CN110458102A - A kind of facial image recognition method and device, electronic equipment and storage medium - Google Patents
A kind of facial image recognition method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN110458102A CN110458102A CN201910739381.8A CN201910739381A CN110458102A CN 110458102 A CN110458102 A CN 110458102A CN 201910739381 A CN201910739381 A CN 201910739381A CN 110458102 A CN110458102 A CN 110458102A
- Authority
- CN
- China
- Prior art keywords
- face
- image data
- feature
- facial image
- face image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
This disclosure relates to a kind of facial image recognition method and device, electronic equipment and storage medium, wherein this method comprises: extracting the image data to be processed for belonging to different faces from face image data；According to the image data to be processed for belonging to different faces, the face image data pair of non-matching is obtained；Wherein, the face image data of the non-matching is to for characterizing the feature for belonging to two facial images of different faces；According to the face image data pair of the non-matching, facial image identification network is trained, the target identification network of facial image for identification is obtained.Using the disclosure, recognition efficiency and accuracy rate to facial image can be improved.
Description
Technical field
This disclosure relates to technical field of computer vision more particularly to a kind of facial image recognition method and device, electronics
Equipment and storage medium.
Background technique
Face recognition application based on deep learning, it is very universal at present.The recognition of face mould obtained by deep learning
The performance of type trains data type used closely bound up with it, in order to improve recognition efficiency and the accuracy of recognition of face, needs
Human face recognition model is improved or improve the validity of the training data for the model training (validity, which refers to, to be had
Help the improvement of human face recognition model, and can help to excavate more to the helpful information of model training), however, related
There is no effective solution in technology.
Summary of the invention
The present disclosure proposes a kind of facial image identification technology schemes.
According to the one side of the disclosure, a kind of facial image recognition method is provided, which comprises
The image data to be processed for belonging to different faces is extracted from face image data；
According to the image data to be processed for belonging to different faces, the face image data pair of non-matching is obtained；Wherein,
The face image data of the non-matching is to for characterizing the feature for belonging to two facial images of different faces；
According to the face image data pair of the non-matching, facial image identification network is trained, is obtained for knowing
The target identification network of others' face image.
Using the disclosure, belong to the facial image feature two-by-two of different faces, due to that can be directed to form non-matching
Face image data pair, so as to obtain belonging to similar in different faces image but feature facial image feature two-by-two, because
This is trained facial image identification network, obtains face figure for identification according to the face image data pair of the non-matching
The target identification network of picture, can be more perfect compared to facial image identification network before, can use target identification network
When identifying to facial image to be identified, the recognition efficiency and accuracy rate to facial image are improved.Possible implementation
In, it is described that the image data to be processed for belonging to different faces is extracted from face image data, comprising:
Network is identified according to the facial image, extracts the feature of facial image in the face image data；
The feature of different faces image will be belonged to as the image data to be processed.
Using the disclosure, network can be identified according to the facial image, extract face figure in the face image data
The feature of picture, due to needing for the facial image feature two-by-two for belonging to different faces, to form the facial image number of non-matching
According to right, therefore, the feature of different faces image will be belonged to as the image data to be processed.
In possible implementation, the image data to be processed that different faces are belonged to according to obtains non-matching
Face image data pair, comprising:
The feature for belonging to different faces includes at least second in fisrt feature and the second face in the first face
Feature；
It, will be described in the case where meeting preset condition according to the similarity that the fisrt feature and the second feature obtain
First face and second face are configured to the face image data pair.
Using the disclosure, the feature due to belonging to different faces has very much, needs to filter out and belongs to different faces but more
Similar feature so forms the face image data pair of non-matching according to the more similar feature, is just conducive to people
Face image identifies that the training of network therefore can be according at least to the fisrt feature of the first face and the second feature of the second face
In the case that obtained similarity meets preset condition, first face and second face are configured to the face figure
As data pair.
In possible implementation, it is described to facial image identification network be trained before, the method also includes:
According to the feature correlation between the face image data pair, sampling order is obtained.
It is needed before being trained to facial image identification network according to the face image data pair using the disclosure
Between feature correlation, obtain sampling order, to extract sample data from training sample according to the sampling order, thus
Be conducive to the training of facial image identification network.If not considering sampling order, such as stochastical sampling, facial image certainly will be reduced
Identify the training effect of network.
In possible implementation, the feature correlation according between the face image data pair is sampled
Sequentially, comprising:
According to the feature between the face image data pair, characteristic set is obtained；
According to the characteristic set construction feature tree KD-Tree, feature correlation between face image data pair is close
Adjacent node of the feature as the KD-Tree；
Traverse path that the KD-Tree is obtained will be traversed as the sampling order.
Using the disclosure, make using KD-Tree, and according to the close feature of the feature correlation between face image data pair
For the adjacent node of KD-Tree, then, it, can be suitable as sampling using the traverse path after traversal KD-Tree obtains traverse path
Sequence extracts sample data according to the sampling order from training sample, is conducive to the training of facial image identification network.It is possible
In implementation, it is described obtain sampling order after, the method also includes:
The face image data pair that will be read according to the sampling order identifies network as the facial image is inputted
Training sample.
Using the disclosure, sample data is extracted from training sample according to the sampling order, is conducive to facial image identification
The training of network.
In possible implementation, the face image data pair, at least from the first face for face training
The second face image set that image collection and acquisition face obtain, and the face in two face image sets is not identical.
Using the disclosure, face image data can be obtained to from preparatory ready-portioned two face image sets,
Face in two face image sets be it is not identical, so as to avoid different faces are screened from a face image set
Processing cost can faster obtain the sample data " face image data of non-matching for training facial image identification network
It is right ".
It is described that facial image identification network is trained in possible implementation, comprising:
In each iteration being trained to facial image identification network, sample characteristics are saved；
The sample characteristics include the feature extracted from first face image set, obtain sample through successive ignition
Feature set.
It will be from the first face image set during being trained to facial image identification network using the disclosure
The feature of extraction is as sample characteristics and saves, can more keeping characteristics, thus for next time to facial image identify net
The more fixed reference features of the repetitive exercise of network are conducive to the training of facial image identification network.It is described right in possible implementation
Facial image identification network is trained, further includes:
According to the current face's feature extracted from second face image set in each iteration and last iteration
All sample characteristics of sample characteristics concentration are obtained, loss function is calculated；
The facial image identification network is trained according to the backpropagation of the loss function.
It, in each iteration, can be according to the current face's feature extracted in the second face image set using the disclosure
The sample characteristics saved with last iteration calculate loss function, thus according to the backpropagation of the loss function come training of human
Face image identifies network, and obtained target identification network can be more perfect compared to facial image identification network before, Ke Yi
When being identified using the facial image that target identification network handles identify, improve to the recognition efficiency of facial image and accurate
Rate.
According to the one side of the disclosure, a kind of facial image identification device is provided, described device includes:
Extraction unit, for extracting the image data to be processed for belonging to different faces from face image data；
First processing units obtain the people of non-matching for belonging to the image data to be processed of different faces according to
Face image data pair；Wherein, the face image data of the non-matching is to for characterizing two face figures for belonging to different faces
The feature of picture；
The second processing unit, for the face image data pair according to the non-matching, to facial image identify network into
Row training, obtains the target identification network of facial image for identification.
In possible implementation, the extraction unit is further used for:
Network is identified according to the facial image, extracts the feature of facial image in the face image data；
The feature of different faces image will be belonged to as the image data to be processed.
In possible implementation, the first processing units are further used for:
The feature for belonging to different faces includes at least second in fisrt feature and the second face in the first face
Feature；
It, will be described in the case where meeting preset condition according to the similarity that the fisrt feature and the second feature obtain
First face and second face are configured to the face image data pair.
In possible implementation, described device further includes third processing unit, is used for:
According to the feature correlation between the face image data pair, sampling order is obtained.
In possible implementation, the third processing unit is further used for:
According to the feature between the face image data pair, characteristic set is obtained；
According to the characteristic set construction feature tree KD-Tree, feature correlation between face image data pair is close
Adjacent node of the feature as the KD-Tree；
Traverse path that the KD-Tree is obtained will be traversed as the sampling order.
In possible implementation, described the second processing unit is further used for:
The face image data pair that will be read according to the sampling order identifies network as the facial image is inputted
Training sample.
In possible implementation, the face image data pair, at least from the first face for face training
The second face image set that image collection and acquisition face obtain, and the face in two face image sets is not identical.
In possible implementation, described the second processing unit is further used for:
In each iteration being trained to facial image identification network, sample characteristics are saved；
The sample characteristics include the feature extracted from first face image set, obtain sample through successive ignition
Feature set.
In possible implementation, described the second processing unit is further used for:
According to the current face's feature extracted from second face image set in each iteration and last iteration
All sample characteristics of sample characteristics concentration are obtained, loss function is calculated；
The facial image identification network is trained according to the backpropagation of the loss function.
According to the one side of the disclosure, a kind of electronic equipment is provided, comprising:
Processor；
Memory for storage processor executable instruction；
Wherein, the processor is configured to: execute above-mentioned facial image recognition method.
According to the one side of the disclosure, a kind of computer readable storage medium is provided, computer program is stored thereon with
Instruction, the computer program instructions realize above-mentioned facial image recognition method when being executed by processor.
In the embodiments of the present disclosure, the image data to be processed for belonging to different faces is extracted from face image data；Root
According to the image data to be processed for belonging to different faces, the face image data pair of non-matching is obtained；Wherein, the non-matching
Face image data to for characterizing the feature for belonging to two facial images of different faces；According to the face of the non-matching
Image data pair is trained facial image identification network, obtains the target identification network of facial image for identification.Using
The disclosure belongs to the facial image feature two-by-two of different faces due to that can be directed to, to form the face image data of non-matching
It is right, so as to obtain belonging to similar in different faces image but feature facial image feature two-by-two, therefore, non-matched according to this
Pair face image data pair, to facial image identification network be trained, obtain the target identification of facial image for identification
Network, can be more perfect compared to facial image identification network before, the subsequent face figure in the identification of target identification network handles
As the recognition efficiency and accuracy rate to facial image can be improved when being identified.
It should be understood that above general description and following detailed description is only exemplary and explanatory, rather than
Limit the disclosure.
According to below with reference to the accompanying drawings to detailed description of illustrative embodiments, the other feature and aspect of the disclosure will become
It is clear.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and those figures show meet this public affairs
The embodiment opened, and together with specification it is used to illustrate the technical solution of the disclosure.
Fig. 1 shows the flow chart of the facial image recognition method according to the embodiment of the present disclosure.
Fig. 2 shows the flow charts according to the facial image recognition method of the embodiment of the present disclosure.
Fig. 3, which is shown, identifies network training process flow chart according to the facial image of the embodiment of the present disclosure.
Fig. 4, which is shown, identifies network training process flow chart according to the facial image of the embodiment of the present disclosure.
Fig. 5 shows the block diagram of the facial image identification device according to the embodiment of the present disclosure.
Fig. 6 shows the block diagram of the electronic equipment according to the embodiment of the present disclosure.
Fig. 7 shows the block diagram of the electronic equipment according to the embodiment of the present disclosure.
Specific embodiment
Various exemplary embodiments, feature and the aspect of the disclosure are described in detail below with reference to attached drawing.It is identical in attached drawing
Appended drawing reference indicate element functionally identical or similar.Although the various aspects of embodiment are shown in the attached drawings, remove
It non-specifically points out, it is not necessary to attached drawing drawn to scale.
Dedicated word " exemplary " means " being used as example, embodiment or illustrative " herein.Here as " exemplary "
Illustrated any embodiment should not necessarily be construed as preferred or advantageous over other embodiments.
The terms "and/or", only a kind of incidence relation for describing affiliated partner, indicates that there may be three kinds of passes
System, for example, A and/or B, can indicate: individualism A exists simultaneously A and B, these three situations of individualism B.In addition, herein
Middle term "at least one" indicate a variety of in any one or more at least two any combination, it may for example comprise A,
B, at least one of C can indicate to include any one or more elements selected from the set that A, B and C are constituted.
In addition, giving numerous details in specific embodiment below to better illustrate the disclosure.
It will be appreciated by those skilled in the art that without certain details, the disclosure equally be can be implemented.In some instances, for
Method, means, element and circuit well known to those skilled in the art are not described in detail, in order to highlight the purport of the disclosure.
With the development of deep learning, face recognition technology is all applied in many application scenarios, is especially being pacified
It is even more an indispensable module in anti-monitoring.The performance of the recognition of face network obtained by deep learning trains institute with it
Data type is closely bound up, in order to preferably be known under a certain scene (such as video analysis, security protection face monitor)
The performances such as other treatment effeciency and accuracy can do training to recognition of face network by acquiring the human face data under the scene
(such as incremental training).The incremental training refers to: being trained according to new training sample, constantly from new training sample
The new knowledge of middle study, and most of historical knowledge learnt in the past can be saved, for example obtained according to same face
Pairing face image data is trained obtained historical record two-by-two, and the disclosure is increased on this basis according to different people
The process that the face image data of non-matching two-by-two that face obtains is trained.
It should be pointed out that the facial image construction nothing of acquisition can be used in order to not introduce noise in incremental training
Label data does unsupervised training.In the training process, facial image is delivered into recognition of face net in the way of " pairs of "
Network is trained.Since this unsupervised training method is only constrained between pairs of face, even if there is multipair face figure
As being delivered into recognition of face network, due to not constrained between different pairs of facial image, therefore, it is impossible to excavate more
Multipair training helpful effective information of recognition of face network, so as to cause the training descendant obtained using this training method
Face identifies that the treatment effeciency of network (such as the target identification network of facial image for identification) is more inefficient, and accuracy of identification is not yet
It is high.
In the disclosure, can using belonging to the image datas to be processed of different faces, according to belong to different faces wait locate
Image data is managed, the face image data pair of non-matching is obtained, thus, it is carried out according to the face image data of the non-matching above-mentioned
Incremental training, due to being constrained between different pairs of facial image, can excavate more to training face knowledge
The other helpful effective information of network causes recognition of face network after the training obtained using disclosure training method (to be such as used for
Identify facial image target identification network) treatment effeciency it is more efficient, improve accuracy of identification.
Fig. 1 shows the flow chart of the facial image recognition method according to the embodiment of the present disclosure, the facial image recognition method
Applied to facial image identification device, for example, facial image identification device can be by terminal device or server or other processing
Equipment executes, wherein terminal device can be user equipment (UE, User Equipment), mobile device, cellular phone, nothing
Rope phone, handheld device, calculates equipment, vehicle-mounted sets personal digital assistant (PDA, Personal Digital Assistant)
Standby, wearable device etc..In some possible implementations, which can be called by processor and be deposited
The mode of the computer-readable instruction stored in reservoir is realized.As shown in Figure 1, the process includes:
Step S101, the image data to be processed for belonging to different faces is extracted from face image data.
In one example, face image data is obtained, face image data is the image data of multiple and different faces.According to people
Face image identification network extracts the feature of facial image in the face image data, for example, can identify net using facial image
Feature extraction functions module in network carries out feature extraction to the feature of facial image in the face image data.It will belong to not
As the image data to be processed, which is made of feature with facial image multiple face characteristics, including
Multiple face characteristics of same face and multiple face characteristics of different faces.
Step S102, the image data to be processed for belonging to different faces according to obtains the facial image number of non-matching
According to right；Wherein, the face image data of the non-matching is to for characterizing the feature for belonging to two facial images of different faces.
In possible implementation, belongs to the image data to be processed of different faces, can be to multiple and different faces
Image data carries out obtained multiple features after feature extraction, calculates in multiple feature the similarity between feature two-by-two,
If the similarity between feature meets preset condition two-by-two, inquiry has the corresponding people of the institute of feature two-by-two of similarity
Face image constructs the face image data pair according to the facial image inquired, and the face image data is to (such as non-matching people
Face image to) be referred to as " pairs of " without label data, i.e., in subsequent training process by the non-matching facial image make
For no label data, and input facial image identifies network in couples, and with training, the facial image identifies network.
In one example, the different characteristic from different faces image is distinguished with " first ", " second " this reference.On
The feature for belonging to different faces is stated including at least the second feature in the fisrt feature and the second face in the first face, according to institute
It states in the case that the similarity that fisrt feature and the second feature obtain meets preset condition, by first face and described
Second face is configured to the face image data pair.
Step S103, according to the face image data pair of the non-matching, facial image identification network is trained, is obtained
To the target identification network of facial image for identification.
In one example, by multiple face image datas to as no label data, and in couples, input facial image is identified
Network, with training, the facial image identifies network.
Using the disclosure, S101- step S102, is available for facial image identification network instruction through the above steps
Experienced training sample, it may be assumed that multiple face image datas are to (such as non-matching facial image to), wherein non-matching facial image pair
Refer to: two facial images are not belonging to the same person.S103 through the above steps can use different between can face image data
The binding character (or correlation) that can be generated, the face image data for obtaining non-matching more effectively train the facial image to rear
Identify network.It in practical applications, can be according to the target for example, in intelligent video analysis or security protection face monitoring scene
The facial image of identification network handles identification is identified, recognition result is obtained.Due to the face image data according to non-matching
To can more effectively train the facial image identification network and its network parameter is carried out it is perfect, therefore, pass through training the people
After face image identification network obtains the target identification network of facial image for identification, image is carried out according to the target identification network
Identification, identifying processing effect is higher, and improves accuracy of identification.
Fig. 2 shows the flow chart according to the facial image recognition method of the embodiment of the present disclosure, the facial image recognition methods
Applied to facial image identification device, for example, facial image identification device can be by terminal device or server or other processing
Equipment executes, wherein terminal device can be user equipment (UE, User Equipment), mobile device, cellular phone, nothing
Rope phone, handheld device, calculates equipment, vehicle-mounted sets personal digital assistant (PDA, Personal Digital Assistant)
Standby, wearable device etc..In some possible implementations, which can be called by processor and be deposited
The mode of the computer-readable instruction stored in reservoir is realized.As shown in Fig. 2, the process includes:
Step S201, the image data to be processed for belonging to different faces is extracted from face image data.
In one example, face image data is obtained, face image data is the image data of multiple and different faces.According to people
Face image identification network extracts the feature of facial image in the face image data, for example, can identify net using facial image
Feature extraction functions module in network carries out feature extraction to the feature of facial image in the face image data.It will belong to not
As the image data to be processed, which is made of feature with facial image multiple face characteristics, including
Multiple face characteristics of same face and multiple face characteristics of different faces.
Step S202, the image data to be processed for belonging to different faces according to obtains the facial image number of non-matching
According to right；Wherein, the face image data of the non-matching is to for characterizing the feature for belonging to two facial images of different faces.
In possible implementation, belongs to the image data to be processed of different faces, can be to multiple and different faces
Image data carries out obtained multiple features after feature extraction, calculates in multiple feature the similarity between feature two-by-two,
If the similarity between feature meets preset condition two-by-two, inquiry has the corresponding people of the institute of feature two-by-two of similarity
Face image constructs the face image data pair according to the facial image inquired, and the face image data is to (such as non-matching people
Face image to) be referred to as " pairs of " without label data, i.e., in subsequent training process by the non-matching facial image make
For no label data, and input facial image identifies network in couples, and with training, the facial image identifies network.
In one example, the different characteristic from different faces image is distinguished with " first ", " second " this reference.On
The feature for belonging to different faces is stated including at least the second feature in the fisrt feature and the second face in the first face, according to institute
It states in the case that the similarity that fisrt feature and the second feature obtain meets preset condition, by first face and described
Second face is configured to the face image data pair.
Step S203, according to the feature correlation between the face image data pair, sampling order is obtained.
In one example, before incremental training, it can determine that the sampling of face picture is suitable according to the correlation of face characteristic
Sequence, for example, obtaining characteristic set according to the feature between the face image data pair.It is constructed according to the characteristic set special
Sign sets KD-Tree, adjacent node of the close feature of the feature correlation between face image data pair as the KD-Tree.
Traverse path that the KD-Tree is obtained will be traversed as the sampling order.According to the correlation calculations of facial image feature
To the sampling order of facial image, the facial image of adjacent reading can be made to have biggish correlation, that is to say, that according to
The sampling order reads facial image, is compared to random reading to facial image, available difference is to facial image
More restrictive (or the correlations) generated between data.And more restrictive (or correlations) can be trained more effectively and be somebody's turn to do
Facial image identification network simultaneously carries out its network parameter perfect.It, can be in binding characteristic memory module in subsequent example
The sample characteristics of preservation further increase effective training to facial image identification network, improve facial image identification net
The training effectiveness of network and accuracy.
Step S204, the face image data pair that will be read according to sampling order is identified as the facial image is inputted
The training sample of network.
In one example, the face image data pair, at least from for face training the first face image set and
Acquire obtained the second face image set of face under true environment, and the face in two face image sets be it is not identical,
Set A can be denoted as in subsequent applications example by extracting the characteristic set that the feature of the first facial image obtains, and extract the second people
The characteristic set that the feature of face image obtains can be denoted as set B in subsequent applications example, be not repeated herein.
Step S205, facial image identification network is trained according to the training sample, obtains face for identification
The target identification network of image.
In one example, by multiple face image datas to as no label data, and in couples, input facial image is identified
Network, with training, the facial image identifies network.
Using the disclosure, S201- step S204, is available for facial image identification network instruction through the above steps
Experienced training sample, it may be assumed that multiple face image datas are to (such as non-matching facial image to), wherein non-matching facial image pair
Refer to: two facial images are not belonging to the same person.S205 through the above steps can use different between can face image data
The binding character (or correlation) that can be generated, obtains the face image data of non-matching to the training sample constituted, later, can
More effectively to train the facial image to identify network according to the training sample.In practical applications, for example, in intelligent video point
In analysis or security protection face monitoring scene, it can be identified, be obtained according to the facial image that the target identification network handles identify
Recognition result.Due to according to the training sample can more effectively train the facial image identification network and to its network parameter into
Row is perfect, therefore, after obtaining the target identification network of facial image for identification by training facial image identification network, root
Image recognition is carried out according to the target identification network, identifying processing effect is higher, and improves accuracy of identification.
In possible implementation, facial image identification network is trained, comprising: net is identified to the facial image
In each iteration that network is trained, sample characteristics are saved.The sample characteristics include from first face image set
The feature of extraction obtains sample characteristics collection through successive ignition.Sample characteristics can be denoted as F in subsequent applications exampleA, FAIt can be with
It is stored in feature memory module, sample characteristics collection can be denoted as F in subsequent applications exampleM, FAThe collection constituted is combined into FM.,
It is not repeated herein.
It is described that facial image identification network is trained, further includes: according in each iteration in possible implementation
The current face's feature and last iteration extracted from second face image set obtain all of sample characteristics concentration
Sample characteristics calculate loss function.The facial image identification network is trained according to the backpropagation of the loss function.It can
To understand are as follows: the face characteristic of iteration retains each time face characteristic and last iteration calculates loss function, i.e., with each
The face characteristic that secondary iteration retains and the face characteristic of last iteration are constrained, to obtain more constraint informations.And these
Constraint information is referred to as " effective information " due to can more effectively train the facial image to identify network.If instructing
Practice during only use current iteration the feature calculation loss function of facial image two-by-two, compared to the disclosure implementation and
Speech cannot get more effective informations, and the disclosure can be by identifying that increasing feature memory module in network (uses in facial image
In the preservation sample characteristics), in the training process, by the sample characteristics in the face characteristic of current iteration and feature memory module
Loss function is calculated together, then more effective informations can be provided, thus, it can use in the training process more effective
Information more effectively trains the facial image to identify network, improves training effectiveness.
Using example:
Fig. 3, which is shown, identifies network training process flow chart according to the facial image of the embodiment of the present disclosure, as shown in figure 3, packet
Contain:
Step S301, feature is extracted respectively to collected different faces image, construction is by non-matching facial image to structure
At training sample, the image in training sample is properly termed as training image.
Step S302, according to the feature of non-matching facial image pair, the training image in training sample is calculated in training
Sampling order.
Step S303, according to the sampling order calculated, the training image in training sample is read, and binding characteristic is remembered
Sample characteristics in module train facial image to identify network together.
Fig. 4, which is shown, identifies network training process flow chart according to the facial image of the embodiment of the present disclosure, is based on Fig. 3-Fig. 4 institute
Show, related specific implementation be described as follows:
One, feature is extracted respectively to collected different faces image, construction is by non-matching facial image to the instruction constituted
Practice sample.
Input: from practical application scene acquire facial image, the original face training image of system, and two set
Face need to ensure that there is no identical face；
Output: facial image feature, non-matching facial image pair.
Specific implementation includes: to do face alignment to the facial image of input；Using current face's identification model to right
Facial image after neat extracts feature, obtains face recognition features, and the facial image feature acquired from practical application scene is denoted as
Set A, the original facial image feature of system are denoted as set B；More than the feature calculation two-by-two in characteristic set B and characteristic set A
String similarity, and to obtained cosine similarity set according to sorting from large to small, taking preceding 10%, (percentage is not unique, can
It adjusts according to the actual situation, percentage is bigger, the mesh trained bigger to the training difficulty of facial image identification network
The performance for identifying other network is also better) cosine similarity corresponding image combination be used as the non-matching facial image pair, and will
An optimization aim threshold value of the cosine similarity of critical point as subsequent facial image identification network training
(threshold)。
Two, according to the feature of non-matching facial image pair, sampling of the training image in training in training sample is calculated
Sequentially.
Input: the information of characteristic set A, non-matching facial image pair；
Output: facial image identifies image sampling sequence when network training.
Specific implementation include: according to the information of the non-matching facial image pair, construction feature set C=A1,
A2 ..., An }, the element in set C is the feature of the selected original face training image of system of training；Use characteristic set C
KD-Tree is established, KD-Tree is traversed, image sampling sequence when traverse path is then trained.
Three, according to the sampling order calculated, the training image in training sample is read, and in binding characteristic memory module
Sample characteristics together train facial image identify network.
Input: current face's image recognition network, the non-matching facial image pair, image sampling sequence；
Output: obtaining target identification network after training, i.e., new facial image identifies network.
Specific implementation includes: to initialize the facial image using the network parameter of current face's image recognition network
Identify network；The non-matching facial image pair is read according to the sampling order calculated, for iteration each time, this of reading is non-
Facial image is matched to including at least two parts: IAAnd IB, IAFrom the original face training image of system, IBFrom adopting
The facial image of collection.Image IAAnd IBBy the calculating of facial image identification network, feature F is obtainedAAnd FB, then by FAIt is saved in
Feature memory module.By FBWith all characteristic set F in feature memory moduleMLoss function is calculated, and updates facial image knowledge
The network parameter of other network.The formula for calculating loss function can be as shown in formula (1), wherein L is loss function；N, M distinguishes
For the various corresponding quantity summations of different characteristic；FMFor sample characteristics FAThe sample feature set of composition；FBFor image IBThrough remarkable
Face image identifies feature obtained by the calculating of network；Threshold is according to special two-by-two in characteristic set B and characteristic set A
Sign calculates the cosine similarity of the critical point obtained when cosine similarity, i.e. one as facial image identification network training
Optimization aim threshold value.
It is noted that the sample characteristics in feature memory module there are timeliness, need periodically to delete, to realize
Update to the sample characteristics in feature memory module.For example, the sample characteristics in feature memory module are more than 100 there are the time
(numerical value is not unique, can be adjusted according to hands-on effect) secondary iteration, then by the sample characteristics from feature memory module
It removes, carries out the iterative processing always until meeting preset the number of iterations.
It will be understood by those skilled in the art that each step writes sequence simultaneously in the above method of specific embodiment
It does not mean that stringent execution sequence and any restriction is constituted to implementation process, the specific execution sequence of each step should be with its function
It can be determined with possible internal logic.
Above-mentioned each embodiment of the method that the disclosure refers to can phase each other without prejudice to principle logic
The embodiment formed after combining is mutually combined, as space is limited, the disclosure repeats no more.
In addition, the disclosure additionally provides facial image identification device, electronic equipment, computer readable storage medium, program,
The above-mentioned any facial image recognition method that can be used to realize disclosure offer, corresponding technical solution is with description and referring to side
The corresponding record of method part, repeats no more.
Fig. 5 shows the block diagram of the facial image identification device according to the embodiment of the present disclosure, as shown in figure 5, the disclosure is implemented
Example facial image identification device include: extraction unit 31, for from face image data extract belong to different faces to
Handle image data；First processing units 32 obtain non-match for belonging to the image data to be processed of different faces according to
Pair face image data pair；Wherein, the face image data of the non-matching belongs to two of different faces to for characterizing
The feature of facial image；The second processing unit 33 knows facial image for the face image data pair according to the non-matching
Other network is trained, and obtains the target identification network of facial image for identification.
It can also include recognition unit, for being identified according to the target identification network handles in possible implementation
Facial image identified, obtain recognition result.
In possible implementation, the extraction unit is further used for: identifying network according to the facial image, mentions
Take the feature of facial image in the face image data；The feature of different faces image will be belonged to as the image to be processed
Data.
In possible implementation, the first processing units are further used for: the feature for belonging to different faces is extremely
It less include the second feature in the fisrt feature and the second face in the first face；According to the fisrt feature and second spy
In the case that the similarity obtained meets preset condition, first face and second face are configured to the face
Image data pair.
In possible implementation, described device further includes third processing unit, is used for: according to the face image data
Feature correlation between, obtains sampling order.
In possible implementation, the third processing unit is further used for: according to the face image data to it
Between feature, obtain characteristic set；According to the characteristic set construction feature tree KD-Tree, between face image data pair
Adjacent node of the close feature of feature correlation as the KD-Tree；The traverse path that the KD-Tree is obtained will be traversed to make
For the sampling order.
In possible implementation, described the second processing unit is further used for: by what is read according to the sampling order
Face image data pair, as the training sample for inputting the facial image identification network.
In possible implementation, the face image data pair, at least from the first face for face training
The second face image set that face obtains, and the face in two face image sets are acquired under image collection and true environment
It is not identical.
In possible implementation, described the second processing unit is further used for: to the facial image identify network into
In each iteration of row training, sample characteristics are saved；The sample characteristics include extracting from first face image set
Feature, obtain sample characteristics collection through successive ignition.
In possible implementation, described the second processing unit is further used for: according in each iteration from described second
The current face's feature and last iteration extracted in face image set obtain all sample characteristics of sample characteristics concentration, meter
Calculate loss function；The facial image identification network is trained according to the backpropagation of the loss function.
In some embodiments, the embodiment of the present disclosure provides the function that has of device or comprising module can be used for holding
The method of row embodiment of the method description above, specific implementation are referred to the description of embodiment of the method above, for sake of simplicity, this
In repeat no more.
The embodiment of the present disclosure also proposes a kind of computer readable storage medium, is stored thereon with computer program instructions, institute
It states when computer program instructions are executed by processor and realizes the above method.Computer readable storage medium can be non-volatile meter
Calculation machine readable storage medium storing program for executing.
The embodiment of the present disclosure also proposes a kind of electronic equipment, comprising: processor；For storage processor executable instruction
Memory；Wherein, the processor is configured to the above method.
The equipment that electronic equipment may be provided as terminal, server or other forms.
Fig. 6 is the block diagram of a kind of electronic equipment 800 shown according to an exemplary embodiment.For example, electronic equipment 800 can
To be mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, Medical Devices are good for
Body equipment, the terminals such as personal digital assistant.
Referring to Fig. 6, electronic equipment 800 may include following one or more components: processing component 802, memory 804,
Power supply module 806, multimedia component 808, audio component 810, the interface 812 of input/output (I/O), sensor module 814,
And communication component 816.
The integrated operation of the usual controlling electronic devices 800 of processing component 802, such as with display, call, data are logical
Letter, camera operation and record operate associated operation.Processing component 802 may include one or more processors 820 to hold
Row instruction, to perform all or part of the steps of the methods described above.In addition, processing component 802 may include one or more moulds
Block, convenient for the interaction between processing component 802 and other assemblies.For example, processing component 802 may include multi-media module, with
Facilitate the interaction between multimedia component 808 and processing component 802.
Memory 804 is configured as storing various types of data to support the operation in electronic equipment 800.These data
Example include any application or method for being operated on electronic equipment 800 instruction, contact data, telephone directory
Data, message, image, video etc..Memory 804 can by any kind of volatibility or non-volatile memory device or it
Combination realize, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM) is erasable
Except programmable read only memory (EPROM), programmable read only memory (PROM), read-only memory (ROM), magnetic memory, fastly
Flash memory, disk or CD.
Power supply module 806 provides electric power for the various assemblies of electronic equipment 800.Power supply module 806 may include power supply pipe
Reason system, one or more power supplys and other with for electronic equipment 800 generate, manage, and distribute the associated component of electric power.
Multimedia component 808 includes the screen of one output interface of offer between the electronic equipment 800 and user.
In some embodiments, screen may include liquid crystal display (LCD) and touch panel (TP).If screen includes touch surface
Plate, screen may be implemented as touch screen, to receive input signal from the user.Touch panel includes one or more touches
Sensor is to sense the gesture on touch, slide, and touch panel.The touch sensor can not only sense touch or sliding
The boundary of movement, but also detect duration and pressure associated with the touch or slide operation.In some embodiments,
Multimedia component 808 includes a front camera and/or rear camera.When electronic equipment 800 is in operation mode, as clapped
When taking the photograph mode or video mode, front camera and/or rear camera can receive external multi-medium data.It is each preposition
Camera and rear camera can be a fixed optical lens system or have focusing and optical zoom capabilities.
Audio component 810 is configured as output and/or input audio signal.For example, audio component 810 includes a Mike
Wind (MIC), when electronic equipment 800 is in operation mode, when such as call mode, recording mode, and voice recognition mode, microphone
It is configured as receiving external audio signal.The received audio signal can be further stored in memory 804 or via logical
Believe that component 816 is sent.In some embodiments, audio component 810 further includes a loudspeaker, is used for output audio signal.
I/O interface 812 provides interface between processing component 802 and peripheral interface module, and above-mentioned peripheral interface module can
To be keyboard, click wheel, button etc..These buttons may include, but are not limited to: home button, volume button, start button and lock
Determine button.
Sensor module 814 includes one or more sensors, for providing the state of various aspects for electronic equipment 800
Assessment.For example, sensor module 814 can detecte the state that opens/closes of electronic equipment 800, the relative positioning of component, example
As the component be electronic equipment 800 display and keypad, sensor module 814 can also detect electronic equipment 800 or
The position change of 800 1 components of electronic equipment, the existence or non-existence that user contacts with electronic equipment 800, electronic equipment 800
The temperature change of orientation or acceleration/deceleration and electronic equipment 800.Sensor module 814 may include proximity sensor, be configured
For detecting the presence of nearby objects without any physical contact.Sensor module 814 can also include optical sensor,
Such as CMOS or ccd image sensor, for being used in imaging applications.In some embodiments, which may be used also
To include acceleration transducer, gyro sensor, Magnetic Sensor, pressure sensor or temperature sensor.
Communication component 816 is configured to facilitate the communication of wired or wireless way between electronic equipment 800 and other equipment.
Electronic equipment 800 can access the wireless network based on communication standard, such as WiFi, 2G or 3G or their combination.Show at one
In example property embodiment, communication component 816 receives broadcast singal or broadcast from external broadcasting management system via broadcast channel
Relevant information.In one exemplary embodiment, the communication component 816 further includes near-field communication (NFC) module, short to promote
Cheng Tongxin.For example, radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra wide band can be based in NFC module
(UWB) technology, bluetooth (BT) technology and other technologies are realized.
In the exemplary embodiment, electronic equipment 800 can be by one or more application specific integrated circuit (ASIC), number
Word signal processor (DSP), digital signal processing appts (DSPD), programmable logic device (PLD), field programmable gate array
(FPGA), controller, microcontroller, microprocessor or other electronic components are realized, for executing the above method.
In the exemplary embodiment, a kind of non-volatile computer readable storage medium storing program for executing is additionally provided, for example including calculating
The memory 804 of machine program instruction, above-mentioned computer program instructions can be executed by the processor 820 of electronic equipment 800 to complete
The above method.
Fig. 7 is the block diagram of a kind of electronic equipment 900 shown according to an exemplary embodiment.For example, electronic equipment 900 can
To be provided as a server.Referring to Fig. 7, it further comprises one or more that electronic equipment 900, which includes processing component 922,
Processor, and the memory resource as representated by memory 932, for store can by the instruction of the execution of processing component 922,
Such as application program.The application program stored in memory 932 may include it is one or more each correspond to one
The module of group instruction.In addition, processing component 922 is configured as executing instruction, to execute the above method.
Electronic equipment 900 can also include that a power supply module 926 is configured as executing the power supply pipe of electronic equipment 900
Reason, a wired or wireless network interface 950 are configured as electronic equipment 900 being connected to network and an input and output (I/
O) interface 958.Electronic equipment 900 can be operated based on the operating system for being stored in memory 932, such as Windows
ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM or similar.
In the exemplary embodiment, a kind of non-volatile computer readable storage medium storing program for executing is additionally provided, for example including calculating
The memory 932 of machine program instruction, above-mentioned computer program instructions can be executed by the processing component 922 of electronic equipment 900 with complete
At the above method.
The disclosure can be system, method and/or computer program product.Computer program product may include computer
Readable storage medium storing program for executing, containing for making processor realize the computer-readable program instructions of various aspects of the disclosure.
Computer readable storage medium, which can be, can keep and store the tangible of the instruction used by instruction execution equipment
Equipment.Computer readable storage medium for example can be-- but it is not limited to-- storage device electric, magnetic storage apparatus, optical storage
Equipment, electric magnetic storage apparatus, semiconductor memory apparatus or above-mentioned any appropriate combination.Computer readable storage medium
More specific example (non exhaustive list) includes: portable computer diskette, hard disk, random access memory (RAM), read-only deposits
It is reservoir (ROM), erasable programmable read only memory (EPROM or flash memory), static random access memory (SRAM), portable
Compact disk read-only memory (CD-ROM), digital versatile disc (DVD), memory stick, floppy disk, mechanical coding equipment, for example thereon
It is stored with punch card or groove internal projection structure and the above-mentioned any appropriate combination of instruction.Calculating used herein above
Machine readable storage medium storing program for executing is not interpreted that instantaneous signal itself, the electromagnetic wave of such as radio wave or other Free propagations lead to
It crosses the electromagnetic wave (for example, the light pulse for passing through fiber optic cables) of waveguide or the propagation of other transmission mediums or is transmitted by electric wire
Electric signal.
Computer-readable program instructions as described herein can be downloaded to from computer readable storage medium it is each calculate/
Processing equipment, or outer computer or outer is downloaded to by network, such as internet, local area network, wide area network and/or wireless network
Portion stores equipment.Network may include copper transmission cable, optical fiber transmission, wireless transmission, router, firewall, interchanger, gateway
Computer and/or Edge Server.Adapter or network interface in each calculating/processing equipment are received from network to be counted
Calculation machine readable program instructions, and the computer-readable program instructions are forwarded, for the meter being stored in each calculating/processing equipment
In calculation machine readable storage medium storing program for executing.
Computer program instructions for executing disclosure operation can be assembly instruction, instruction set architecture (ISA) instructs,
Machine instruction, machine-dependent instructions, microcode, firmware instructions, condition setup data or with one or more programming languages
The source code or object code that any combination is write, the programming language include the programming language-of object-oriented such as
Smalltalk, C++ etc., and conventional procedural programming languages-such as " C " language or similar programming language.Computer
Readable program instructions can be executed fully on the user computer, partly execute on the user computer, be only as one
Vertical software package executes, part executes on the remote computer or completely in remote computer on the user computer for part
Or it is executed on server.In situations involving remote computers, remote computer can pass through network-packet of any kind
It includes local area network (LAN) or wide area network (WAN)-is connected to subscriber computer, or, it may be connected to outer computer (such as benefit
It is connected with ISP by internet).In some embodiments, by utilizing computer-readable program instructions
Status information carry out personalized customization electronic circuit, such as programmable logic circuit, field programmable gate array (FPGA) or can
Programmed logic array (PLA) (PLA), the electronic circuit can execute computer-readable program instructions, to realize each side of the disclosure
Face.
Referring herein to according to the flow chart of the method, apparatus (system) of the embodiment of the present disclosure and computer program product and/
Or block diagram describes various aspects of the disclosure.It should be appreciated that flowchart and or block diagram each box and flow chart and/
Or in block diagram each box combination, can be realized by computer-readable program instructions.
These computer-readable program instructions can be supplied to general purpose computer, special purpose computer or other programmable datas
The processor of processing unit, so that a kind of machine is produced, so that these instructions are passing through computer or other programmable datas
When the processor of processing unit executes, function specified in one or more boxes in implementation flow chart and/or block diagram is produced
The device of energy/movement.These computer-readable program instructions can also be stored in a computer-readable storage medium, these refer to
It enables so that computer, programmable data processing unit and/or other equipment work in a specific way, thus, it is stored with instruction
Computer-readable medium then includes a manufacture comprising in one or more boxes in implementation flow chart and/or block diagram
The instruction of the various aspects of defined function action.
Computer-readable program instructions can also be loaded into computer, other programmable data processing units or other
In equipment, so that series of operation steps are executed in computer, other programmable data processing units or other equipment, to produce
Raw computer implemented process, so that executed in computer, other programmable data processing units or other equipment
Instruct function action specified in one or more boxes in implementation flow chart and/or block diagram.
The flow chart and block diagram in the drawings show system, method and the computer journeys according to multiple embodiments of the disclosure
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
One module of table, program segment or a part of instruction, the module, program segment or a part of instruction include one or more use
The executable instruction of the logic function as defined in realizing.In some implementations as replacements, function marked in the box
It can occur in a different order than that indicated in the drawings.For example, two continuous boxes can actually be held substantially in parallel
Row, they can also be executed in the opposite order sometimes, and this depends on the function involved.It is also noted that block diagram and/or
The combination of each box in flow chart and the box in block diagram and or flow chart, can the function as defined in executing or dynamic
The dedicated hardware based system made is realized, or can be realized using a combination of dedicated hardware and computer instructions.
The presently disclosed embodiments is described above, above description is exemplary, and non-exclusive, and
It is not limited to disclosed each embodiment.Without departing from the scope and spirit of illustrated each embodiment, for this skill
Many modifications and changes are obvious for the those of ordinary skill in art field.The selection of term used herein, purport
In the principle, practical application or technological improvement to technology in market for best explaining each embodiment, or make the art
Other those of ordinary skill can understand each embodiment disclosed herein.
Claims (10)
1. a kind of facial image recognition method, which is characterized in that the described method includes:
The image data to be processed for belonging to different faces is extracted from face image data；
According to the image data to be processed for belonging to different faces, the face image data pair of non-matching is obtained；Wherein, described
The face image data of non-matching is to for characterizing the feature for belonging to two facial images of different faces；
According to the face image data pair of the non-matching, facial image identification network is trained, people for identification is obtained
The target identification network of face image.
2. the method according to claim 1, wherein described extract from face image data belongs to different faces
Image data to be processed, comprising:
Network is identified according to the facial image, extracts the feature of facial image in the face image data；
The feature of different faces image will be belonged to as the image data to be processed.
3. according to the method described in claim 2, it is characterized in that, the image to be processed for belonging to different faces according to
Data obtain the face image data pair of non-matching, comprising:
The feature for belonging to different faces includes at least the second feature in fisrt feature and the second face in the first face；
In the case where meeting preset condition according to the similarity that the fisrt feature and the second feature obtain, by described first
Face and second face are configured to the face image data pair.
4. according to the method described in claim 2, it is characterized in that, it is described to facial image identification network be trained before,
The method also includes:
According to the feature correlation between the face image data pair, sampling order is obtained.
5. according to the method described in claim 4, it is characterized in that, the feature according between the face image data pair
Correlation obtains sampling order, comprising:
According to the feature between the face image data pair, characteristic set is obtained；
According to the characteristic set construction feature tree KD-Tree, the close feature of feature correlation between face image data pair
Adjacent node as the KD-Tree；
Traverse path that the KD-Tree is obtained will be traversed as the sampling order.
6. method according to claim 4 or 5, which is characterized in that it is described obtain sampling order after, the method is also wrapped
It includes:
By the face image data read according to the sampling order to as the training for inputting the facial image identification network
Sample.
7. method according to claim 1 to 6, which is characterized in that the face image data pair is at least come
The second face image set obtained derived from the first face image set for face training with acquisition face, and two faces
Face in image collection is not identical.
8. a kind of facial image identification device, which is characterized in that described device includes:
Extraction unit, for extracting the image data to be processed for belonging to different faces from face image data；
First processing units obtain the face figure of non-matching for belonging to the image data to be processed of different faces according to
As data pair；Wherein, the face image data of the non-matching is to for characterizing two facial images for belonging to different faces
Feature；
The second processing unit instructs facial image identification network for the face image data pair according to the non-matching
Practice, obtains the target identification network of facial image for identification.
9. a kind of electronic equipment characterized by comprising
Processor；
Memory for storage processor executable instruction；
Wherein, the processor is configured to: perform claim require any one of 1 to 7 described in method.
10. a kind of computer readable storage medium, is stored thereon with computer program instructions, which is characterized in that the computer
Method described in any one of claim 1 to 7 is realized when program instruction is executed by processor.
Priority Applications (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910739381.8A CN110458102A (en) | 2019-08-12 | 2019-08-12 | A kind of facial image recognition method and device, electronic equipment and storage medium |
| PCT/CN2020/089012 WO2021027343A1 (en) | 2019-08-12 | 2020-05-07 | Human face image recognition method and apparatus, electronic device, and storage medium |
| KR1020217026325A KR20210114511A (en) | 2019-08-12 | 2020-05-07 | Face image recognition method and apparatus, electronic device and storage medium |
| JP2021547720A JP2022520120A (en) | 2019-08-12 | 2020-05-07 | Face image recognition methods and devices, electrical equipment and storage media |
| TW109122357A TW202107337A (en) | 2019-08-12 | 2020-07-02 | Face image recognition method and device, electronic device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910739381.8A CN110458102A (en) | 2019-08-12 | 2019-08-12 | A kind of facial image recognition method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110458102A true CN110458102A (en) | 2019-11-15 |
Family
ID=68485929
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910739381.8A Pending CN110458102A (en) | 2019-08-12 | 2019-08-12 | A kind of facial image recognition method and device, electronic equipment and storage medium |
Country Status (5)
| Country | Link |
|---|---|
| JP (1) | JP2022520120A (en) |
| KR (1) | KR20210114511A (en) |
| CN (1) | CN110458102A (en) |
| TW (1) | TW202107337A (en) |
| WO (1) | WO2021027343A1 (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111339964A (en) * | 2020-02-28 | 2020-06-26 | 北京市商汤科技开发有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN112149732A (en) * | 2020-09-23 | 2020-12-29 | 上海商汤智能科技有限公司 | Image protection method, device, electronic device and storage medium |
| WO2021027343A1 (en) * | 2019-08-12 | 2021-02-18 | 深圳市商汤科技有限公司 | Human face image recognition method and apparatus, electronic device, and storage medium |
| CN112784823A (en) * | 2021-03-17 | 2021-05-11 | 中国工商银行股份有限公司 | Face image recognition method, face image recognition device, computing equipment and medium |
| CN112949634A (en) * | 2021-03-08 | 2021-06-11 | 北京交通大学 | Bird nest detection method for railway contact network |
| CN113269425A (en) * | 2021-05-18 | 2021-08-17 | 北京航空航天大学 | Quantitative evaluation method for health state of equipment under unsupervised condition and electronic equipment |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20230032092A (en) | 2021-08-30 | 2023-03-07 | 주식회사 엘지에너지솔루션 | A solid electrolyte membrane and all solid-state lithium secondary battery comprising the same |
| CN113807253A (en) * | 2021-09-17 | 2021-12-17 | 上海商汤智能科技有限公司 | Face recognition method and device, electronic equipment and storage medium |
| KR102705532B1 (en) * | 2021-11-30 | 2024-09-09 | 연세대학교 산학협력단 | Method and apparatus for evaluating 3D facial shape similarity with deep multiview perceptual representations |
| CN114255502B (en) * | 2021-12-23 | 2024-03-29 | 中国电信股份有限公司 | Face image generation method and device, face recognition method, equipment and medium |
| CN115909434B (en) * | 2022-09-07 | 2023-07-04 | 以萨技术股份有限公司 | Data processing system for acquiring facial image characteristics |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103679158A (en) * | 2013-12-31 | 2014-03-26 | 北京天诚盛业科技有限公司 | Face authentication method and device |
| CN109753875A (en) * | 2018-11-28 | 2019-05-14 | 北京的卢深视科技有限公司 | Face recognition method, device and electronic device based on perceptual loss of face attribute |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4606779B2 (en) * | 2004-06-07 | 2011-01-05 | グローリー株式会社 | Image recognition apparatus, image recognition method, and program causing computer to execute the method |
| JP4591215B2 (en) * | 2005-06-07 | 2010-12-01 | 株式会社日立製作所 | Facial image database creation method and apparatus |
| US8224042B2 (en) * | 2009-03-12 | 2012-07-17 | Seiko Epson Corporation | Automatic face recognition |
| JP6312485B2 (en) * | 2014-03-25 | 2018-04-18 | キヤノン株式会社 | Information processing apparatus, authentication apparatus, and methods thereof |
| CN110458102A (en) * | 2019-08-12 | 2019-11-15 | 深圳市商汤科技有限公司 | A kind of facial image recognition method and device, electronic equipment and storage medium |
-
2019
- 2019-08-12 CN CN201910739381.8A patent/CN110458102A/en active Pending
-
2020
- 2020-05-07 KR KR1020217026325A patent/KR20210114511A/en not_active Abandoned
- 2020-05-07 WO PCT/CN2020/089012 patent/WO2021027343A1/en active Application Filing
- 2020-05-07 JP JP2021547720A patent/JP2022520120A/en active Pending
- 2020-07-02 TW TW109122357A patent/TW202107337A/en unknown
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103679158A (en) * | 2013-12-31 | 2014-03-26 | 北京天诚盛业科技有限公司 | Face authentication method and device |
| CN109753875A (en) * | 2018-11-28 | 2019-05-14 | 北京的卢深视科技有限公司 | Face recognition method, device and electronic device based on perceptual loss of face attribute |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021027343A1 (en) * | 2019-08-12 | 2021-02-18 | 深圳市商汤科技有限公司 | Human face image recognition method and apparatus, electronic device, and storage medium |
| CN111339964A (en) * | 2020-02-28 | 2020-06-26 | 北京市商汤科技开发有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN112149732A (en) * | 2020-09-23 | 2020-12-29 | 上海商汤智能科技有限公司 | Image protection method, device, electronic device and storage medium |
| CN112949634A (en) * | 2021-03-08 | 2021-06-11 | 北京交通大学 | Bird nest detection method for railway contact network |
| CN112949634B (en) * | 2021-03-08 | 2024-04-26 | 北京交通大学 | A method for detecting bird nests in railway contact network |
| CN112784823A (en) * | 2021-03-17 | 2021-05-11 | 中国工商银行股份有限公司 | Face image recognition method, face image recognition device, computing equipment and medium |
| CN113269425A (en) * | 2021-05-18 | 2021-08-17 | 北京航空航天大学 | Quantitative evaluation method for health state of equipment under unsupervised condition and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2022520120A (en) | 2022-03-28 |
| WO2021027343A1 (en) | 2021-02-18 |
| KR20210114511A (en) | 2021-09-23 |
| TW202107337A (en) | 2021-02-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110458102A (en) | A kind of facial image recognition method and device, electronic equipment and storage medium | |
| CN109614613A (en) | The descriptive statement localization method and device of image, electronic equipment and storage medium | |
| CN110909815B (en) | Neural network training method, neural network training device, neural network processing device, neural network training device, image processing device and electronic equipment | |
| CN110348537A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN110472091B (en) | Image processing method and device, electronic equipment and storage medium | |
| CN109800737A (en) | Face recognition method and device, electronic equipment and storage medium | |
| CN109948494A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN109816764A (en) | Image generating method and device, electronic equipment and storage medium | |
| CN110532956A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN110188236A (en) | A kind of recommended method of music, apparatus and system | |
| CN108985176A (en) | image generating method and device | |
| CN109934275A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN109711546A (en) | Neural network training method and device, electronic equipment and storage medium | |
| CN110659690B (en) | Neural network construction method and device, electronic equipment and storage medium | |
| CN109543537A (en) | Weight identification model increment training method and device, electronic equipment and storage medium | |
| CN111259967A (en) | Image classification and neural network training method, device, equipment and storage medium | |
| CN109977860A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN110070049A (en) | Facial image recognition method and device, electronic equipment and storage medium | |
| CN110175546A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN110378312A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN109543536A (en) | Image identification method and device, electronic equipment and storage medium | |
| CN109920016A (en) | Image generating method and device, electronic equipment and storage medium | |
| CN110019960A (en) | Data processing method and device, electronic equipment and storage medium | |
| CN109934240A (en) | Feature update method and device, electronic equipment and storage medium | |
| CN109685041A (en) | Image analysis method and device, electronic equipment and storage medium |
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 | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 40015798 Country of ref document: HK |
|
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191115 |
|
| RJ01 | Rejection of invention patent application after publication |