 110210393A CN 201910472084 A CN201910472084 A CN 201910472084A CN 110210393 A CN110210393 A CN 110210393A
- Authority
- CN
- China
- Prior art keywords
- face
- image
- human face
- sample image
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Abstract
The embodiment of the present application discloses detection method, device, electronic equipment and the computer-readable medium of facial image.One specific embodiment of this method includes: to obtain the picture frame comprising face object；Face datection is carried out to picture frame, orients human face region；Use trained change face detection model to face object included in the human face region oriented whether for synthesis forgery face detect.The embodiment can effectively detect the forgery facial image generated based on the face in nerual network technique replacement image.
Description
Technical field
The invention relates to field of computer technology, and in particular to field of artificial intelligence more particularly to face
The detection method and device of image.
Background technique
Face switching technology is the technology for replacing face in an image or a video, and generalling use artificial intelligence technology will scheme
The face of a people replaces with the face of another person automatically in picture or video, these images or video can not be by human eyes accurately
Identify.If it is tight to generate portraiture right infringement, stroll spoofing, initiation public accident etc. by improper use for this technology
Weight consequence.
Summary of the invention
The embodiment of the present application proposes detection method, device, electronic equipment and the computer-readable medium of facial image.
In a first aspect, embodiment of the disclosure provides a kind of detection method of facial image, comprising: obtaining includes face
The picture frame of object；Face datection is carried out to picture frame, orients human face region；Using the detection model of changing face trained to fixed
Whether face object included in the human face region that position goes out is that the forgery face synthesized is detected.
In some embodiments, above-mentioned that Face datection is carried out to picture frame, orient human face region, comprising: to video frame
Carry out face critical point detection, face key area is oriented based on the face key point detected, by video frame to face
Key area extends to the outside the region of preset range as the human face region oriented.
In some embodiments, the above-mentioned detection model of changing face trained is obtained by executing following first training operation
: first sample image collection is obtained, the first sample image in first sample image collection includes real human face image and conjunction
At facial image, synthesizing facial image is that the face object in image is replaced with the image generated after another face object；
To first sample image into face critical point detection is carried out, face key area is oriented based on the face key point detected,
The region for extending to the outside preset range in first sample image to face key area is positioned as from first sample image
Human face region out；Based on the human face region oriented from first sample image to the image classification neural network constructed into
Row is trained, and is in the face object that image classification neural network is included to the human face region oriented from first sample image
Deconditioning when the prediction error of the no forgery face for synthesis reaches preset first condition of convergence, the image that training is completed
What Classification Neural was determined as having trained change face detection model.
In some embodiments, above-mentioned first training operation further include: by the real human face in first sample image collection
Image and synthesis facial image add after carrying out geometric transformation respectively as new real human face image and new synthesis facial image
It adds in first sample image collection.
In some embodiments, above-mentioned detection model of changing face includes disaggregated model；The above method further include: use has been trained
Human face recognition model extract the feature of human face region oriented；And the detection model of changing face that above-mentioned use has been trained is to fixed
Whether face object included in the human face region that position goes out is that the forgery face synthesized is detected, comprising: is known based on face
The feature for the human face region that other model extraction goes out, using detection model of changing face to face included in the human face region oriented
Whether object is that the forgery face synthesized is classified.
In some embodiments, above-mentioned detection model of changing face is obtained by following second training operation: obtaining second
Sample image set, the second sample image in the second sample image set include real human face image and synthesis facial image,
Synthesis facial image is that the face object in image is replaced with the image generated after another face object；To the second sample graph
As carrying out Face datection, the human face region of the second sample image is oriented；Using the human face recognition model trained to the second sample
The human face region of this image carries out feature extraction；The disaggregated model constructed is trained based on the feature extracted, is being divided
Whether face object included in human face region of the class model to the second sample image is that the prediction for forging face synthesized misses
Deconditioning when difference reaches preset second condition of convergence, the disaggregated model completed based on training generate the detection of changing face trained
Model.
Second aspect, embodiment of the disclosure provide a kind of detection device of facial image, comprising: acquiring unit, quilt
It is configured to obtain the picture frame comprising face object；Positioning unit is configured as carrying out Face datection to picture frame, orients people
Face region；Detection unit is configured as using the detection model of changing face trained to included in the human face region oriented
Whether face object is that the forgery face synthesized is detected.
In some embodiments, above-mentioned positioning unit is configured to carry out face to picture frame as follows
Detection, orients human face region: carrying out face critical point detection to video frame, orients people based on the face key point detected
Face key area, using the region for extending to the outside preset range in video frame to face key area as the face area oriented
Domain.
In some embodiments, the above-mentioned detection model of changing face trained is obtained by executing following first training operation
: first sample image collection is obtained, the first sample image in first sample image collection includes real human face image and conjunction
At facial image, synthesizing facial image is that the face object in image is replaced with the image generated after another face object；
To first sample image into face critical point detection is carried out, face key area is oriented based on the face key point detected,
The region for extending to the outside preset range in first sample image to face key area is positioned as from first sample image
Human face region out；Based on the human face region oriented from first sample image to the image classification neural network constructed into
Row is trained, and is in the face object that image classification neural network is included to the human face region oriented from first sample image
Deconditioning when the prediction error of the no forgery face for synthesis reaches preset first condition of convergence, the image that training is completed
What Classification Neural was determined as having trained change face detection model.
In some embodiments, above-mentioned first training operation further include: by the real human face in first sample image collection
Image and synthesis facial image add after carrying out geometric transformation respectively as new real human face image and new synthesis facial image
It adds in first sample image collection.
In some embodiments, above-mentioned detection model of changing face includes disaggregated model；Above-mentioned apparatus further include: extraction unit,
It is configured as extracting the feature for the human face region oriented using the human face recognition model trained；And above-mentioned detection unit into
Whether one step is configured as being as follows the forgery synthesized to face object included in the human face region oriented
Face is detected: the feature based on the human face region that human face recognition model extracts, using detection model of changing face to orienting
Human face region included in face object whether be synthesis forgery face classify.
In some embodiments, above-mentioned detection model of changing face is obtained by following second training operation: obtaining second
Sample image set, the second sample image in the second sample image set include real human face image and synthesis facial image,
Synthesis facial image is that the face object in image is replaced with the image generated after another face object；To the second sample graph
As carrying out Face datection, the human face region of the second sample image is oriented；Using the human face recognition model trained to the second sample
The human face region of this image carries out feature extraction；The disaggregated model constructed is trained based on the feature extracted, is being divided
Whether face object included in human face region of the class model to the second sample image is that the prediction for forging face synthesized misses
Deconditioning when difference reaches preset second condition of convergence, the disaggregated model completed based on training generate the detection of changing face trained
Model.
The third aspect, embodiment of the disclosure provide a kind of electronic equipment, comprising: one or more processors；Storage
Device, for storing one or more programs, when one or more programs are executed by one or more processors so that one or
Multiple processors realize the detection method of the facial image provided such as first aspect.
Fourth aspect, embodiment of the disclosure provide a kind of computer-readable medium, are stored thereon with computer program,
Wherein, the detection method for the facial image that first aspect provides is realized when program is executed by processor.
The detection method and device of the facial image of above-described embodiment of the disclosure, electronic equipment and computer-readable Jie
Matter, by obtain include face object picture frame, Face datection is carried out to picture frame, human face region is oriented, using having instructed
Whether experienced detection model of changing face is that the forgery face synthesized carries out to face object included in the human face region oriented
Detection, realize the accurate detection of " changing face ", can effectively detect based on nerual network technique replace image in face and
The forgery image of generation.
Detailed description of the invention
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, the application's is other
Feature, objects and advantages will become more apparent upon:
Fig. 1 is that the embodiment of the present application can be applied to exemplary system architecture figure therein；
Fig. 2 is the flow chart according to one embodiment of the detection method of the facial image of the application；
Fig. 3 is the flow chart according to another embodiment of the detection method of the facial image of the application；
Fig. 4 is the flow chart according to another embodiment of the detection method of the facial image of the application；
Fig. 5 is the structural schematic diagram of one embodiment of the detection device of the facial image of the application；
Fig. 6 is adapted for the structural schematic diagram for the computer system for realizing the electronic equipment of the embodiment of the present application.
Specific embodiment
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to
Convenient for description, part relevant to related invention is illustrated only in attached drawing.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.The application is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
Fig. 1 is shown can be using the example of the detection device of the detection method or facial image of the facial image of the application
Property system architecture.
As shown in Figure 1, may include terminal device 101,102,103, network 104 and server in system architecture 100
105.Network 104 between terminal device 101,102,103 and server 105 to provide the medium of communication link.Network can
To include various connection types, such as wired, wireless communication link or fiber optic cables etc..
Terminal device 101,102,103 can be the electronic equipment with display screen, can be smart phone, notebook electricity
Brain, desktop computer, tablet computer, smartwatch, etc..Various network moneys can be installed on terminal device 101,102,103
Source application, such as audio and video playing application, information client, browser application, etc..User can be used terminal device 101,
102,103 Internet resources, image, video in browse network, etc. are accessed.
Server 105 can be the content shown for terminal device 101,102,103 and provide the server of back-office support.Clothes
The resource access request or data analysis that business device 105 can be sent by 104 receiving terminal apparatus 101,102,103 of network are asked
It asks, search related resource or feeds back to terminal device 101,102,103 after obtaining processing result to Data Analysis Services.
In the application scenarios of the disclosure, server 105 can be to provide the clothes that true and false facial image/video identifies service
Business device.Server 105 can be to receiving or from database from terminal device 101,102,103 or other servers
Whether the face in image or video found out is to be identified by the synthesis face that operation generates of changing face.Server 105
Identification result can be fed back to terminal device 101,102,103, user can know mirror by terminal device 101,102,103
Other result.
In some scenes, terminal device 101,102,103 can also run true and false facial image/video evaluator,
It whether is to identify by the synthesis face that operation generates of changing face to the face in image or video.
It should be noted that server 105 can be hardware, it is also possible to software.It, can when server 105 is hardware
To be implemented as the distributed server cluster that multiple servers form, individual server also may be implemented into.When server 105 is
When software, multiple softwares or software module may be implemented into (such as providing multiple softwares of Distributed Services or software mould
Block), single software or software module also may be implemented into.It is not specifically limited herein.
Above-mentioned terminal device 101,102,103 is also possible to software.It, can when terminal device 101,102,103 is software
To be mounted in above-mentioned cited electronic equipment.Multiple softwares or software module may be implemented into (such as providing point in it
The multiple softwares or software module of cloth service), single software or software module also may be implemented into.Specific limit is not done herein
It is fixed.
It should be noted that the detection method of facial image provided by embodiment of the disclosure can be by terminal device
101,102,103 or server 105 execute, correspondingly, the detection device of facial image can be set in terminal device 101,
102,103 or server 105.
It should be understood that the terminal device, network, the number of server in Fig. 1 are only schematical.According to realization need
It wants, can have any number of terminal device, network, server.
With continued reference to Fig. 2, it illustrates the processes according to one embodiment of the detection method of the facial image of the application
200.The detection method of the facial image, comprising the following steps:
Step 201, the picture frame comprising face object is obtained.
In the present embodiment, the available image comprising face object of the executing subject of the detection method of facial image
Frame, as picture frame to be detected.The video resource that who object can be obtained from database, extracts from video resource
Picture frame comprising face object；Also in the video to be detected of other available electronic equipments comprising face object
Picture frame is as picture frame to be detected.
The above-mentioned picture frame comprising face object can also be the picture frame that user specifies, such as the image mirror that user submits
The picture frame comprising face object specified in asking is not invited.
In practice, user can initiate facial image to terminal device or whether video passes through the adulterator for generation of changing face
The identification of face image or video is requested, and uploads facial image or video to be identified.Above-mentioned executing subject can be by upload
Facial image is as picture frame to be detected, or can extract one or more picture frames from the face video of upload and make
For picture frame to be detected.
Step 202, Face datection is carried out to picture frame, orients human face region.
Other than face object, other objects, such as other body of people can also be included in above-mentioned picture frame
Position (such as neck, four limbs, trunk).In the present embodiment, above-mentioned executing subject can include face object to what is got
Picture frame carries out Face datection, to orient human face region.Specifically, the feature in picture frame, creation detection can be extracted
Whether window uses classifier to judge the object in detection window for face object based on the feature extracted.Alternatively, can examine
The point for meeting face key point (such as canthus, mouth, nose, eyebrow) feature in image is measured, then by the way of returning
Determine the boundary of human face region.Or it is also based on the phase between features of skin colors and head sizes, face key point
The boundary for orienting face from picture frame to position feature, to orient human face region.
Step 203, what use had been trained changes face detection model to face object included in the human face region oriented
It whether is that the forgery face synthesized is detected.
It can be syncopated as human face region image from the picture frame that step 201 obtains, the human face region image that will be syncopated as
The detection model of changing face trained is inputted to be detected.Wherein, detection model of changing face can predict the face in the image of input
Whether object is the forgery face of synthesis, or can determine that the face object in the image inputted is the forgery face synthesized
Probability value.Herein, the forgery face of synthesis can be the face object in image or video through neural network model
Replace with the face object in the image or video generated after another face object.
The above-mentioned detection model of changing face trained can be the set based on the facial image with true and false label and train
's.Wherein, it is to carry out Image Acquisition generation to true face that true and false label, which is used to mark the face object in facial image,
Or by operation (the face object in image is replaced with to the operation of the face object of another person) generation of changing face.
Facial image in above-mentioned set may come from multiple and different face objects.In practice, it can collect more
The facial image of a different people is changed face as the facial image with label " real human face ", and using neural network based
Method carries out operation of changing face to the facial image being collected into, and generates the facial image with label " forging face ".
It in the training process, can be based on to be trained detection model of changing face to the true and false of the face object in facial image
Difference iteration between testing result and corresponding true and false label adjusts the parameter for detection model of changing face, so that detection of changing face
The prediction error of model in an iterative process is gradually reduced to certain section.
The above-mentioned detection model of changing face trained, which can be based on neural network framework, to be constructed, such as VGG,
The neural networks framework such as RestNet.It is synthesis that the output of the detection model of changing face, which can be the face object in input picture,
The classification results for the forgery the face whether probability or the face object in input picture for forging face synthesize.
The detection method of the facial image of the above embodiments of the present application is right by obtaining the picture frame comprising face object
Picture frame carries out Face datection, orients human face region, what use had been trained changes face detection model to the human face region oriented
Included in face object whether be synthesis forgery face detected, realize the accurate detection of " changing face ", Neng Gouyou
Effect detects the forgery facial image generated based on the face in nerual network technique replacement image.
With continued reference to Fig. 3, it illustrates the streams according to another embodiment of the detection method of the facial image of the application
Cheng Tu.As shown in figure 3, the process 300 of the detection method of the facial image of the present embodiment, comprising the following steps:
Step 301, the picture frame comprising face object is obtained.
In the present embodiment, the available image comprising face object of the executing subject of the detection method of facial image
Frame, as picture frame to be detected.The picture frame comprising face object can be specifically extracted from database, or can receive
The picture frame comprising face object in the video to be detected of other electronic equipments is as picture frame to be detected.
Step 302, face critical point detection is carried out to video frame, face is oriented based on the face key point detected and is closed
Key range will extend to the outside the region of preset range as the human face region oriented to face key area in video frame.
Above-mentioned executing subject can orient human face region from video frame.Specifically, it can detecte the face in video
Key point.Herein, face key point can be the point of characterization face key position feature, position, shape including characterizing face
The point of the features such as shape, or can also include the point of the features of faces determinant attribute such as characterization cheekbone, chin, dimple.
Optionally, above-mentioned face key point may include on the profile of the key positions such as eyes, nose, mouth, eyebrow
Point, such as canthus, the corners of the mouth, eyebrow peak, etc..
After detecting face key point, the minimum external contact zone for being included all face key points is determined
For face key area.In practice, can outer boundary based on eyebrow or the tail of the eye and mouth lower end creation it is minimum outer
Rectangle is connect as the face key area oriented.
Usual method for detecting human face is less frequently utilized the marginal information of face, therefore the face key area oriented may not
Include face edge.In the present embodiment, the region after face key area can be extended to the outside to preset range is as positioning
Face edge is included in the human face region oriented by human face region out.Wherein, preset range can be based on positioning
What the size of face key area out determined, for example, the boundary of face key area can be expanded to presupposition multiple work outward
For human face region, it is ensured that the faces such as chin, cheek edge is comprised in the region after expanding.
The transfer and replacement of the current faces key position such as technical concerns eyes, eyebrow, mouth, nose of changing face, in people
Effect of changing face in face key area is more true to nature, but the fringe regions such as cheek, chin are difficult to and are accurately replaced.
It is extended to the outside by the face key area that will test in the present embodiment to be included in and orient by the fringe region of face
In human face region, can efficiently use the feature of face fringe region in subsequent detection come the face object identified in image is
It is no for forgery, to promote the accuracy and reliability of testing result.
Step 303, what use had been trained changes face detection model to face object included in the human face region oriented
It whether is that the forgery face synthesized is detected.
Human face region image can be syncopated as from picture frame from the human face region oriented based on step 302, by cutting
The detection model of changing face that human face region image input out has been trained is detected.Wherein, detection model of changing face, which can be, to be based on
What the set of the facial image with true and false label trained, detection model of changing face can detecte out the human face region image inputted
In face object whether be synthesis forgery face, or can determine input human face region image in face object
It is the probability value of the forgery face of synthesis.
The step 301 and step 303 of the present embodiment are corresponding with the step 201 of previous embodiment, step 203 respectively, step
301, the specific implementation of step 303 can be respectively with reference to the description in previous embodiment to step 201, step 203, herein
It repeats no more.
In some optional implementations of the present embodiment, above-mentioned detection model of changing face be can be through the first training behaviour
What work obtained.As shown in figure 3, the first training operation may include:
Firstly, in step 311, obtaining first sample image collection.
First sample image collection is the sample data set for training detection model of changing face.In first sample image collection
First sample image may include real human face image and synthesis facial image.Synthesizing facial image is by the face in image
Object replaces with the image generated after another face object, that is, the image of the forgery face generated after changing face.
In the present embodiment, the image of face can be acquired as the real human face image in first sample image collection,
Or the video of real human face is obtained, therefrom extract real human face image.It is raw after acquisition being concentrated to change face from data with existing
At forgery face image as the synthesis facial image in first sample image collection.Can also to real human face image into
Capable processing of changing face, replaces with other face objects for the face object in real human face image to generate synthesis facial image.
Herein, the first sample image in first sample image collection has label, which can be used as the first sample
Face object in this image is the synthesis face object of true face object (without the face object changed face) or forgery
The markup information of (the face object after changing face).
Then, in step 312, to first sample image into face critical point detection is carried out, based on the face detected
Key point orients face key area, will extend to the outside the region of preset range in first sample image to face key area
As the human face region oriented from first sample image.
Method identical with step 302 can be used, orients human face region from first sample image.It specifically can be right
First sample image carries out face critical point detection, detects the position of the key points such as eyebrow, eyes, nose, mouth, Huo Zheke
To extract the position feature of these face key points.The position feature for being then based on face key point determines face key area
Domain.It is then possible to which the region after face key area is expanded presupposition multiple is as the face for the first sample image oriented
Region.
Later, in step 313, based on the human face region oriented from first sample image to the image constructed point
Neural network is trained, and is included to the human face region oriented from first sample image in image classification neural network
Face object whether be deconditioning when the prediction error of forgery face of synthesis reaches preset first condition of convergence, will instruct
Practice the detection model of changing face that the image classification neural network completed is determined as having trained.
Image Classification Neural can be constructed based on neural network model.Image classification neural network can be one two
Sorter network is that real human face or forgery face are classified for the face object in the image to input.Then using
The face object in human face region that the image classification network of building orients step 312 is classified, and first sample is obtained
The corresponding face object classification result of image.Then by comparing image Classification Neural to the classification knot of first sample image
The label of fruit and first sample image determines image classification neural network to the face area oriented from first sample image
The face object that domain is included whether be synthesis forgery face prediction error.If the prediction error is not up to preset first
The condition of convergence, then using the parameter of gradient descent method adjustment image classification neural network, to update image classification neural network.
Included to the human face region oriented from first sample image again using updated image classification neural network later
Face object whether be that the forgery face of synthesis is predicted, compare between prediction result and the label of first sample image
Whether difference meets preset first condition of convergence.If not satisfied, then repeat the parameter of above-mentioned adjustment image classification neural network,
It updates image classification neural network, oriented using updated image classification neural network prediction to from first sample image
The human face region face object that is included whether be that the forgery face of synthesis is predicted, compare prediction result and first sample
Whether the difference between the label of image meets the operation of preset first condition of convergence.It is pre- until image classification neural network
When the difference surveyed between result and the label of first sample image meets preset first condition of convergence, stop updating image classification
The parameter of neural network completes training.Image classification neural network the changing face of being determined as having trained that training is completed is detected into mould
Type.
Above-mentioned preset first condition of convergence can be difference value less than preset threshold value, be also possible to parameter update times
More than preset frequency threshold value.
Change face the synthesis face after operation edge include replacement before face characteristic, by real human face image and
It forges facial image to carry out face critical point detection and determine the human face region comprising face edge, so that image classification is neural
Network obtains the feature resolution real human face using face edge in training and forges the ability of face, guarantees what training obtained
It changes face the reliability of detection model, to promote the accuracy of Face datection result.
Optionally, it in order to which lift scheme is to the practicability of the facial image of different scale or different direction, changes face in training
It, can be with spread training sample, so that including the facial image of various scales and orientation in training sample, to increase when detection model
The ability of strong detection model of changing face.Above-mentioned first training operation can also include: by the true people in first sample image collection
Respectively as new real human face image and new synthesis facial image after face image and synthesis facial image progress geometric transformation
It is added in first sample image collection.It can be by being revolved to existing facial image in first sample image collection
Turn, the geometric transformations such as scaling, shear transformation at random, obtains new facial image.Due to during geometric transformation not to people
The key point of face is replaced or moves, so the facial image obtained after geometric transformation and the facial image before corresponding transformation
True and false label having the same.Namely the image that real human face image generates after geometric transformation is another width real human face figure
Picture, the image that synthesis facial image generates after geometric transformation are another width synthesis facial images.
In this way, by the geometric transformation of image in training sample come the diversity of training for promotion sample, so that training obtains
Change face detection model have for various forms facial image detectability, thus further improve facial image inspection
Survey the reliability of result.
With continued reference to Fig. 4, it illustrates the streams according to another embodiment of the detection method of the facial image of the application
Cheng Tu.As shown in figure 4, the process 400 of the detection method of the facial image of the present embodiment, comprising the following steps:
Step 401, the picture frame comprising face object is obtained.
In the present embodiment, the available image comprising face object of the executing subject of the detection method of facial image
Frame, as picture frame to be detected.The picture frame comprising face object can be specifically extracted from database, or can receive
The picture frame comprising face object in the video to be detected of other electronic equipments is as picture frame to be detected.
Step 402, Face datection is carried out to picture frame, orients human face region.
In the present embodiment, candidate face detection block can be determined first, and the image extracted in candidate face detection block is special
Sign is judged using classifier with the presence or absence of face object in candidate face detection block, if so, can be by candidate face detection block
Certain size is reduced, feature is extracted in the candidate face detection block of minification and judges whether there is face object, is recycled
Candidate face detection block of the above process until determining size minimum and comprising face object is executed, as the face detected
Boundary.Candidate face detection block can be slided and repeated the above process whether there is other faces pair in detection image frame
As.Alternatively, if classifier determines in candidate face detection block, there is no face objects, can slide candidate face detection
It is repeated the above process after frame or amplification candidate face detection block.Thus, it is possible to which the human face region in picture frame is oriented to come.
The step 401, step 402 of the present embodiment are consistent with the step 201 of previous embodiment, step 202 respectively, step
401, the specific implementation of step 402 can also refer to the description in previous embodiment to step 201, step 202 respectively, this
Place repeats no more.
Optionally, human face region can also be oriented by the way of step 302 in the present embodiment: video frame is carried out
Face critical point detection orients face key area based on the face key point detected, will be crucial to face in video frame
Region extends to the outside the region of preset range as the human face region oriented.
Optionally, after orienting human face region, human face region can also be aligned, can be specifically based on face
Key point is normalized the size of human face region using affine transformation by human face posture and position correction, to eliminate as far as possible
Posture, the size difference bring error of human face region of different faces.
Step 403, the human face recognition model that use has been trained extracts the feature for the human face region oriented.
It can use preparatory trained human face recognition model to propose the human face region progress feature that step 402 is oriented
It takes.Herein, human face recognition model can be based on the model that obtains of machine learning method training, can be with when identifying face
Feature extraction is carried out to face, and is compared to obtain face recognition result with feature templates library based on the feature extracted.This
In embodiment, existing human face recognition model can use to extract the feature of human face region.The human face recognition model can be from
Other equipment obtain, and run in above-mentioned executing subject.
Since human face recognition model has good ability in feature extraction, the differentiation feature of different people can be extracted,
And synthesizing face in the scene of the present embodiment is to replace another face and original face, it is also desirable to utilize the difference of different people
Alienation feature, so human face recognition model is applied to also to can achieve preferable effect in the classification of real human face and forgery face
Fruit.Also, the present embodiment can use mature human face recognition model, eliminate in detection model training of changing face to feature extraction
The study of ability simplifies the framework for detection model of changing face.
Step 404, the feature of the human face region extracted based on human face recognition model, using detection model of changing face to positioning
Whether face object included in human face region out is that the forgery face synthesized is classified.
In the present embodiment, detection model of changing face may include disaggregated model.The disaggregated model can be based on SVM
The building such as (Support Vector Machine, support vector machines), neural network.Human face recognition model can be extracted
Feature inputs disaggregated model, obtains classification prediction result.
In some optional implementations of the present embodiment, above-mentioned detection model of changing face be can be through the second training behaviour
Make what training obtained.Second training operation may include step 411, step 412, step 413 and step 414.
Firstly, obtaining the second sample image set in step 411.
Second sample image set is the sample data set for training detection model of changing face.In second sample image set
The second sample image include real human face image and synthesis facial image, synthesis facial image is by the face object in image
Replace with the image generated after another face object.
The acquisition methods of second sample image set can be with reference to above to the description of step 311, and details are not described herein again.And
And optionally, first sample image collection and the second sample image set can have intersection or first sample image collection with
Second sample image collection is combined into the same set.
In step 412, Face datection is carried out to the second sample image, orients the human face region of the second sample image.
Face datection can be carried out to the second sample image using the method similar with step 402, orient the second sample
Human face region in image.
Optionally, the human face region in the second sample image can also be aligned, utilizes the affine transformation school of image
Positive face, to reduce error.
In step 413, the human face recognition model that use has been trained carries out feature to the human face region of the second sample image
It extracts.
Human face recognition model can be the model of preparatory trained face for identification, can be stored in advance in above-mentioned hold
In row main body, or it can be obtained by above-mentioned executing subject from other electronic equipments.The human face recognition model trained has good
Good face characteristic extractability, can extract the differentiation feature of different faces.Can use with it is consistent in step 403
Human face recognition model carries out feature extraction to the human face region of the second sample image, obtains the face object in the second sample image
Feature.Or feature extraction can also be carried out using human face region of another human face recognition model to the second sample image.
In step 414, the disaggregated model constructed is trained based on the feature extracted, in disaggregated model to
Whether face object included in the human face region of two sample images is that the prediction error of forgery face of synthesis reaches default
Second condition of convergence when deconditioning, the disaggregated model completed based on training generates the detection model of changing face trained.
Herein, step 413 can be extracted based on structural classifications models such as SVM, neural networks, and using disaggregated model
Feature out is classified, and obtains whether face object included in the human face region of the second sample image is the forgery synthesized
The prediction result of face.It then, whether is the forgery face synthesized according to the prediction result of disaggregated model and the second sample image
Markup information (markup information of real human face be "No", synthesize face markup information be "Yes") between difference determine point
The prediction error of class model.When the prediction error for determining disaggregated model is not up to preset second condition of convergence, can be based on
The parameter of gradient descent algorithm iteration adjustment disaggregated model again extracts step 413 based on the disaggregated model after adjusting parameter
Feature out carries out classification prediction.When the prediction error of disaggregated model reaches preset second condition of convergence, can stop adjusting
The parameter of whole disaggregated model completes training.The detection model of changing face that the disaggregated model for completing training is determined as having trained.It is above-mentioned
The number that preset second condition of convergence can be iteration adjustment parameter reaches preset frequency threshold value, or prediction error reaches
Preset convergency value.
As can be seen that not needing to repeat to extract the face area in the second sample image during train classification models
The feature in domain, it is only necessary to the parameter of iteration update disaggregated model.Thus, it is possible to promote the training speed of disaggregated model, subtract
Few resource consumption, while guaranteeing the accuracy of disaggregated model.
With further reference to Fig. 5, as the realization to method shown in above-mentioned each figure, this application provides a kind of facial images
One embodiment of detection device, the Installation practice is corresponding with Fig. 2, Fig. 3 and embodiment of the method shown in Fig. 4, the device
It specifically can be applied in various electronic equipments.
As shown in figure 5, the detection device 500 of the facial image of the present embodiment includes: acquiring unit 501, positioning unit 502
With detection unit 503.Wherein, acquiring unit 501 is configured as obtaining the picture frame comprising face object；502 quilt of positioning unit
It is configured to carry out Face datection to picture frame, orients human face region；Detection unit 503 is configured as changing face using what is trained
Whether detection model is that the forgery face synthesized detects to face object included in the human face region oriented.
In some embodiments, above-mentioned positioning unit 502 can be configured to as follows to picture frame
Face datection is carried out, human face region is oriented: face critical point detection being carried out to video frame, based on the face key point detected
Face key area is oriented, using the region that extends to the outside preset range in video frame to face key area as orienting
Human face region.
In some embodiments, the above-mentioned detection model of changing face trained is obtained by executing following first training operation
: first sample image collection is obtained, the first sample image in first sample image collection includes real human face image and conjunction
At facial image, synthesizing facial image is that the face object in image is replaced with the image generated after another face object；
To first sample image into face critical point detection is carried out, face key area is oriented based on the face key point detected,
The region for extending to the outside preset range in first sample image to face key area is positioned as from first sample image
Human face region out；Based on the human face region oriented from first sample image to the image classification neural network constructed into
Row is trained, and is in the face object that image classification neural network is included to the human face region oriented from first sample image
Deconditioning when the prediction error of the no forgery face for synthesis reaches preset first condition of convergence, the image that training is completed
What Classification Neural was determined as having trained change face detection model.
In some embodiments, above-mentioned first training operation further include: by the real human face in first sample image collection
Image and synthesis facial image add after carrying out geometric transformation respectively as new real human face image and new synthesis facial image
It adds in first sample image collection.
In some embodiments, above-mentioned detection model of changing face includes disaggregated model；Above-mentioned apparatus 500 can also include: to mention
Unit is taken, is configured as extracting the feature for the human face region oriented using the human face recognition model trained；And above-mentioned inspection
Survey unit 503 can be configured to
The no forgery face for synthesis is detected: the feature based on the human face region that human face recognition model extracts, using inspection of changing face
Whether survey model is that the forgery face synthesized is classified to face object included in the human face region oriented.
In some embodiments, above-mentioned detection model of changing face is obtained by following second training operation: obtaining second
Sample image set, the second sample image in the second sample image set include real human face image and synthesis facial image,
Synthesis facial image is that the face object in image is replaced with the image generated after another face object；To the second sample graph
As carrying out Face datection, the human face region of the second sample image is oriented；Using the human face recognition model trained to the second sample
The human face region of this image carries out feature extraction；The disaggregated model constructed is trained based on the feature extracted, is being divided
Whether face object included in human face region of the class model to the second sample image is that the prediction for forging face synthesized misses
Deconditioning when difference reaches preset second condition of convergence, the disaggregated model completed based on training generate the detection of changing face trained
Model.
The detection device 500 of the facial image of the above embodiments of the present application, by obtaining the picture frame comprising face object,
Face datection is carried out to picture frame, orients human face region, what use had been trained changes face detection model to the face area oriented
Whether face object included in domain is that the forgery face synthesized is detected, and realizes the accurate detection of " changing face ", can
Effectively detect the forgery image generated based on the face in nerual network technique replacement image.
Below with reference to Fig. 6, it illustrates the electronic equipment that is suitable for being used to realize embodiment of the disclosure, (example is as shown in figure 1
Server) 600 structural schematic diagram.Electronic equipment shown in Fig. 6 is only an example, should not be to embodiment of the disclosure
Function and use scope bring any restrictions.
As shown in fig. 6, electronic equipment 600 may include processing unit (such as central processing unit, graphics processor etc.)
601, random access can be loaded into according to the program being stored in read-only memory (ROM) 602 or from storage device 608
Program in memory (RAM) 603 and execute various movements appropriate and processing.In RAM 603, it is also stored with electronic equipment
Various programs and data needed for 600 operations.Processing unit 601, ROM 602 and RAM 603 pass through the phase each other of bus 604
Even.Input/output (I/O) interface 605 is also connected to bus 604.
In general, following device can connect to I/O interface 605: including such as touch screen, touch tablet, keyboard, mouse, taking the photograph
As the input unit 606 of head, microphone, accelerometer, gyroscope etc.；Including such as liquid crystal display (LCD), loudspeaker, vibration
The output device 607 of dynamic device etc.；Storage device 608 including such as hard disk etc.；And communication device 609.Communication device 609 can
To allow electronic equipment 600 wirelessly or non-wirelessly to be communicated with other equipment to exchange data.Although Fig. 6 is shown with various
The electronic equipment 600 of device, it should be understood that being not required for implementing or having all devices shown.It can be alternatively
Implement or have more or fewer devices.Each box shown in Fig. 6 can represent a device, also can according to need
Represent multiple devices.
Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description
Software program.For example, embodiment of the disclosure includes a kind of computer program product comprising be carried on computer-readable medium
On computer program, which includes the program code for method shown in execution flow chart.In such reality
It applies in example, which can be downloaded and installed from network by communication device 609, or from storage device 608
It is mounted, or is mounted from ROM 602.When the computer program is executed by processing unit 601, the implementation of the disclosure is executed
The above-mentioned function of being limited in the method for example.It should be noted that computer-readable medium described in embodiment of the disclosure can
To be computer-readable signal media or computer readable storage medium either the two any combination.Computer can
Reading storage medium for example may be-but not limited to-the system of electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor, device
Or device, or any above combination.The more specific example of computer readable storage medium can include but is not limited to: tool
There are electrical connection, the portable computer diskette, hard disk, random access storage device (RAM), read-only memory of one or more conducting wires
(ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-
ROM), light storage device, magnetic memory device or above-mentioned any appropriate combination.In embodiment of the disclosure, computer
Readable storage medium storing program for executing can be any tangible medium for including or store program, which can be commanded execution system, device
Either device use or in connection.And in embodiment of the disclosure, computer-readable signal media may include
In a base band or as the data-signal that carrier wave a part is propagated, wherein carrying computer-readable program code.It is this
The data-signal of propagation can take various forms, including but not limited to electromagnetic signal, optical signal or above-mentioned any appropriate
Combination.Computer-readable signal media can also be any computer-readable medium other than computer readable storage medium, should
Computer-readable signal media can send, propagate or transmit for by instruction execution system, device or device use or
Person's program in connection.The program code for including on computer-readable medium can transmit with any suitable medium,
Including but not limited to: electric wire, optical cable, RF (radio frequency) etc. or above-mentioned any appropriate combination.
Above-mentioned computer-readable medium can be included in above-mentioned electronic equipment；It is also possible to individualism, and not
It is fitted into the electronic equipment.Above-mentioned computer-readable medium carries one or more program, when said one or more
When a program is executed by the electronic equipment, so that the electronic equipment: obtaining the picture frame comprising face object；Picture frame is carried out
Face datection orients human face region；Using the detection model of changing face trained to included in the human face region oriented
Whether face object is that the forgery face synthesized is detected.
The behaviour for executing embodiment of the disclosure can be write with one or more programming languages or combinations thereof
The computer program code of work, programming language include object oriented program language-such as Java,
Smalltalk, C++ further include conventional procedural programming language-such as " C " language or similar program design language
Speech.Program code can be executed fully on the user computer, partly be executed on the user computer, as an independence
Software package execute, part on the user computer part execute on the remote computer or completely in remote computer or
It is executed on server.In situations involving remote computers, remote computer can pass through the network of any kind --- packet
Include local area network (LAN) or wide area network (WAN) --- it is connected to subscriber computer, or, it may be connected to outer computer (such as
It is connected using ISP by internet).
Flow chart and block diagram in attached drawing are illustrated according to the system of the various embodiments of the application, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part of one module, program segment or code of table, a part of the module, program segment or code include one or more use
The executable instruction of the logic function as defined in realizing.It should also be noted that in some implementations as replacements, being marked in box
The function of note can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are actually
It can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it to infuse
Meaning, the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart can be with holding
The dedicated hardware based system of functions or operations as defined in row is realized, or can use specialized hardware and computer instruction
Combination realize.
Being described in unit involved in the embodiment of the present application can be realized by way of software, can also be by hard
The mode of part is realized.Described unit also can be set in the processor, for example, can be described as: a kind of processor packet
Include acquiring unit, positioning unit and detection unit.Wherein, the title of these units is not constituted under certain conditions to the unit
The restriction of itself, for example, acquiring unit is also described as " obtaining the unit of the picture frame comprising face object ".
Above description is only the preferred embodiment of the application and the explanation to institute's application technology principle.Those skilled in the art
Member is it should be appreciated that invention scope involved in the application, however it is not limited to technology made of the specific combination of above-mentioned technical characteristic
Scheme, while should also cover in the case where not departing from foregoing invention design, it is carried out by above-mentioned technical characteristic or its equivalent feature
Any combination and the other technical solutions formed.Such as features described above has similar function with (but being not limited to) disclosed herein
Can technical characteristic replaced mutually and the technical solution that is formed.
Claims (14)
1. a kind of detection method of facial image, comprising:
Obtain the picture frame comprising face object；
Face datection is carried out to described image frame, orients human face region；
Use the detection model of changing face trained to face object included in the human face region oriented whether for synthesis
Face is forged to be detected.
2. it is described that Face datection is carried out to described image frame according to the method described in claim 1, wherein, orient face area
Domain, comprising:
Face critical point detection is carried out to the video frame, face key area is oriented based on the face key point detected,
The region of preset range is extended to the outside as the human face region oriented to the face key area using in the video frame.
3. according to the method described in claim 2, wherein, the detection model of changing face trained is by executing following first
Training operation obtains:
First sample image collection is obtained, the first sample image in the first sample image collection includes real human face image
With synthesis facial image, the synthesis facial image is to generate after the face object in image is replaced with another face object
Image；
To the first sample image into face critical point detection is carried out, face is oriented based on the face key point detected and is closed
Key range will extend to the outside the region of preset range as from described to the face key area in the first sample image
The human face region oriented in first sample image；
The image classification neural network constructed is instructed based on the human face region oriented from the first sample image
Practice, in the face pair that described image Classification Neural is included to the human face region oriented from the first sample image
As if the no prediction error of forgerys face for synthesis deconditioning when reaching preset first condition of convergence, it will training completion
Image classification neural network is determined as the detection model of changing face trained.
4. according to the method described in claim 3, wherein, first training operates further include:
Make respectively by the real human face image in the first sample image collection and after synthesizing facial image progress geometric transformation
It is added in the first sample image collection for new real human face image and new synthesis facial image.
5. according to the method described in claim 1, wherein, the detection model of changing face includes disaggregated model；
The method also includes:
The feature for the human face region oriented is extracted using the human face recognition model trained；And
It is described to use the detection model of changing face trained to face object included in the human face region oriented whether for conjunction
At forgery face detected, comprising:
Feature based on the human face region that the human face recognition model extracts, using the detection model of changing face to orienting
Whether face object included in human face region is that the forgery face synthesized is classified.
6. according to the method described in claim 5, wherein, the detection model of changing face is obtained by following second training operation
:
The second sample image set is obtained, the second sample image in the second sample image set includes real human face image
With synthesis facial image, the synthesis facial image is to generate after the face object in image is replaced with another face object
Image；
Face datection is carried out to second sample image, orients the human face region of second sample image；
Feature extraction is carried out using human face region of the human face recognition model trained to second sample image；
The disaggregated model constructed is trained based on the feature extracted, in the disaggregated model to second sample graph
Face object included in the human face region of picture whether be synthesis forgerys face prediction error reach it is preset second receipts
Deconditioning when holding back condition, the disaggregated model completed based on training generate the detection model of changing face trained.
7. a kind of detection device of facial image, comprising:
Acquiring unit is configured as obtaining the picture frame comprising face object；
Positioning unit is configured as carrying out Face datection to described image frame, orients human face region；
Detection unit is configured as using the detection model of changing face trained to face included in the human face region oriented
Whether object is that the forgery face synthesized is detected.
8. device according to claim 7, wherein the positioning unit is configured to as follows to institute
It states picture frame and carries out Face datection, orient human face region:
Face critical point detection is carried out to the video frame, face key area is oriented based on the face key point detected,
The region of preset range is extended to the outside as the human face region oriented to the face key area using in the video frame.
9. device according to claim 8, wherein the detection model of changing face trained is by executing following first
Training operation obtains:
First sample image collection is obtained, the first sample image in the first sample image collection includes real human face image
With synthesis facial image, the synthesis facial image is to generate after the face object in image is replaced with another face object
Image；
To the first sample image into face critical point detection is carried out, face is oriented based on the face key point detected and is closed
Key range will extend to the outside the region of preset range as from described to the face key area in the first sample image
The human face region oriented in first sample image；
The image classification neural network constructed is instructed based on the human face region oriented from the first sample image
Practice, in the face pair that described image Classification Neural is included to the human face region oriented from the first sample image
As if the no prediction error of forgerys face for synthesis deconditioning when reaching preset first condition of convergence, it will training completion
Image classification neural network is determined as the detection model of changing face trained.
10. device according to claim 9, wherein the first training operation further include:
Make respectively by the real human face image in the first sample image collection and after synthesizing facial image progress geometric transformation
It is added in the first sample image collection for new real human face image and new synthesis facial image.
11. device according to claim 7, wherein the detection model of changing face includes disaggregated model；
Described device further include:
Extraction unit is configured as extracting the feature for the human face region oriented using the human face recognition model trained；And
The detection unit is configured to as follows to face pair included in the human face region oriented
As if the no forgery face for synthesis is detected:
Feature based on the human face region that the human face recognition model extracts, using the detection model of changing face to orienting
Whether face object included in human face region is that the forgery face synthesized is classified.
12. device according to claim 11, wherein the detection model of changing face is operated by following second training
Out:
The second sample image set is obtained, the second sample image in the second sample image set includes real human face image
With synthesis facial image, the synthesis facial image is to generate after the face object in image is replaced with another face object
Image；
Face datection is carried out to second sample image, orients the human face region of second sample image；
Feature extraction is carried out using human face region of the human face recognition model trained to second sample image；
The disaggregated model constructed is trained based on the feature extracted, in the disaggregated model to second sample graph
Face object included in the human face region of picture whether be synthesis forgerys face prediction error reach it is preset second receipts
Deconditioning when holding back condition, the disaggregated model completed based on training generate the detection model of changing face trained.
13. a kind of electronic equipment, comprising:
One or more processors；
Storage device, for storing one or more programs,
When one or more of programs are executed by one or more of processors, so that one or more of processors are real
Now such as method as claimed in any one of claims 1 to 6.
14. a kind of computer-readable medium, is stored thereon with computer program, wherein real when described program is executed by processor
Now such as method as claimed in any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910472084.1A CN110210393A (en) | 2019-05-31 | 2019-05-31 | The detection method and device of facial image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910472084.1A CN110210393A (en) | 2019-05-31 | 2019-05-31 | The detection method and device of facial image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110210393A true CN110210393A (en) | 2019-09-06 |
Family
ID=67790092
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910472084.1A Pending CN110210393A (en) | 2019-05-31 | 2019-05-31 | The detection method and device of facial image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110210393A (en) |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110533775A (en) * | 2019-09-18 | 2019-12-03 | 广州智美科技有限公司 | A kind of glasses matching process, device and terminal based on 3D face |
| CN110909750A (en) * | 2019-11-14 | 2020-03-24 | 展讯通信（上海）有限公司 | Image difference detection method and device, storage medium and terminal |
| CN110929617A (en) * | 2019-11-14 | 2020-03-27 | 北京神州绿盟信息安全科技股份有限公司 | Face-changing composite video detection method and device, electronic equipment and storage medium |
| CN110969572A (en) * | 2019-11-29 | 2020-04-07 | 广州华多网络科技有限公司 | Face changing model training method, face exchanging device and electronic equipment |
| CN111191571A (en) * | 2019-12-26 | 2020-05-22 | 新绎健康科技有限公司 | Traditional Chinese medicine facial diagnosis face partitioning method and system based on face feature point detection |
| CN111209855A (en) * | 2020-01-06 | 2020-05-29 | 电子科技大学 | Face image identification method based on two-channel dense convolution neural network with contour enhancement |
| CN111291863A (en) * | 2020-01-20 | 2020-06-16 | 腾讯科技（深圳）有限公司 | Training method of face changing identification model, face changing identification method, device and equipment |
| CN111339832A (en) * | 2020-02-03 | 2020-06-26 | 中国人民解放军国防科技大学 | Method and device for detecting face synthetic image |
| CN111353392A (en) * | 2020-02-18 | 2020-06-30 | 腾讯科技（深圳）有限公司 | Face change detection method, device, equipment and storage medium |
| CN111444881A (en) * | 2020-04-13 | 2020-07-24 | 中国人民解放军国防科技大学 | Fake face video detection method and device |
| CN111611873A (en) * | 2020-04-28 | 2020-09-01 | 平安科技（深圳）有限公司 | Face replacement detection method and device, electronic equipment and computer storage medium |
| CN111652878A (en) * | 2020-06-16 | 2020-09-11 | 腾讯科技（深圳）有限公司 | Image detection method, image detection device, computer equipment and storage medium |
| CN111739046A (en) * | 2020-06-19 | 2020-10-02 | 百度在线网络技术（北京）有限公司 | Method, apparatus, device and medium for model update and image detection |
| CN111783644A (en) * | 2020-06-30 | 2020-10-16 | 百度在线网络技术（北京）有限公司 | Detection method, device, equipment and computer storage medium |
| CN111814697A (en) * | 2020-07-13 | 2020-10-23 | 伊沃人工智能技术（江苏）有限公司 | Real-time face recognition method and system and electronic equipment |
| CN111862030A (en) * | 2020-07-15 | 2020-10-30 | 北京百度网讯科技有限公司 | Face synthetic image detection method and device, electronic equipment and storage medium |
| CN112036317A (en) * | 2020-08-31 | 2020-12-04 | 成都新潮传媒集团有限公司 | Face image intercepting method and device and computer equipment |
| CN112052831A (en) * | 2020-09-25 | 2020-12-08 | 北京百度网讯科技有限公司 | Face detection method, device and computer storage medium |
| CN112101281A (en) * | 2020-09-25 | 2020-12-18 | 北京百度网讯科技有限公司 | Face image detection method and device |
| CN112116525A (en) * | 2020-09-24 | 2020-12-22 | 百度在线网络技术（北京）有限公司 | Face-changing identification method, device, equipment and computer-readable storage medium |
| CN112132133A (en) * | 2020-06-16 | 2020-12-25 | 杭州中科睿鉴科技有限公司 | Identification image data enhancement method and authenticity intelligent identification method |
| CN112560753A (en) * | 2020-12-23 | 2021-03-26 | 平安银行股份有限公司 | Face recognition method, device and equipment based on feature fusion and storage medium |
| CN112633259A (en) * | 2021-03-08 | 2021-04-09 | 北京远鉴信息技术有限公司 | Face abnormity authentication identification method and device and electronic equipment |
| CN112767303A (en) * | 2020-08-12 | 2021-05-07 | 腾讯科技（深圳）有限公司 | Image detection method, device, equipment and computer readable storage medium |
| CN112884761A (en) * | 2021-03-19 | 2021-06-01 | 东营市阔海水产科技有限公司 | Aquatic economic animal head identification method, terminal device and readable storage medium |
| CN113014914A (en) * | 2021-01-27 | 2021-06-22 | 北京市博汇科技股份有限公司 | Neural network-based single face-changing short video identification method and system |
| CN113221786A (en) * | 2021-05-21 | 2021-08-06 | 深圳市商汤科技有限公司 | Data classification method and device, electronic equipment and storage medium |
| CN113449542A (en) * | 2020-03-24 | 2021-09-28 | 浙江宇视科技有限公司 | Face changing identification method, device, equipment and medium |
| CN113554045A (en) * | 2020-04-23 | 2021-10-26 | 国家广播电视总局广播电视科学研究院 | Data set manufacturing method, device, equipment and storage medium |
| CN114170689A (en) * | 2021-09-06 | 2022-03-11 | 北京天德科技有限公司 | Method and system for detecting forged face image in block chain prediction machine |
Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101593425A (en) * | 2009-05-06 | 2009-12-02 | 深圳市汉华安道科技有限责任公司 | A kind of fatigue driving monitoring method and system based on machine vision |
| CN101739555A (en) * | 2009-12-01 | 2010-06-16 | 北京中星微电子有限公司 | Method and system for detecting false face, and method and system for training false face model |
| CN102348116A (en) * | 2010-08-03 | 2012-02-08 | 株式会社理光 | Video processing method, video processing device and video processing system |
| CN103679118A (en) * | 2012-09-07 | 2014-03-26 | 汉王科技股份有限公司 | Human face in-vivo detection method and system |
| CN105404846A (en) * | 2014-09-15 | 2016-03-16 | 中国移动通信集团广东有限公司 | Image processing method and apparatus |
| CN105426867A (en) * | 2015-12-11 | 2016-03-23 | 小米科技有限责任公司 | Face identification verification method and apparatus |
| CN106778525A (en) * | 2016-11-25 | 2017-05-31 | 北京旷视科技有限公司 | Identity identifying method and device |
| CN106778518A (en) * | 2016-11-24 | 2017-05-31 | 汉王科技股份有限公司 | A kind of human face in-vivo detection method and device |
| CN107220635A (en) * | 2017-06-21 | 2017-09-29 | 北京市威富安防科技有限公司 | Human face in-vivo detection method based on many fraud modes |
| CN107292267A (en) * | 2017-06-21 | 2017-10-24 | 北京市威富安防科技有限公司 | Photo fraud convolutional neural networks training method and human face in-vivo detection method |
| CN107590482A (en) * | 2017-09-29 | 2018-01-16 | 百度在线网络技术（北京）有限公司 | information generating method and device |
| CN108509915A (en) * | 2018-04-03 | 2018-09-07 | 百度在线网络技术（北京）有限公司 | The generation method and device of human face recognition model |
| CN108921071A (en) * | 2018-06-24 | 2018-11-30 | 深圳市中悦科技有限公司 | Human face in-vivo detection method, device, storage medium and processor |
| CN108985200A (en) * | 2018-07-02 | 2018-12-11 | 中国科学院半导体研究所 | A kind of In vivo detection algorithm of the non-formula based on terminal device |
| CN108985134A (en) * | 2017-06-01 | 2018-12-11 | 重庆中科云丛科技有限公司 | Face In vivo detection and brush face method of commerce and system based on binocular camera |
| CN109035167A (en) * | 2018-07-17 | 2018-12-18 | 北京新唐思创教育科技有限公司 | Method, apparatus, equipment and the medium that multiple faces in image are handled |
| CN109344709A (en) * | 2018-08-29 | 2019-02-15 | 中国科学院信息工程研究所 | A kind of face generates the detection method of forgery image |
| CN109376659A (en) * | 2018-10-26 | 2019-02-22 | 北京陌上花科技有限公司 | Training method, face critical point detection method, apparatus for face key spot net detection model |
| CN109670491A (en) * | 2019-02-25 | 2019-04-23 | 百度在线网络技术（北京）有限公司 | Identify method, apparatus, equipment and the storage medium of facial image |
| CN109711384A (en) * | 2019-01-09 | 2019-05-03 | 江苏星云网格信息技术有限公司 | A kind of face identification method based on depth convolutional neural networks |
| CN109815797A (en) * | 2018-12-17 | 2019-05-28 | 北京飞搜科技有限公司 | Biopsy method and device |
-
2019
- 2019-05-31 CN CN201910472084.1A patent/CN110210393A/en active Pending
Patent Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101593425A (en) * | 2009-05-06 | 2009-12-02 | 深圳市汉华安道科技有限责任公司 | A kind of fatigue driving monitoring method and system based on machine vision |
| CN101739555A (en) * | 2009-12-01 | 2010-06-16 | 北京中星微电子有限公司 | Method and system for detecting false face, and method and system for training false face model |
| CN102348116A (en) * | 2010-08-03 | 2012-02-08 | 株式会社理光 | Video processing method, video processing device and video processing system |
| CN103679118A (en) * | 2012-09-07 | 2014-03-26 | 汉王科技股份有限公司 | Human face in-vivo detection method and system |
| CN105404846A (en) * | 2014-09-15 | 2016-03-16 | 中国移动通信集团广东有限公司 | Image processing method and apparatus |
| CN105426867A (en) * | 2015-12-11 | 2016-03-23 | 小米科技有限责任公司 | Face identification verification method and apparatus |
| CN106778518A (en) * | 2016-11-24 | 2017-05-31 | 汉王科技股份有限公司 | A kind of human face in-vivo detection method and device |
| CN106778525A (en) * | 2016-11-25 | 2017-05-31 | 北京旷视科技有限公司 | Identity identifying method and device |
| CN108985134A (en) * | 2017-06-01 | 2018-12-11 | 重庆中科云丛科技有限公司 | Face In vivo detection and brush face method of commerce and system based on binocular camera |
| CN107220635A (en) * | 2017-06-21 | 2017-09-29 | 北京市威富安防科技有限公司 | Human face in-vivo detection method based on many fraud modes |
| CN107292267A (en) * | 2017-06-21 | 2017-10-24 | 北京市威富安防科技有限公司 | Photo fraud convolutional neural networks training method and human face in-vivo detection method |
| CN107590482A (en) * | 2017-09-29 | 2018-01-16 | 百度在线网络技术（北京）有限公司 | information generating method and device |
| CN108509915A (en) * | 2018-04-03 | 2018-09-07 | 百度在线网络技术（北京）有限公司 | The generation method and device of human face recognition model |
| CN108921071A (en) * | 2018-06-24 | 2018-11-30 | 深圳市中悦科技有限公司 | Human face in-vivo detection method, device, storage medium and processor |
| CN108985200A (en) * | 2018-07-02 | 2018-12-11 | 中国科学院半导体研究所 | A kind of In vivo detection algorithm of the non-formula based on terminal device |
| CN109035167A (en) * | 2018-07-17 | 2018-12-18 | 北京新唐思创教育科技有限公司 | Method, apparatus, equipment and the medium that multiple faces in image are handled |
| CN109344709A (en) * | 2018-08-29 | 2019-02-15 | 中国科学院信息工程研究所 | A kind of face generates the detection method of forgery image |
| CN109376659A (en) * | 2018-10-26 | 2019-02-22 | 北京陌上花科技有限公司 | Training method, face critical point detection method, apparatus for face key spot net detection model |
| CN109815797A (en) * | 2018-12-17 | 2019-05-28 | 北京飞搜科技有限公司 | Biopsy method and device |
| CN109711384A (en) * | 2019-01-09 | 2019-05-03 | 江苏星云网格信息技术有限公司 | A kind of face identification method based on depth convolutional neural networks |
| CN109670491A (en) * | 2019-02-25 | 2019-04-23 | 百度在线网络技术（北京）有限公司 | Identify method, apparatus, equipment and the storage medium of facial image |
Non-Patent Citations (8)
| Title |
|---|
| AMIN JOURABLOO等: "Face De-Spoofing: Anti-Spoofing via Noise Modeling", 《ARXIV:1807.09968V1 [CS.CV]》 * |
| ANDREAS ROSSLER等: "FaceForensics:A Large-scale Video Dataset for Forgery Detection in Human Faces", 《ARXIV:1803.09179V1 [CS.CV]》 * |
| JIANWEI YANG等: "Learn Convolutional Neural Network for Face Anti-Spoofing", 《ARXIV:1408.5601V2 [CS.CV]》 * |
| SHIFENG ZHANG等: "A Dataset and Benchmark for Large-scale Multi-modal Face Anti-spoofing", 《ARXIV:1812.00408V3 [CS.CV]》 * |
| 党永芳: "基于分层随机森林的面部特征点检测技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 崔滢: "基于统计学习的人脸特征点定位算法研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
| 张旭东: "基于Adaboost和主动性状模型相结合的驾驶员疲劳检测算法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 陈宇: "人脸检测与跟踪技术的研究", 《中国优秀博硕士学位论文全文数据库 (硕士) 信息科技辑》 * |
Cited By (46)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110533775A (en) * | 2019-09-18 | 2019-12-03 | 广州智美科技有限公司 | A kind of glasses matching process, device and terminal based on 3D face |
| CN110533775B (en) * | 2019-09-18 | 2023-04-18 | 广州智美科技有限公司 | Glasses matching method and device based on 3D face and terminal |
| CN110909750A (en) * | 2019-11-14 | 2020-03-24 | 展讯通信（上海）有限公司 | Image difference detection method and device, storage medium and terminal |
| CN110929617A (en) * | 2019-11-14 | 2020-03-27 | 北京神州绿盟信息安全科技股份有限公司 | Face-changing composite video detection method and device, electronic equipment and storage medium |
| CN110909750B (en) * | 2019-11-14 | 2022-08-19 | 展讯通信（上海）有限公司 | Image difference detection method and device, storage medium and terminal |
| CN110969572A (en) * | 2019-11-29 | 2020-04-07 | 广州华多网络科技有限公司 | Face changing model training method, face exchanging device and electronic equipment |
| CN110969572B (en) * | 2019-11-29 | 2023-04-18 | 广州方硅信息技术有限公司 | Face changing model training method, face exchange device and electronic equipment |
| CN111191571A (en) * | 2019-12-26 | 2020-05-22 | 新绎健康科技有限公司 | Traditional Chinese medicine facial diagnosis face partitioning method and system based on face feature point detection |
| CN111209855A (en) * | 2020-01-06 | 2020-05-29 | 电子科技大学 | Face image identification method based on two-channel dense convolution neural network with contour enhancement |
| CN111291863A (en) * | 2020-01-20 | 2020-06-16 | 腾讯科技（深圳）有限公司 | Training method of face changing identification model, face changing identification method, device and equipment |
| CN111291863B (en) * | 2020-01-20 | 2022-11-08 | 腾讯科技（深圳）有限公司 | Training method of face changing identification model, face changing identification method, device and equipment |
| CN111339832B (en) * | 2020-02-03 | 2023-09-12 | 中国人民解放军国防科技大学 | Face synthetic image detection method and device |
| CN111339832A (en) * | 2020-02-03 | 2020-06-26 | 中国人民解放军国防科技大学 | Method and device for detecting face synthetic image |
| CN111353392A (en) * | 2020-02-18 | 2020-06-30 | 腾讯科技（深圳）有限公司 | Face change detection method, device, equipment and storage medium |
| CN111353392B (en) * | 2020-02-18 | 2022-09-30 | 腾讯科技（深圳）有限公司 | Face change detection method, device, equipment and storage medium |
| CN113449542A (en) * | 2020-03-24 | 2021-09-28 | 浙江宇视科技有限公司 | Face changing identification method, device, equipment and medium |
| CN111444881A (en) * | 2020-04-13 | 2020-07-24 | 中国人民解放军国防科技大学 | Fake face video detection method and device |
| CN113554045B (en) * | 2020-04-23 | 2024-04-09 | 国家广播电视总局广播电视科学研究院 | Data set manufacturing method, device, equipment and storage medium |
| CN113554045A (en) * | 2020-04-23 | 2021-10-26 | 国家广播电视总局广播电视科学研究院 | Data set manufacturing method, device, equipment and storage medium |
| CN111611873A (en) * | 2020-04-28 | 2020-09-01 | 平安科技（深圳）有限公司 | Face replacement detection method and device, electronic equipment and computer storage medium |
| CN111652878A (en) * | 2020-06-16 | 2020-09-11 | 腾讯科技（深圳）有限公司 | Image detection method, image detection device, computer equipment and storage medium |
| CN112132133A (en) * | 2020-06-16 | 2020-12-25 | 杭州中科睿鉴科技有限公司 | Identification image data enhancement method and authenticity intelligent identification method |
| CN112132133B (en) * | 2020-06-16 | 2023-11-17 | 中国科学院计算技术研究所数字经济产业研究院 | Identification image data enhancement method and true-false intelligent identification method |
| CN111739046A (en) * | 2020-06-19 | 2020-10-02 | 百度在线网络技术（北京）有限公司 | Method, apparatus, device and medium for model update and image detection |
| CN111783644A (en) * | 2020-06-30 | 2020-10-16 | 百度在线网络技术（北京）有限公司 | Detection method, device, equipment and computer storage medium |
| CN111814697B (en) * | 2020-07-13 | 2024-02-13 | 伊沃人工智能技术（江苏）有限公司 | Real-time face recognition method and system and electronic equipment |
| CN111814697A (en) * | 2020-07-13 | 2020-10-23 | 伊沃人工智能技术（江苏）有限公司 | Real-time face recognition method and system and electronic equipment |
| CN111862030A (en) * | 2020-07-15 | 2020-10-30 | 北京百度网讯科技有限公司 | Face synthetic image detection method and device, electronic equipment and storage medium |
| CN111862030B (en) * | 2020-07-15 | 2024-02-09 | 北京百度网讯科技有限公司 | Face synthetic image detection method and device, electronic equipment and storage medium |
| US11881050B2 (en) | 2020-07-15 | 2024-01-23 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method for detecting face synthetic image, electronic device, and storage medium |
| CN112767303A (en) * | 2020-08-12 | 2021-05-07 | 腾讯科技（深圳）有限公司 | Image detection method, device, equipment and computer readable storage medium |
| CN112767303B (en) * | 2020-08-12 | 2023-11-28 | 腾讯科技（深圳）有限公司 | Image detection method, device, equipment and computer readable storage medium |
| CN112036317A (en) * | 2020-08-31 | 2020-12-04 | 成都新潮传媒集团有限公司 | Face image intercepting method and device and computer equipment |
| CN112116525A (en) * | 2020-09-24 | 2020-12-22 | 百度在线网络技术（北京）有限公司 | Face-changing identification method, device, equipment and computer-readable storage medium |
| CN112101281A (en) * | 2020-09-25 | 2020-12-18 | 北京百度网讯科技有限公司 | Face image detection method and device |
| CN112052831A (en) * | 2020-09-25 | 2020-12-08 | 北京百度网讯科技有限公司 | Face detection method, device and computer storage medium |
| CN112101281B (en) * | 2020-09-25 | 2023-06-16 | 北京百度网讯科技有限公司 | Face image detection method and device |
| CN112052831B (en) * | 2020-09-25 | 2023-08-08 | 北京百度网讯科技有限公司 | Method, device and computer storage medium for face detection |
| CN112560753A (en) * | 2020-12-23 | 2021-03-26 | 平安银行股份有限公司 | Face recognition method, device and equipment based on feature fusion and storage medium |
| CN113014914A (en) * | 2021-01-27 | 2021-06-22 | 北京市博汇科技股份有限公司 | Neural network-based single face-changing short video identification method and system |
| CN113014914B (en) * | 2021-01-27 | 2022-11-01 | 北京市博汇科技股份有限公司 | Neural network-based single face-changing short video identification method and system |
| CN112633259A (en) * | 2021-03-08 | 2021-04-09 | 北京远鉴信息技术有限公司 | Face abnormity authentication identification method and device and electronic equipment |
| CN112884761A (en) * | 2021-03-19 | 2021-06-01 | 东营市阔海水产科技有限公司 | Aquatic economic animal head identification method, terminal device and readable storage medium |
| WO2022242032A1 (en) * | 2021-05-21 | 2022-11-24 | 深圳市商汤科技有限公司 | Data classification method and apparatus, electronic device, storage medium. and computer program product |
| CN113221786A (en) * | 2021-05-21 | 2021-08-06 | 深圳市商汤科技有限公司 | Data classification method and device, electronic equipment and storage medium |
| CN114170689A (en) * | 2021-09-06 | 2022-03-11 | 北京天德科技有限公司 | Method and system for detecting forged face image in block chain prediction machine |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110210393A (en) | The detection method and device of facial image | |
| CN109816589B (en) | Method and apparatus for generating cartoon style conversion model | |
| CN108898186B (en) | Method and device for extracting image | |
| CN109214343B (en) | Method and device for generating face key point detection model | |
| US20230081645A1 (en) | Detecting forged facial images using frequency domain information and local correlation | |
| WO2021036059A1 (en) | Image conversion model training method, heterogeneous face recognition method, device and apparatus | |
| CN108197618B (en) | Method and device for generating human face detection model | |
| CN108171207A (en) | Face identification method and device based on video sequence | |
| CN108073910B (en) | Method and device for generating human face features | |
| CN110245612A (en) | The detection method and device of facial image | |
| CN107578017A (en) | Method and apparatus for generating image | |
| CN108985259A (en) | Human motion recognition method and device | |
| WO2020078119A1 (en) | Method, device and system for simulating user wearing clothing and accessories | |
| US20190147224A1 (en) | Neural network based face detection and landmark localization | |
| CN112967212A (en) | Virtual character synthesis method, device, equipment and storage medium | |
| CN108269254A (en) | Image quality measure method and apparatus | |
| CN107609506A (en) | Method and apparatus for generating image | |
| CN110741377A (en) | Face image processing method and device, storage medium and electronic equipment | |
| CN108494778A (en) | Identity identifying method and device | |
| CN110428399A (en) | Method, apparatus, equipment and storage medium for detection image | |
| CN108388889B (en) | Method and device for analyzing face image | |
| CN108062544A (en) | For the method and apparatus of face In vivo detection | |
| CN109919244A (en) | Method and apparatus for generating scene Recognition model | |
| CN111539903B (en) | Method and device for training face image synthesis model | |
| CN110298850B (en) | Segmentation method and device for fundus image |
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 |