CN110472491A - Abnormal face detecting method, abnormality recognition method, device, equipment and medium - Google Patents
Abnormal face detecting method, abnormality recognition method, device, equipment and medium Download PDFInfo
- Publication number
- CN110472491A CN110472491A CN201910603391.9A CN201910603391A CN110472491A CN 110472491 A CN110472491 A CN 110472491A CN 201910603391 A CN201910603391 A CN 201910603391A CN 110472491 A CN110472491 A CN 110472491A
- Authority
- CN
- China
- Prior art keywords
- image
- face
- abnormal
- video information
- detected
- 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
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 178
- 238000000034 method Methods 0.000 title claims abstract description 85
- 230000005856 abnormality Effects 0.000 title claims abstract description 21
- 230000001815 facial effect Effects 0.000 claims abstract description 109
- 238000001514 detection method Methods 0.000 claims abstract description 33
- 239000000284 extract Substances 0.000 claims abstract description 21
- 238000005516 engineering process Methods 0.000 claims abstract description 17
- 238000000605 extraction Methods 0.000 claims abstract description 14
- 238000005520 cutting process Methods 0.000 claims abstract description 12
- 239000013598 vector Substances 0.000 claims description 89
- 238000004590 computer program Methods 0.000 claims description 20
- 238000004422 calculation algorithm Methods 0.000 claims description 17
- 238000003860 storage Methods 0.000 claims description 15
- 230000002547 anomalous effect Effects 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000011218 segmentation Effects 0.000 claims description 5
- 210000000887 face Anatomy 0.000 description 29
- 238000010586 diagram Methods 0.000 description 8
- 239000000203 mixture Substances 0.000 description 8
- 230000006870 function Effects 0.000 description 7
- 238000012545 processing Methods 0.000 description 4
- 238000009826 distribution Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 230000010485 coping Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 210000004709 eyebrow Anatomy 0.000 description 1
- 230000008921 facial expression Effects 0.000 description 1
- 210000004209 hair Anatomy 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
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 invention discloses a kind of abnormal face detecting method, abnormality recognition method, device, equipment and media, which comprises video information cutting is N number of sub-video information by the video information for obtaining user；N number of sub-video information is separately disassembled into image to be detected, obtains N number of image to be detected sequence；Face datection is carried out using feature sub-face technology to N number of image to be detected sequence, extracts facial image；Characteristics extraction is carried out to the corresponding facial image of each spliting node, obtains the face characteristic of each facial image；Each face characteristic corresponding to each spliting node carries out similarity mode, judges in image to be detected sequence with the presence or absence of different faces image；If there are different faces images in image to be detected sequence, indicating risk operation is carried out to the video information of user.Using this method, the accuracy rate of abnormal face detection can be improved, meanwhile, improve the efficiency of abnormal face detection.
Description
Technical field
The present invention relates to field of image detection more particularly to a kind of abnormal face detecting method, abnormality recognition method, dresses
It sets, equipment and medium.
Background technique
With the fast development of computer technology, more and more scenes or task realize and turn under line on line
Change, for many scenes or task carry out provide biggish convenience, do not need user completed into fixed place it is cumbersome
The typing of information.However, the authenticity in order to guarantee data input, needs to detect the operating process on the line, especially
It is the detection to abnormal face (non-user), to guarantee the safety of operating process.Abnormal face detection, is computer
One of most active research topic in visual field, at present in the security and early warning of supermarket, bank, the hub of communication and hospital
Intelligent aspect has a wide range of applications.However, there is efficiency and accuracy rate are low for current abnormal face detecting method
Under problem.
Summary of the invention
The embodiment of the present invention provides a kind of abnormal face detecting method, device, computer equipment and storage medium to solve
The lower problem of abnormal face detection efficiency.
In addition, the embodiment of the present invention provides abnormality recognition method, device, computer equipment and storage medium to solve exception
The lower problem of accuracy of identification.
A kind of abnormal face detecting method, comprising:
The video information of user is obtained, the video information is provided with N number of spliting node, and N is positive integer；
It is N number of sub-video information by the video information cutting based on N number of spliting node；
N number of sub-video information is separately disassembled into image to be detected, obtains N number of image to be detected sequence；
Face datection is carried out using feature sub-face technology to N number of image to be detected sequence, extracts facial image；
The facial image corresponding to each spliting node carries out characteristics extraction, obtains each face figure
The face characteristic of picture；
Each face characteristic corresponding to each spliting node carries out similarity mode, judges described to be detected
It whether there is different faces image in image sequence；
If there are different faces images in image to be detected sequence, risk is carried out to the video information of the user
Prompt operation.
A kind of abnormal face detection device, comprising:
Acquiring video information module, for obtaining the video information of user, the video information is provided with N number of segmented section
Point, N are positive integer；
The video information cutting is N number of sub-video letter for being based on N number of spliting node by Video segmentation module
Breath；
Image sequence decomposing module obtains N number of for N number of sub-video information to be separately disassembled into image to be detected
Image to be detected sequence；
Facial image obtains module, for carrying out face using feature sub-face technology to N number of image to be detected sequence
Detection, extracts facial image；
Face characteristic obtains module, mentions for carrying out characteristic value to the corresponding facial image of each spliting node
It takes, obtains the face characteristic of each facial image；
Facial image determining module, it is similar for being carried out to the corresponding each face characteristic of each spliting node
Degree matching judges in image to be detected sequence with the presence or absence of different faces image；
Indicating risk module, if for there are different faces images in image to be detected sequence, to the user
Video information carry out indicating risk operation.
A kind of abnormality recognition method, comprising:
The video information of M progress indicating risk operation is obtained, M is positive integer, wherein the progress indicating risk operation
Video information detected to obtain using above-mentioned abnormal face detecting method；
Extract the doubtful abnormal face image in the M video informations for carrying out indicating risk operation；
Using each doubtful abnormal people of the facial modeling algorithm positioning of Coarse-to-fine CNN network
The key feature points of face image；
The key feature points of each doubtful abnormal face image are linked in sequence, are obtained each described doubtful different
The feature vector of ordinary person's face image；
The distance of the feature vector of the abnormal face image is calculated two-by-two, if the distance is less than or equal to preset
Distance threshold, it is determined that the doubtful abnormal face image is abnormal face image.
A kind of anomalous identification device, comprising:
Acquiring video information module, for obtaining the video information of user, the video information is provided with N number of segmented section
Point, N are positive integer, wherein it is described carry out indicating risk operation video information be using above-mentioned abnormal face detecting method into
Row detection obtains；
The video information cutting is N number of sub-video letter for being based on N number of spliting node by Video segmentation module
Breath；
Image sequence decomposing module obtains N number of for N number of sub-video information to be separately disassembled into image to be detected
Image to be detected sequence；
Facial image obtains module, for carrying out face using feature sub-face technology to N number of image to be detected sequence
Detection, extracts facial image；
Face characteristic obtains module, mentions for carrying out characteristic value to the corresponding facial image of each spliting node
It takes, obtains the face characteristic of each facial image；
Facial image determining module, it is similar for being carried out to the corresponding each face characteristic of each spliting node
Degree matching judges in image to be detected sequence with the presence or absence of different faces image；
Indicating risk module, if for there are different faces images in image to be detected sequence, to the user
Video information carry out indicating risk operation.
A kind of computer equipment, including memory, processor and storage are in the memory and can be in the processing
The computer program run on device, the processor realize above-mentioned abnormal face detecting method when executing the computer program,
Alternatively, the processor realizes above-mentioned abnormality recognition method when executing the computer program.
A kind of computer readable storage medium, the computer-readable recording medium storage have computer program, the meter
Calculation machine program realizes above-mentioned abnormal face detecting method when being executed by processor, alternatively, the processor executes the computer
Above-mentioned abnormality recognition method is realized when program.
In above-mentioned abnormal face detecting method, device, computer equipment and storage medium, firstly, obtaining the video of user
Information, video information are provided with N number of spliting node, obtain significantly more efficient video information according to spliting node so as to subsequent；So
Afterwards, be based on N number of spliting node, be N number of sub-video information by video information cutting, be conducive to accelerate subsequent sub-video information into
Row is further processed, and improves the efficiency to video information process；Then, N number of sub-video information is separately disassembled into mapping to be checked
Picture obtains N number of image to be detected sequence；Next, carrying out face inspection using feature sub-face technology to N number of image to be detected sequence
It surveys, extracts facial image, so that the quality of the facial image extracted is further promoted, so as to improve abnormal face
The efficiency of detection；In turn, characteristics extraction is carried out to the corresponding facial image of each spliting node, obtains each facial image
Face characteristic improves the acquisition efficiency of face characteristic；Followed by each face characteristic corresponding to each spliting node carries out phase
It matches, is judged in image to be detected sequence with the presence or absence of different faces image like degree；Finally, if existing in image to be detected sequence
Different faces image then carries out indicating risk operation to video information, can detect abnormal face image easily and fast, can
Abnormal face image is detected easily and fast.At the same time, also to there are the corresponding video informations of different faces image to carry out wind
Danger prompt operation, so as to the subsequent fraud bring risk that can evade the corresponding abnormal user of abnormal face image, mentions
The high efficiency of abnormal face detection.
In above-mentioned exception people recognition methods, device, computer equipment and storage medium, mentioned firstly, obtaining M progress risk
Show the video information of operation, the video information of the carry out indicating risk operation is detect using abnormal face detecting method
It arrives；Then, the doubtful abnormal face image in the video information of M progress indicating risk operation is extracted；Then, it uses
The facial modeling algorithm of Coarse-to-fine CNN network positions the key feature of each doubtful abnormal face image
Point improves the accuracy of the key feature points of abnormal face image；Next, by the key of each doubtful abnormal face image
Characteristic point is linked in sequence, and obtains the feature vector of each doubtful abnormal face image；Finally, calculating feature vector two-by-two
Distance, if distance is less than or equal to distance threshold, it is determined that doubtful abnormal face image is abnormal face image, is greatly improved
The efficiency of anomalous identification.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below by institute in the description to the embodiment of the present invention
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is the application environment signal of abnormal face detecting method and abnormality recognition method provided in an embodiment of the present invention
Figure；
Fig. 2 is one exemplary diagram of abnormal face detecting method provided in an embodiment of the present invention；
Fig. 3 is another exemplary diagram of abnormal face detecting method provided in an embodiment of the present invention；
Fig. 4 is another exemplary diagram of abnormal face detecting method provided in an embodiment of the present invention；
Fig. 5 is a functional block diagram of abnormal face detection device provided in an embodiment of the present invention；
Fig. 6 is another functional block diagram of abnormal face detection device provided in an embodiment of the present invention；
Fig. 7 is one exemplary diagram of abnormality recognition method provided in an embodiment of the present invention；
Fig. 8 is a functional block diagram of anomalous identification device provided in an embodiment of the present invention；
Fig. 9 is a schematic diagram of computer equipment provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are some of the embodiments of the present invention, instead of all the embodiments.Based on this hair
Embodiment in bright, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall within the protection scope of the present invention.
Abnormal face detecting method provided by the present application can be applicable in the application environment such as Fig. 1, wherein client is logical
It crosses network to be communicated with server-side, server-side receives the video information that client is sent, and is then N number of by video information cutting
N number of sub-video information is then separately disassembled into image to be detected, N number of image to be detected sequence is obtained, to N by sub-video information
A image to be detected sequence carries out Face datection using feature sub-face technology, extracts facial image, and then to each segmented section
The corresponding facial image of point carries out characteristics extraction, the face characteristic of each facial image is obtained, finally, to each spliting node
Corresponding each face characteristic carries out similarity mode, judges to whether there is different faces image in image to be detected sequence, when
There are different faces images in image to be detected sequence, then carry out indicating risk operation to video information.Wherein, client can be with
But it is not limited to various personal computers, laptop, smart phone, tablet computer and portable wearable device.Service
End can be realized with the server cluster of the either multiple server compositions of independent server.
In one embodiment, as shown in Fig. 2, being applied to be illustrated for the server-side in Fig. 1 in this way, including
Following steps:
S10: obtaining the video information of user, and video information is provided with N number of spliting node, and N is positive integer.
Wherein, video information refers to that user carries out the video information of task processing.The acquisition channel of the video information includes
But be not limited to APP, the webpage of website or background data base etc..Spliting node refers to according to user in processing different type task
And the timing node being arranged, by taking the video information of user's demand for credit product as an example, which, which can be, fills in job note
The spliting node of position fills in the spliting node of annual income or fills in the spliting node of contact information or fill in guarantor etc.
The spliting node of information.N is the number of spliting node, for example, when user is when filling in the information material of a certain credit product, it should
The application system of credit product will fill in the spliting node of annual income, fills in the spliting node of contact information and fill in guarantor
Etc. information timing node as spliting node, i.e. N=3.Specifically, the different product of the size concrete foundation of N is configured,
It is not construed as limiting herein.It is to be appreciated that the amount of image information that the entire video of video information includes is very complicated, thus by pair
Spliting node is arranged in video information, obtains significantly more efficient video information according to spliting node so as to subsequent.
S20: being based on N number of spliting node, is N number of sub-video information by video information cutting.
Wherein, sub-video information referred to from each spliting node time started to one section in the spliting node end time
Video information namely N number of spliting node correspond to N number of sub-video information, are handled for subsequent sub-video information, and figure is improved
As treatment effeciency.Specifically, video information can be carried out by cutting using MP4 file division algorithm, implements process are as follows:
Firstly, parsing to video information, Key Frames List is got；Then, the period to be divided is selected according to spliting node
(such as since key frame)；Then, key frame is regenerated into metadata moov box；Finally, copying corresponding metadata
Moov box generates new file namely sub-video information.It is to be appreciated that since sub-video information only includes a segmented section
The sub-video information of point is conducive to accelerate subsequent antithetical phrase view so that the effective information that sub-video information includes more is simplified
Frequency information is further processed, and improves the efficiency to video information process.
S30: N number of sub-video information is separately disassembled into image to be detected, obtains N number of image to be detected sequence.
Wherein, image to be detected sequence refer to composition sub-video information in multiple frame images combination, for image into
Row recognition of face and face alignment, correspondingly, the corresponding image to be detected sequence of each sub-video information.It specifically, can be with
Sub-video information is decomposed into image to be detected by the AviFileInfo () function carried in Opencv, by sub-video information
Video flowing is read out, and saves come out frame by frame, and multiple frame images namely image to be detected sequence can be obtained.
S40: Face datection is carried out using feature sub-face technology to N number of image to be detected sequence, extracts facial image.
Wherein, feature sub-face technology is a kind of viewpoint from statistics, finds the basic element of facial image distribution, i.e. face
The feature vector of image pattern collection covariance matrix, the method for detecting human face of facial image is approximatively characterized with this.Spy therein
Sign vector is also referred to as eigenface, and this feature face reflects the structure pass for lying in information and face inside face sample set
System.The feature vector of sample set covariance matrix of eyes, cheek, lower jaw is known as eigen eyes, feature jaw and feature lip, is referred to as
Sub-face of feature.Sub-face of feature generated subspace, referred to as sub-face space in corresponding image space.Specifically, calculate it is N number of to
In detection image sequence each image to be detected sub-face space projector distance, if projector distance meets threshold value comparison condition,
Then judge image to be detected for face.Wherein, facial image refers to include facial image, refers to that face is clear in the present embodiment
Degree reaches the facial image of preset threshold.
It should be noted that also judging the clarity of image to be detected in the present embodiment, and weed out clarity
Lower image to be detected.Definition judgment method therein is that channel (such as gray value) a certain in image to be detected is passed through drawing
This mask of pula carries out convolution algorithm, then calculates the numerical value of the standard deviation of convolution results as picture clarity size.It can be with
Understand ground, Laplace operator is used to measure the second dervative of image to be detected, and intensity quickly changes in prominent image to be detected
Region, such as the edge of image to be detected, if image to be detected variance is smaller, which has relatively narrow frequency response model
It encloses, i.e., the amount of edge in image to be detected is seldom, that is, picture is fuzzyyer.It is special by being used to N number of image to be detected sequence
It levies sub- face technology and carries out Face datection, and the clarity of image to be detected is judged, so that the matter of the facial image extracted
Further promotion is measured, so as to improve the efficiency of abnormal face detection.
S50: characteristics extraction is carried out to the corresponding facial image of each spliting node, obtains the face of each facial image
Feature.
Wherein, face characteristic refers to the key message of reflection face information, such as geometrical characteristic (such as face five of facial image
Official's characteristic point and facial contour feature point) and facial image gray feature (such as face complexion), for knowing to facial image
Not.Preferably, by extracting geometrical characteristic namely crucial point location including human face five-sense-organ and facial contour in the present embodiment
The characteristic point of crucial point location is as face characteristic value.Specifically, Eigenvalue Extraction Method can be Coarse-to-fine
The facial modeling algorithm of CNN network, the facial modeling for being also possible to ASM (Active Shape Model) are calculated
Method, the algorithm are that Global Face appearance establishes universal model, are steady keys to local image damage, but its calculating cost is very
Height needs a large amount of iterative steps.It can also be that the facial modeling algorithm of AAM (Active Appreance Model) obtains
Face characteristic is taken, positioning feature point is directly regarded as a recurrence task by this method, calculates spy with a global recurrence device
Levy the coordinate of point.Since facial modeling is easy to be influenced by factors such as human face expression, posture, illumination, so as to cause fixed
Position accuracy rate reduces.Meanwhile the positioning difficulty of face different location characteristic point is different, and is carried out only by single model
Positioning is difficult to ensure the accuracy rate of positioning.
As the preferred of the present embodiment, the characteristic point of different location is positioned using multi-model, i.e., face is divided into inside
Point group (51 points) positions respectively with boundary point group (17 points).Namely the face characteristic using Coarse-to-fine CNN network
Point location algorithm obtain the human face five-sense-organ on each facial image 51 Inner points and predict face outer profile 17
A Contour points obtains face characteristic.Specifically, DCNN model is divided into two parallel CNN cascade networks, wherein
One group of mould be 4 grades of cascade CNN networks, for calculating the position of inter characteristic points () at 51 points, the 1st grade for estimating internal spy
The bounding box of point is levied, the 2nd grade of position for characteristic point according to a preliminary estimate, the 3rd group for further accurate estimation each group
At the position (left and right eye, left and right eyebrow, nose, 6 parts of mouth) of Partial Feature point, the 4th grade is to having carried out rotation school
6 parts after just carry out characteristic point accurate positioning.Another group model is 2 grades of cascade CNN networks, and the 1st group for estimating side
The bounding box of boundary's point (17 points), the 2nd grade for estimating the accurate location of 17 edge feature points.It is to be appreciated that by
In traditional DCNN model in priori knowledge deficiency, the most strength of convolutional network can be made all to be wasted in searching face
On, the efficiency of facial modeling is reduced, therefore, in the present embodiment, by the 1st grade of two groups of CNN in parallel all fixed
On position bounding box (bounding box), Face datection efficiency can be improved, and then improve the acquisition efficiency of face characteristic.
S60: each face characteristic corresponding to each spliting node carries out similarity mode, judges image to be detected sequence
In whether there is different faces image.
Wherein, similarity mode refer to calculate any two face characteristic similarity degree as matching index, for pair
Facial image is compared.Similarity therein can be Euclidean distance, can also can also be Hamming distance with COS distance.
Specifically, each face characteristic is converted into feature vector, then calculates the similarity of any feature vector two-by-two, works as similarity
When greater than similarity threshold, then confirm that the two width facial image is otherwise same people confirms that the two width facial image is not same
People.Similarity threshold therein is for measuring the minimum similarity degree that two width facial images are same people, for example, the similarity threshold
Value can be 87.5%.
S70: if there are different faces images in image to be detected sequence, indicating risk is carried out to the video information of user
Operation.
Specifically, when there are at least have 2 in multiple people namely image to be detected sequence in image to be detected sequence
People can then determine that there are abnormal faces in the sub-video information of the corresponding user of the spliting node.Ground is readily appreciated that, in credit
In application process, whole process should carry out application materials by same user (user) and fill in, and more people, then can determine if it exists
There are scene namely fraud that intermediary carries out task processing in corresponding task, there are abnormal face images, therefore, right
The video information that user fills in application materials carries out indicating risk operation.In the present embodiment, by detecting image to be detected sequence
In whether there is different faces image, abnormal face image can be detected easily and fast, at the same time, also to there are different peoples
The corresponding video information of face image carries out indicating risk operation, can evade the corresponding abnormal use of abnormal face image so as to subsequent
The fraud bring risk at family improves the efficiency of abnormal face detection.
In the present embodiment, firstly, obtaining the video information of user, video information is provided with N number of spliting node, so as to subsequent
Significantly more efficient video information is obtained according to spliting node；Then, it is based on N number of spliting node, is N number of son by video information cutting
Video information, is conducive to accelerate subsequent sub-video information to be further processed, and improves the efficiency to video information process；It connects
, N number of sub-video information is separately disassembled into image to be detected, obtains N number of image to be detected sequence；Next, to N number of to be checked
Altimetric image sequence carries out Face datection using feature sub-face technology, extracts facial image, so that the matter of the facial image extracted
Further promotion is measured, so as to improve the efficiency of abnormal face detection；In turn, to the corresponding face of each spliting node
Image carries out characteristics extraction, obtains the face characteristic of each facial image, improves the acquisition efficiency of face characteristic；Followed by,
Each face characteristic corresponding to each spliting node carries out similarity mode, judges in image to be detected sequence with the presence or absence of not
Same facial image；Finally, carrying out indicating risk behaviour to video information if there are different faces images in image to be detected sequence
Make, abnormal face image can be detected easily and fast, abnormal face image can be detected easily and fast.At the same time, also
To there are the corresponding video informations of different faces image to carry out indicating risk operation, abnormal face image can be evaded so as to subsequent
The fraud bring risk of corresponding abnormal user improves the efficiency of abnormal face detection.
In one embodiment, as shown in figure 3, in step S40, feature sub-face technology is used to N number of image to be detected sequence
Face datection is carried out, facial image is extracted, comprising:
S41: the matrix based on preset face space vector obtains each image to be detected in image to be detected sequence
Human face region.
Wherein, face space vector refers to the space vector for presetting the distribution element composition of face template, for making
For the basis on location of human face region in image to be detected.Specifically, by the corresponding matrix of facial image to be measured and preset face
The matrix of space vector carries out template matching, determines the human face region of each image to be detected.
S42: human face region is calculated in the projector distance in preset face space.
Wherein, projector distance refers in the corresponding vector set of human face region, vector distance corresponding with face space
Minimum value, namely the region nearest apart from face template.Specifically, the corresponding vector of each human face region and face space are calculated
The distance of vector, and will be apart from minimum value as projector distance.It is to be appreciated that by calculating human face region in preset face
The projector distance in space can effectively overcome illumination, block, the shadow that the factors such as lesser visual angle change detect human face region
It rings, therefore the accuracy and speed of abnormal face detection can be improved.
S43: if projector distance is in preset projector distance threshold range, it is determined that be checked in image to be detected sequence
Altimetric image is facial image, and extracts facial image.
Specifically, when projector distance is in preset projector distance threshold range, it is determined that in image to be detected sequence
Image to be detected is facial image, and extracts facial image.It is to be appreciated that when projector distance satisfaction judges the item of facial image
When part, so that the corresponding human face region of the projector distance be extracted, facial image is obtained, ensure that the accurate of facial image
Property, facial image is detected so as to subsequent.
In the present embodiment, firstly, the matrix based on preset face space vector, obtains every in image to be detected sequence
The human face region of one image to be detected；Then, human face region is calculated in the projector distance in preset face space, effectively overcomes light
According to, block, the influence that the factors such as lesser visual angle change detect human face region, therefore the standard of abnormal face detection can be improved
True property and speed；Finally, if projector distance is in preset projector distance threshold range, it is determined that in image to be detected sequence
Image to be detected is facial image, and extracts facial image, ensure that the accuracy of facial image, so as to subsequent to facial image
It is detected.
In one embodiment, as shown in figure 4, in step S60, each face characteristic corresponding to each spliting node is carried out
Similarity mode judges to specifically comprise the following steps: in image to be detected sequence with the presence or absence of different faces image
S61: each face characteristic is converted into face feature vector, obtains the face being made of L face feature vector
Feature vector set, wherein L is positive integer.
Face characteristic in the present embodiment is geometrical characteristic, including 51 face five features points and 17 facial contour spies
Point is levied, i.e., one shares 68 characteristic points, under conventional coordinates, the position coordinates of each feature can be determined, then according to cross
68 coordinates are linked in sequence by the size of coordinate or ordinate, obtain the face feature vector that length is 68.It is therein
L refers to the number of face feature vector, since a face characteristic corresponds to a face vector, L namely one is to be detected
Facial image number in sequence, it is laggard which can carry out Face datection by the feature sub-face technology in step S40
Row determines, such as has detected 8 facial images, that is, is corresponding with 8 face characteristics, then L is 8.
S62: being successively directed to each face feature vector, calculates separately in face feature vector and face feature vector set
Remaining L-1 face feature vector distance, obtain L* (L-1)/2 target range.
Specifically, all face feature vectors are calculated two-by-two respectively, that is, is directed to each face feature vector, respectively
Face feature vector is calculated at a distance from remaining L-1 face feature vector in face feature vector set, obtains L-1 mesh
Therefore subject distance when successively calculating L face feature vector, is for L face feature vector, needs to carry out L* (L- altogether
1)/2 time distance calculates, and obtains L* (L-1)/2 target range.
S63: being greater than the number of preset vector distance threshold value in statistics target range, if number is more than or equal to 2, it is determined that
There are different faces images in image to be detected sequence.
Wherein, preset vector distance threshold value refers to for measuring whether any two image to be detected is the same person's
The critical value of vector distance, when target range be greater than preset distance threshold when, illustrate the target range it is corresponding two it is to be checked
Surveying facial image is not same people.Specifically, judge target range and preset distance threshold size, and to greater than target range
In be greater than preset distance threshold number counted, when the number of statistics be more than or equal to 2 when, then can determine mapping to be checked
As at least existing in 2 different people namely image to be detected sequence in sequence, there are different faces images.It is to be appreciated that
In the present embodiment, it is only necessary to the number for being greater than preset distance threshold in target range in target range is simply judged, according to system
The number of meter can determine in image to be detected sequence with the presence or absence of different faces image.Further, when counting on target
When the number for being greater than preset vector distance threshold value in distance in target range is 2, it will be able to easily and accurately determine detection figure
As there are different faces images in sequence, and the redundancy of effort for continuing statistics is eliminated, so as in image to be detected sequence
Judgement with the presence or absence of different faces image is more accurate and quick.
In the present embodiment, firstly, each face characteristic is converted into face feature vector, obtain from L face characteristic to
Measure the face feature vector set of composition；Secondly, successively be directed to each face feature vector, calculate separately face feature vector with
The distance of remaining L-1 face feature vector in face feature vector set obtains L* (L-1)/2 target range；Finally,
The number for being greater than preset vector distance threshold value in target range is counted, if number is more than or equal to 2, it is determined that image to be detected sequence
There are different faces images in column, so that more accurate to the judgement that whether there is different faces image in image to be detected sequence
With it is quick.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process
Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present invention constitutes any limit
It is fixed.
In one embodiment, a kind of abnormal face detection device is provided, the abnormal face detection device and above-described embodiment
Middle abnormal face detecting method corresponds.As shown in figure 5, the abnormal face detection device includes acquiring video information module
10, Video segmentation module 20, image sequence decomposing module 30, facial image obtain module 40, face characteristic obtains module 50, people
Face image determining module 60 and indicating risk module 70.Detailed description are as follows for each functional module:
Acquiring video information module 10, for obtaining the video information of user, video information is provided with N number of spliting node, N
For positive integer；
Video information cutting is N number of sub-video information for being based on N number of spliting node by Video segmentation module 20；
Image sequence decomposing module 30, for N number of sub-video information to be separately disassembled into image to be detected, obtain it is N number of to
Detection image sequence；
Facial image obtains module 40, for carrying out face inspection using feature sub-face technology to N number of image to be detected sequence
It surveys, extracts facial image；
Face characteristic obtains module 50, for carrying out characteristics extraction to the corresponding facial image of each spliting node, obtains
To the face characteristic of each facial image；
Facial image determining module 60 carries out similarity for each face characteristic corresponding to each spliting node
Match, judges in image to be detected sequence with the presence or absence of different faces image；
Indicating risk module 70, if for there are different faces images in image to be detected sequence, to the video of user
Information carries out indicating risk operation.
Preferably, as shown in fig. 6, it includes human face region acquiring unit 41, projector distance meter that facial image, which obtains module 40,
Calculate unit 42 and facial image extraction unit 43.
Human face region acquiring unit 41 obtains image to be detected sequence for the matrix based on preset face space vector
The human face region of each image to be detected in column；
Projector distance computing unit 42, for calculating human face region in the projector distance in preset face space；
Facial image extraction unit 43, if for projector distance in preset projector distance threshold range, it is determined that
Image to be detected in detection image sequence is facial image, and extracts facial image.
Preferably, facial image determining module includes feature vector converting unit, target range computing unit and face figure
As detection unit.
Feature vector converting unit obtains special by L face for each face characteristic to be converted into face feature vector
Levy the face feature vector set of vector composition, wherein L is positive integer；
Target range computing unit, for successively be directed to each face feature vector, calculate separately face feature vector with
The distance of remaining L-1 face feature vector in face feature vector set obtains L* (L-1)/2 target range；
Facial image detection unit, for counting the number for being greater than preset vector distance threshold value in target range, if a
Number is more than or equal to 2, it is determined that there are different faces images in image to be detected sequence.
In one embodiment, an abnormality recognition method is provided, which can also apply the application in such as Fig. 1
In environment, wherein client is communicated by network with server-side.Server-side receives the M progress risk that client is sent
The video information of operation is prompted, and the video information for carrying out indicating risk operation is examined using abnormal face detecting method
It measures；Then the doubtful abnormal face image in the video information of M progress indicating risk operation is extracted；Then, it uses
The facial modeling algorithm of Coarse-to-fine CNN network positions the key feature of each doubtful abnormal face image
Point；And then the key feature points of each doubtful abnormal face image are linked in sequence, obtain each doubtful abnormal face figure
The feature vector of picture；Finally, the distance of feature vector is calculated two-by-two, if distance is less than or equal to distance threshold, it is determined that doubt
It is abnormal face image like abnormal face image.Wherein, client can be, but not limited to be various personal computers, notebook electricity
Brain, smart phone, tablet computer and portable wearable device.Server-side can use the either multiple services of independent server
The server cluster of device composition is realized.
In one embodiment, as shown in fig. 7, being applied to be illustrated for the server-side in Fig. 1 in this way, including
Following steps:
S80: the video information of M progress indicating risk operation is obtained, M is positive integer, wherein carry out indicating risk operation
Video information detected to obtain using abnormal face detecting method.
Wherein, the video information for carrying out indicating risk operation refers to that, there are the video information of abnormal face image, M carries out wind
The quantity of the video information of danger prompt operation, can be according to practical specific setting.And the video letter of this M progress indicating risk operation
Breath is detected to obtain using abnormal face detecting method, it is possible to understand that ground, due to the exception of step S10- step S70
Method for detecting human face has stronger detection efficiency, therefore, the video information of the progress indicating risk operation in the present embodiment
It is highly efficient accurate.
S90: the doubtful abnormal face image in the video information of M progress indicating risk operation is extracted.
Wherein, doubtful abnormal face image refers to facial image to be identified.Since the doubtful facial image is from multiple
It is extracted in the user video information of progress indicating risk operation, is after being detected using abnormal face detecting method
User video information obtained in, therefore so that extract doubtful abnormal face image it is more accurate.
S100: using each doubtful abnormal people of the facial modeling algorithm positioning of Coarse-to-fine CNN network
The key feature points of face image.
Wherein, the facial modeling algorithm of Coarse-to-fine CNN network is a kind of for positioning multiple faces
Characteristic point realizes the algorithm of the high accuracy positioning of 68 human face characteristic points.Specifically, the key feature of doubtful abnormal face image
The realization process of point location algorithm is consistent with step S60, and details are not described herein again.Due to the cascade network point of Coarse-to-fine
The complexity and training for having dissipated network are born, and the accuracy of the key feature points of abnormal face image is improved.
S110: the key feature points of each doubtful abnormal face image are linked in sequence, each doubtful exception is obtained
The feature vector of facial image.
Specifically, the key feature points of each doubtful abnormal face image are coordinate form, therefore, by each coordinate into
It goes after being linked in sequence, as the feature vector of a multi-C vector namely doubtful abnormal face image.
S120: calculating the distance of the feature vector of doubtful abnormal face image two-by-two, if distance is less than or equal to distance
Threshold value, it is determined that doubtful abnormal face image is abnormal face image.
Wherein, distance threshold be for measure two abnormal face images whether be the same person feature vector distance
Critical value.Specifically, the doubtful abnormal face image in each video information for carrying out indicating risk operation is calculated separately
Feature vector and the feature vector of the doubtful abnormal face image in the video information of remaining M-1 progress indicating risk operation
Distance can determine doubting in the video information of the carry out indicating risk operation if distance is less than or equal to distance threshold
It is abnormal face image like abnormal face image.It is to be appreciated that abnormal face image if it does not exist, then M progress risk mentions
Showing just has M facial image in the video information of operation, however, by abnormal face detecting method detected into
There is abnormal abnormal image in the video information of row indicating risk operation, therefore, by calculate calculate two-by-two feature vector away from
From can determine which doubtful abnormal face image be abnormal face image.By calculating the distance of feature vector two-by-two, when
When distance is less than or equal to distance threshold, it is determined that doubtful abnormal face image is abnormal face image, is substantially increased different
The other efficiency of common sense.
In the present embodiment, firstly, obtaining the video information of M progress indicating risk operation, the carry out indicating risk operation
Video information detected to obtain using abnormal face detecting method；Then, M progress indicating risk operation is extracted
Doubtful abnormal face image in video information；Then, using the facial modeling of Coarse-to-fine CNN network
Algorithm positions the key feature points of each doubtful abnormal face image, improves the accurate of the key feature points of abnormal face image
Property；Next, the key feature points of each doubtful abnormal face image are linked in sequence, each doubtful abnormal face is obtained
The feature vector of image；Finally, the distance of feature vector is calculated two-by-two, if distance is less than or equal to distance threshold, it is determined that
Doubtful abnormal face image is abnormal face image, substantially increases the efficiency of anomalous identification.
In one embodiment, as shown in fig. 6, after step S120, determining that the doubtful abnormal face image is different
After ordinary person's face image, the abnormality recognition method further include:
Each abnormal face image is formed into abnormal face library, and each abnormal face image is provided with spliting node mark
Know.
Wherein, abnormal face library refers to the database of all abnormal face image compositions, for being used as abnormal face image
The standard of judgement.Specifically, by the way that abnormal face image is formed abnormal face library by each, the knowledge to abnormal face is improved
Other efficiency, meanwhile, the spliting node where quickly determining abnormal face can be convenient according to the spliting node of setting mark, with
Early warning prompting is carried out for the corresponding personnel of abnormal face image in the application case there are fraud.It specifically, can be with
By analyzing abnormal face amount of images in more each spliting node, it is greater than for the abnormal face amount of images identified pre-
If the operation of the spliting node of threshold value carries out early warning, the probability that fraud occurs is reduced.
In the present embodiment, by the way that abnormal face image is formed abnormal face library by each, improve to abnormal face
Recognition efficiency, meanwhile, the spliting node where quickly determining abnormal face can be convenient according to the spliting node of setting mark,
For there are the corresponding personnel of the abnormal face image of fraud to carry out early warning prompting.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process
Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present invention constitutes any limit
It is fixed.
In one embodiment, a kind of anomalous identification device is provided, which knows extremely with above-described embodiment
Other method corresponds.As shown in figure 8, the anomalous identification device includes risk video acquiring module 80, doubtful abnormal face figure
As extraction module 90, positioning feature point module 100, vector obtain module 110 and abnormal face image determining module 120.Each function
Detailed description are as follows for energy module:
Risk video acquiring module 80, for obtaining the video information of M progress indicating risk operation, M is positive integer,
In, the video information for carrying out indicating risk operation is to be detected to obtain using the abnormal face detecting method in above-described embodiment
's；
Doubtful abnormal face image zooming-out module 90, for extracting in the video information that M carries out indicating risk operation
Doubtful abnormal face image；
Positioning feature point module 100, for the facial modeling algorithm using Coarse-to-fine CNN network
Position the key feature points of each doubtful abnormal face image；
Vector obtains module 110, and the key feature points of each doubtful abnormal face image are linked in sequence, and obtains every
The feature vector of one doubtful abnormal face image；
Abnormal face image determining module 120, the distance of the feature vector for calculating abnormal face image two-by-two, if away from
From less than or equal to preset distance threshold, it is determined that doubtful abnormal face image is abnormal face image.
Preferably, which further includes face database comprising modules, for forming each abnormal face image
Abnormal face library, and each abnormal face image is provided with spliting node mark.
Specific about anomalous identification device limits the restriction that may refer to above for abnormality recognition method, herein not
It repeats again.Modules in above-mentioned anomalous identification device can be realized fully or partially through software, hardware and combinations thereof.On
Stating each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also store in a software form
In memory in computer equipment, the corresponding operation of the above modules is executed in order to which processor calls.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition can be as shown in Figure 9.The computer equipment include by system bus connect processor, memory, network interface and
Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment
Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data
Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The database of machine equipment is for storing the data used in abnormal face detecting method.The network interface of the computer equipment is used
It is communicated in passing through network connection with external terminal.To realize a kind of abnormal face inspection when the computer program is executed by processor
Survey method.
In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory
And the computer program that can be run on a processor, processor realize the abnormal people in above-described embodiment when executing computer program
Face detecting method or processor realize the abnormality recognition method in above-described embodiment when executing computer program.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program realizes that abnormal face detecting method or processor in above-described embodiment execute the calculating when being executed by processor
The abnormality recognition method in above-described embodiment is realized when machine program.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each function
Can unit, module division progress for example, in practical application, can according to need and by above-mentioned function distribution by different
Functional unit, module are completed, i.e., the internal structure of described device is divided into different functional unit or module, more than completing
The all or part of function of description.
Embodiment described above is merely illustrative of the technical solution of the present invention, rather than its limitations；Although referring to aforementioned reality
Applying example, invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features；And these are modified
Or replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution should all
It is included within protection scope of the present invention.
Claims (10)
1. a kind of abnormal face detecting method, which is characterized in that the abnormal face detecting method includes:
The video information of user is obtained, the video information is provided with N number of spliting node, and N is positive integer；
It is N number of sub-video information by the video information cutting based on N number of spliting node；
N number of sub-video information is separately disassembled into image to be detected, obtains N number of image to be detected sequence；
Face datection is carried out using feature sub-face technology to N number of image to be detected sequence, extracts facial image；
The facial image corresponding to each spliting node carries out characteristics extraction, obtains each facial image
Face characteristic；
Each face characteristic corresponding to each spliting node carries out similarity mode, judges described image to be detected
It whether there is different faces image in sequence；
If there are different faces images in image to be detected sequence, indicating risk is carried out to the video information of the user
Operation.
2. abnormal face detecting method as described in claim 1, which is characterized in that described to N number of image to be detected sequence
Column carry out Face datection using feature sub-face technology, extract facial image, comprising:
Based on the matrix of preset face space vector, the people of each image to be detected in image to be detected sequence is obtained
Face region；
The human face region is calculated in the projector distance in preset face space；
If the projector distance is in preset projector distance threshold range, it is determined that be checked in image to be detected sequence
Altimetric image is facial image, and extracts the facial image.
3. abnormal face detecting method as described in claim 1, which is characterized in that described corresponding to each spliting node
Each face characteristic carry out similarity mode, judge in image to be detected sequence with the presence or absence of different faces figure
Picture, comprising:
Each face characteristic is converted into face feature vector, obtains the face characteristic being made of L face feature vector
Vector set, wherein L is positive integer；
It is successively directed to each face feature vector, calculates separately the face feature vector and the face feature vector collection
The distance of remaining L-1 face feature vectors in conjunction, obtains L* (L-1)/2 target range；
The number for being greater than preset vector distance threshold value in the target range is counted, if the number is more than or equal to 2, it is determined that
There are different faces images in image to be detected sequence.
4. a kind of abnormality recognition method, which is characterized in that the abnormality recognition method includes:
The video information of M progress indicating risk operation is obtained, M is positive integer, wherein the view for carrying out indicating risk operation
Frequency information is detected to obtain using abnormal face detecting method as described in any one of claims 1 to 3；
Extract the doubtful abnormal face image in the M video informations for carrying out indicating risk operation；
Each doubtful abnormal face figure is positioned using the facial modeling algorithm of Coarse-to-fine CNN network
The key feature points of picture；
The key feature points of each doubtful abnormal face image are linked in sequence, each doubtful abnormal people is obtained
The feature vector of face image；
The distance of the feature vector of the abnormal face image is calculated two-by-two, if the distance is less than or equal to preset distance
Threshold value, it is determined that the doubtful abnormal face image is abnormal face image.
5. abnormality recognition method as claimed in claim 4, which is characterized in that in the determination doubtful abnormal face image
After abnormal face image, the abnormality recognition method further include:
Each abnormal face image is formed into abnormal face library, and each abnormal face image is provided with spliting node
Mark.
6. a kind of abnormal face detection device, which is characterized in that the abnormal face detection device includes:
Acquiring video information module, for obtaining the video information of user, the video information is provided with N number of spliting node, and N is
Positive integer；
The video information cutting is N number of sub-video information for being based on N number of spliting node by Video segmentation module；
Image sequence decomposing module obtains N number of to be checked for N number of sub-video information to be separately disassembled into image to be detected
Altimetric image sequence；
Facial image obtains module, for carrying out Face datection using feature sub-face technology to N number of image to be detected sequence,
Extract facial image；
Face characteristic obtains module, for carrying out characteristics extraction to the corresponding facial image of each spliting node,
Obtain the face characteristic of each facial image；
Facial image determining module, for carrying out similarity to the corresponding each face characteristic of each spliting node
Match, judges in image to be detected sequence with the presence or absence of different faces image；
Indicating risk module, if for there are different faces images in image to be detected sequence, to the view of the user
Frequency information carries out indicating risk operation.
7. abnormal face detection device as claimed in claim 6, which is characterized in that the facial image obtains module, comprising:
Human face region acquiring unit obtains image to be detected sequence for the matrix based on preset face space vector
In each image to be detected human face region；
Projector distance computing unit, for calculating the human face region in the projector distance in preset face space；
Facial image extraction unit, if for the projector distance in preset projector distance threshold range, it is determined that described
Image to be detected in image to be detected sequence is facial image, and extracts the facial image.
8. a kind of anomalous identification device, which is characterized in that the anomalous identification device includes:
Risk video acquiring module, for obtaining the video information of M progress indicating risk operation, M is positive integer, wherein institute
Stating and carrying out the video information of indicating risk operation is using abnormal face detecting method as described in any one of claims 1 to 3
It is detected；
Doubtful abnormal face image zooming-out module, for extracting doubting in the M video informations for carrying out indicating risk operation
Like abnormal face image；
Positioning feature point module, it is each for the facial modeling algorithm positioning using Coarse-to-fine CNN network
The key feature points of the doubtful abnormal face image；
Vector obtains module, and the key feature points of each doubtful abnormal face image are linked in sequence, are obtained each
The feature vector of the doubtful abnormal face image；
Abnormal face image determining module, the distance of the feature vector for calculating the abnormal face image two-by-two, if described
Distance is less than or equal to preset distance threshold, it is determined that the doubtful abnormal face image is abnormal face image.
9. a kind of computer equipment, including memory, processor and storage are in the memory and can be in the processor
The computer program of upper operation, which is characterized in that the processor realized when executing the computer program as claim 1 to
Any one of 3 abnormal face detecting methods or the processor realize such as claim 4 when executing the computer program
To 5 described in any item abnormality recognition methods.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In, the abnormal face detecting method as described in any one of claims 1 to 3 is realized when the computer program is executed by processor,
Or the processor realizes such as claim 4 to 5 described in any item abnormality recognition methods when executing the computer program.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910603391.9A CN110472491A (en) | 2019-07-05 | 2019-07-05 | Abnormal face detecting method, abnormality recognition method, device, equipment and medium |
| PCT/CN2020/085571 WO2021004112A1 (en) | 2019-07-05 | 2020-04-20 | Anomalous face detection method, anomaly identification method, device, apparatus, and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910603391.9A CN110472491A (en) | 2019-07-05 | 2019-07-05 | Abnormal face detecting method, abnormality recognition method, device, equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110472491A true CN110472491A (en) | 2019-11-19 |
Family
ID=68507379
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910603391.9A Pending CN110472491A (en) | 2019-07-05 | 2019-07-05 | Abnormal face detecting method, abnormality recognition method, device, equipment and medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110472491A (en) |
| WO (1) | WO2021004112A1 (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111339990A (en) * | 2020-03-13 | 2020-06-26 | 乐鑫信息科技（上海）股份有限公司 | Face recognition system and method based on dynamic update of face features |
| CN111401197A (en) * | 2020-03-10 | 2020-07-10 | 支付宝(杭州)信息技术有限公司 | Picture risk identification method, device and equipment |
| CN111429638A (en) * | 2020-04-13 | 2020-07-17 | 汪辽宁 | Access control method based on voice recognition and face recognition |
| CN111639689A (en) * | 2020-05-20 | 2020-09-08 | 杭州海康威视系统技术有限公司 | Face data processing method and device and computer readable storage medium |
| CN112084893A (en) * | 2020-08-24 | 2020-12-15 | 中国银联股份有限公司 | Biological recognition terminal abnormity detection method, device, equipment and storage medium |
| CN112200009A (en) * | 2020-09-15 | 2021-01-08 | 青岛邃智信息科技有限公司 | Pedestrian re-identification method based on key point feature alignment in community monitoring scene |
| WO2021004112A1 (en) * | 2019-07-05 | 2021-01-14 | 深圳壹账通智能科技有限公司 | Anomalous face detection method, anomaly identification method, device, apparatus, and medium |
| CN112784712A (en) * | 2021-01-08 | 2021-05-11 | 重庆创通联智物联网有限公司 | Missing child early warning implementation method and device based on real-time monitoring |
| CN113449543A (en) * | 2020-03-24 | 2021-09-28 | 百度在线网络技术（北京）有限公司 | Video detection method, device, equipment and storage medium |
| CN114724074A (en) * | 2022-06-01 | 2022-07-08 | 共道网络科技有限公司 | Method and device for detecting risk video |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112969047B (en) * | 2021-02-22 | 2023-01-24 | 深圳市沃特沃德信息有限公司 | Control method and device of smart watch, smart watch and storage medium |
| CN113182273A (en) * | 2021-04-13 | 2021-07-30 | 深圳市美日净化科技有限公司 | Omnibearing air shower and air shower control method |
| CN113239742A (en) * | 2021-04-23 | 2021-08-10 | 郑州大学 | System and method for monitoring abnormal gathering behavior of people in community |
| CN113221725A (en) * | 2021-05-08 | 2021-08-06 | 上海智蕙林医疗科技有限公司 | Risk processing method and system |
| CN114944004A (en) * | 2022-07-25 | 2022-08-26 | 海易科技(北京)有限公司 | Face image storage method, device, equipment, computer medium and program product |
| CN115396485B (en) * | 2022-08-29 | 2023-06-06 | 浙江华盟电气有限公司 | Tower crane data interaction method and system based on Bluetooth box |
| CN115620210B (en) * | 2022-11-29 | 2023-03-21 | 广东祥利科技有限公司 | Method and system for determining performance of electronic wire material based on image processing |
| CN115810218A (en) * | 2022-12-20 | 2023-03-17 | 山东交通学院 | Method and system for detecting abnormal behaviors of people based on machine vision and target detection |
| CN117541957A (en) * | 2023-11-08 | 2024-02-09 | 继善(广东)科技有限公司 | Method, system and medium for generating event solving strategy based on artificial intelligence |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106599855A (en) * | 2016-12-19 | 2017-04-26 | 四川长虹电器股份有限公司 | Softmax-based face recognizing method |
| CN109543629A (en) * | 2018-11-26 | 2019-03-29 | 平安科技（深圳）有限公司 | A kind of blink recognition methods, device, equipment and readable storage medium storing program for executing |
| CN109871834A (en) * | 2019-03-20 | 2019-06-11 | 北京字节跳动网络技术有限公司 | Information processing method and device |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103577815B (en) * | 2013-11-29 | 2017-06-16 | 中国科学院计算技术研究所 | A kind of face alignment method and system |
| CN109961588A (en) * | 2017-12-26 | 2019-07-02 | 天地融科技股份有限公司 | A kind of monitoring system |
| CN109961587A (en) * | 2017-12-26 | 2019-07-02 | 天地融科技股份有限公司 | A kind of monitoring system of self-service bank |
| CN110472491A (en) * | 2019-07-05 | 2019-11-19 | 深圳壹账通智能科技有限公司 | Abnormal face detecting method, abnormality recognition method, device, equipment and medium |
-
2019
- 2019-07-05 CN CN201910603391.9A patent/CN110472491A/en active Pending
-
2020
- 2020-04-20 WO PCT/CN2020/085571 patent/WO2021004112A1/en active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106599855A (en) * | 2016-12-19 | 2017-04-26 | 四川长虹电器股份有限公司 | Softmax-based face recognizing method |
| CN109543629A (en) * | 2018-11-26 | 2019-03-29 | 平安科技（深圳）有限公司 | A kind of blink recognition methods, device, equipment and readable storage medium storing program for executing |
| CN109871834A (en) * | 2019-03-20 | 2019-06-11 | 北京字节跳动网络技术有限公司 | Information processing method and device |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021004112A1 (en) * | 2019-07-05 | 2021-01-14 | 深圳壹账通智能科技有限公司 | Anomalous face detection method, anomaly identification method, device, apparatus, and medium |
| CN111401197A (en) * | 2020-03-10 | 2020-07-10 | 支付宝(杭州)信息技术有限公司 | Picture risk identification method, device and equipment |
| CN111401197B (en) * | 2020-03-10 | 2023-08-15 | 支付宝(杭州)信息技术有限公司 | Picture risk identification method, device and equipment |
| CN111339990A (en) * | 2020-03-13 | 2020-06-26 | 乐鑫信息科技（上海）股份有限公司 | Face recognition system and method based on dynamic update of face features |
| CN111339990B (en) * | 2020-03-13 | 2023-03-24 | 乐鑫信息科技（上海）股份有限公司 | Face recognition system and method based on dynamic update of face features |
| CN113449543A (en) * | 2020-03-24 | 2021-09-28 | 百度在线网络技术（北京）有限公司 | Video detection method, device, equipment and storage medium |
| CN113449543B (en) * | 2020-03-24 | 2022-09-27 | 百度在线网络技术（北京）有限公司 | Video detection method, device, equipment and storage medium |
| CN111429638A (en) * | 2020-04-13 | 2020-07-17 | 汪辽宁 | Access control method based on voice recognition and face recognition |
| CN111639689A (en) * | 2020-05-20 | 2020-09-08 | 杭州海康威视系统技术有限公司 | Face data processing method and device and computer readable storage medium |
| CN112084893A (en) * | 2020-08-24 | 2020-12-15 | 中国银联股份有限公司 | Biological recognition terminal abnormity detection method, device, equipment and storage medium |
| CN112200009A (en) * | 2020-09-15 | 2021-01-08 | 青岛邃智信息科技有限公司 | Pedestrian re-identification method based on key point feature alignment in community monitoring scene |
| CN112200009B (en) * | 2020-09-15 | 2023-10-17 | 青岛邃智信息科技有限公司 | Pedestrian re-identification method based on key point feature alignment in community monitoring scene |
| CN112784712A (en) * | 2021-01-08 | 2021-05-11 | 重庆创通联智物联网有限公司 | Missing child early warning implementation method and device based on real-time monitoring |
| CN112784712B (en) * | 2021-01-08 | 2023-08-18 | 重庆创通联智物联网有限公司 | Missing child early warning implementation method and device based on real-time monitoring |
| CN114724074A (en) * | 2022-06-01 | 2022-07-08 | 共道网络科技有限公司 | Method and device for detecting risk video |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021004112A1 (en) | 2021-01-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110472491A (en) | Abnormal face detecting method, abnormality recognition method, device, equipment and medium | |
| CN110662484B (en) | System and method for whole body measurement extraction | |
| US10489683B1 (en) | Methods and systems for automatic generation of massive training data sets from 3D models for training deep learning networks | |
| WO2020199479A1 (en) | Human motion recognition method and device | |
| US10176376B2 (en) | System and method for generating and employing short length iris codes | |
| CN108229330A (en) | Face fusion recognition methods and device, electronic equipment and storage medium | |
| CN103703472A (en) | Semantic parsing of objects in video | |
| CN111428552B (en) | Black eye recognition method and device, computer equipment and storage medium | |
| CN110210457A (en) | Method for detecting human face, device, equipment and computer readable storage medium | |
| US20220044070A1 (en) | Methods and systems for automatic generation of massive training data sets from 3d models for training deep learning networks | |
| CN109299658A (en) | Face area detecting method, face image rendering method, device and storage medium | |
| Lee et al. | Robust iris recognition baseline for the grand challenge | |
| CN113780145A (en) | Sperm morphology detection method, sperm morphology detection device, computer equipment and storage medium | |
| Chen et al. | Slender Flexible Object Segmentation Based on Object Correlation Module and Loss Function Optimization | |
| CN110321935A (en) | Business event relationship determines method and device, computer readable storage medium | |
| CN111428553B (en) | Face pigment spot recognition method and device, computer equipment and storage medium | |
| Zappa et al. | Uncertainty of 3D facial features measurements and its effects on personal identification | |
| CN112101185A (en) | Method for training wrinkle detection model, electronic device and storage medium | |
| Paul et al. | Automatic adaptive facial feature extraction using CDF analysis | |
| Rabba et al. | Discriminative robust gaze estimation using kernel-dmcca fusion | |
| CN115601817A (en) | Face recognition method and device, processor and electronic equipment | |
| Hipparagi et al. | Artist Based Video Recommendation System using Machine Learning | |
| Wang et al. | Positioning corners of human mouth based on local gradient operator | |
| Lejbølle et al. | Constraint patch matching for faster person re-identification | |
| CN115082605A (en) | Face image generation and face recognition model robustness evaluation method and device |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191119 |
|
| WD01 | Invention patent application deemed withdrawn after publication |