CN111209878A - Cross-age face recognition method and device - Google Patents
Cross-age face recognition method and device Download PDFInfo
- Publication number
- CN111209878A CN111209878A CN202010028274.7A CN202010028274A CN111209878A CN 111209878 A CN111209878 A CN 111209878A CN 202010028274 A CN202010028274 A CN 202010028274A CN 111209878 A CN111209878 A CN 111209878A
- Authority
- CN
- China
- Prior art keywords
- face
- picture
- age
- recognized
- feature extraction
- 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
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000000605 extraction Methods 0.000 claims abstract description 93
- 239000013598 vector Substances 0.000 claims abstract description 35
- 238000012549 training Methods 0.000 claims description 35
- 230000001815 facial effect Effects 0.000 claims description 15
- 238000013526 transfer learning Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 11
- 230000004927 fusion Effects 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000032683 aging Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Images
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
- G06V40/172—Classification, e.g. identification
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- 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
- G06V40/178—Human faces, e.g. facial parts, sketches or expressions estimating age from face image; using age information for improving recognition
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The embodiment of the invention provides a cross-age face recognition method and device. The method comprises the following steps: inputting a face picture to be recognized into a picture generation model, and acquiring generation pictures of a plurality of age groups corresponding to the face picture to be recognized; respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group, and outputting basic human face features and human face features of each age group; acquiring a feature vector according to the basic face features and the face features of all age groups; and acquiring the recognition result of the face picture to be recognized according to the feature vector and a face library acquired in advance. According to the cross-age face recognition method and device provided by the embodiment of the invention, the faces of different age groups are generated, the features of the faces in the face picture to be recognized and the generated faces of different age groups are respectively extracted and fused, and the face recognition is carried out according to the fused face features, so that the recognition accuracy of the cross-age face recognition can be improved.
Description
Technical Field
The invention relates to the technical field of computers, in particular to an age-crossing face recognition method and device.
Background
Face recognition is becoming more and more widely used as an important biometric feature recognition technology. However, the recognition rate of face recognition is greatly affected by age, and in face recognition, the face difference between different individuals is often smaller than that of the same individual under different conditions (such as face difference under different ages of 0).
Currently, the cross-age face recognition generally converts the faces of different ages into fixed representations at one or more ages in a face synthesis mode, and then performs face recognition. Because the span between different ages is as long as years or even decades, the training of the face recognition model based on the method needs very large data volume, the face data acquisition cannot meet two requirements of large data volume and large age span, the multidimensional face features and the age features are mixed together and are difficult to distinguish, and the high-dimensional face features have no interpretability, so that the selection difficulty of the feature matching method is caused. The deep neural network achieves high recognition accuracy rate based on a large amount of training data, but the amount of trans-age face training data is limited, and the trained face recognition network is used for each age group: children, adults and the elderly lack generalization ability. The data volume is limited and not enough to train the neural network to achieve high recognition accuracy,
in conclusion, the existing age-crossing face recognition method is low in recognition accuracy.
Disclosure of Invention
The embodiment of the invention provides an age-crossing face recognition method and device, which are used for solving or at least partially solving the defect of low recognition accuracy rate in the prior art.
In a first aspect, an embodiment of the present invention provides an age-related face recognition method, including:
inputting a face picture to be recognized into a picture generation model, and acquiring generation pictures of a plurality of age groups corresponding to the face picture to be recognized;
respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group, and outputting basic human face features and human face features of each age group;
acquiring a feature vector according to the basic face features and the face features of all age groups;
acquiring the recognition result of the face picture to be recognized according to the feature vector and a face library acquired in advance;
the image generation model is obtained by training according to sample data of a human face; the basic feature extraction model is obtained by training according to the sample data of the human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
Preferably, before the step of inputting the face picture to be recognized and each of the generated pictures into the basic feature extraction model and the feature extraction model of each of the age groups and outputting the basic face features and the face features of each of the age groups, the method further includes:
training according to the sample data of the human face to obtain the basic feature extraction model;
and for each age group, finely adjusting the basic feature extraction model according to the face data of the age group in the sample data of the face to obtain the feature extraction model of the age group.
Preferably, before the step of inputting the face picture to be recognized into the picture generation model and obtaining the generated pictures of a plurality of age groups corresponding to the face picture to be recognized, the method further includes:
performing countermeasure training according to the sample data of the human face to obtain a discriminator and a picture generation model;
the discriminator is used for judging whether the input face picture is the sample data of the face or the face data generated by the picture generation model.
Preferably, the picture generation model comprises an encoder and a generator;
correspondingly, the specific steps of inputting the face picture to be recognized into the picture generation model and acquiring the generated pictures of a plurality of age groups corresponding to the face picture to be recognized comprise:
inputting the face picture to be recognized into the encoder, and projecting the face picture to be recognized into a feature space corresponding to each age group according to each age group;
and the generator up-samples the characteristics of the face picture to be recognized in the characteristic space corresponding to the age group to obtain the generated picture of the age group.
Preferably, the specific step of obtaining the feature vector according to the basic face features and the face features of each age group includes:
and acquiring the feature vector according to the basic face features, the face features of all age groups and preset weights.
Preferably, before the step of inputting the face picture to be recognized into the generator and acquiring the generated pictures of a plurality of age groups corresponding to the face picture to be recognized, the method further includes:
and carrying out face key point detection and face alignment on the original picture to obtain the face picture to be recognized.
Preferably, the age groups include: children, adults, and the elderly.
In a second aspect, an embodiment of the present invention provides an age-related face recognition apparatus, including:
the image generation module is used for inputting the face image to be recognized into the image generation model and acquiring the generated images of a plurality of age groups corresponding to the face image to be recognized;
the feature extraction module is used for respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group and outputting basic human face features and human face features of each age group;
the feature fusion module is used for acquiring feature vectors according to the basic face features and the face features of all age groups;
the face recognition module is used for acquiring a recognition result of the face picture to be recognized according to the feature vector and a face library acquired in advance;
the image generation model is obtained by training according to sample data of a human face; the basic feature extraction model is obtained by training according to the sample data of the human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
In a third aspect, an embodiment of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the computer program is executed, the steps of the cross-age face recognition method provided in any one of the various possible implementations of the first aspect are implemented.
In a fourth aspect, an embodiment of the present invention provides a non-transitory computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the steps of the cross-age face recognition method as provided in any one of the various possible implementations of the first aspect.
According to the cross-age face recognition method and device provided by the embodiment of the invention, the faces of different age groups are generated, the features of the faces in the face picture to be recognized and the generated faces of different age groups are respectively extracted through the plurality of feature extraction models and are fused, and the face recognition is carried out according to the fused face features, so that the recognition accuracy rate of the cross-age face recognition can be improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
Fig. 1 is a schematic flowchart of a cross-age face recognition method according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a cross-age face recognition apparatus according to an embodiment of the present invention;
fig. 3 is a schematic physical structure diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to solve the above problems in the prior art, embodiments of the present invention provide a method and an apparatus for identifying a cross-age face, where the method and apparatus include generating faces of different age groups corresponding to faces in a face picture to be identified according to the face picture to be identified, obtaining multiple feature extraction models respectively used for extracting face features of different age groups based on transfer learning, and fusing the face features extracted by the feature extraction models to obtain stronger face features, thereby improving the identification accuracy of a cross-age face identification task.
Fig. 1 is a schematic flowchart of a cross-age face recognition method according to an embodiment of the present invention. As shown in fig. 1, the method includes: step S101, inputting the face picture to be recognized into a picture generation model, and obtaining generation pictures of a plurality of age groups corresponding to the face picture to be recognized.
The image generation model is obtained by training according to sample data of the human face.
Specifically, the image generation model is used for generating a generated image of the age group corresponding to the face image to be recognized according to each age group and the face image to be recognized.
The plurality of age groups are predetermined.
The face in the face picture to be recognized and the face in the generated picture of any age group corresponding to the face picture to be recognized are the faces of the same person.
Therefore, a picture is generated, which is a face picture of a person in the face picture to be recognized in a certain age range determined in advance.
And S102, respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group, and outputting basic human face features and human face features of each age group.
The basic feature extraction model is obtained by training according to sample data of a human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
Specifically, each feature extraction model includes a basic feature extraction model and a feature extraction model of each age group.
And inputting the face picture to be recognized into the basic feature extraction model, and outputting the basic face features.
And inputting the generated picture of each age group corresponding to the face picture to be recognized into the feature extraction model of the age group, and outputting the face features of the age group.
It is understood that the basic facial features and the facial features of any age group include the same parameters, but the same parameters may have different values.
And step S103, acquiring a feature vector according to the basic face features and the face features of all age groups.
Specifically, the basic face features and the face features of all age groups are fused or spliced to obtain feature vectors.
And step S104, acquiring the recognition result of the face picture to be recognized according to the feature vector and a face library acquired in advance.
Specifically, the feature vector and a pre-acquired face library can be matched in a ratio of 1:1, so that the face most similar to the face picture to be recognized in the face library and the corresponding similarity are obtained.
And calculating the distance between the obtained feature vector and the feature vector of each face in the pre-acquired face library.
And acquiring the similarity between the face in the face picture to be recognized and the face in the face library according to the distance.
And judging whether the maximum value of the similarity is larger than a preset similarity threshold value.
And if so, taking the face in the face library corresponding to the maximum similarity as a face recognition result of the face picture to be recognized, wherein the face in the face picture to be recognized is the face in the face library.
If the number of the face images is less than the preset number, the face in the face image to be recognized is not the face in the face library, and the face in the face library is not matched with the face image to be recognized.
The human face of each age group generated according to the human face picture to be recognized can effectively approach human face samples of different ages of one person, and the extracted features are more effective by combining the human face recognition model corresponding to each age group, so that the generalization capability of the whole algorithm can be improved
The embodiment of the invention can improve the identification accuracy of the cross-age face identification by generating the faces of different ages, respectively extracting the features of the faces in the picture of the faces to be identified and the generated faces of different ages through a plurality of feature extraction models, fusing the features, and identifying the faces according to the fused face features. And the steps are simpler and more convenient.
Based on the content of each embodiment, before the facial image to be recognized and each generated image are respectively input to the basic feature extraction model and the feature extraction model of each age group and the basic facial features and the facial features of each age group are output, the method further includes: training is carried out according to the sample data of the human face, and a basic feature extraction model is obtained.
The sample data of the face is face data without age span, which does not distinguish age groups. For example, a face database such as ms _ celeb or cascia may be used for model training.
The face features of each age group vary greatly due to the effect of age aging. Under the common conditions, the inter-class variance of the faces of children is small, the face characteristics of adults are stable, the identity characteristics can be embodied to the maximum extent, and the face of the elderly has the problem of large intra-class variance and conflicts with the small inter-class variance of the children. Training multiple ages together can result in a model that does not achieve optimal performance for multiple ages at the same time. Each age group needs to train a corresponding feature extraction model.
Because the age-crossing face data set is very small and is difficult to collect, good feature extraction models cannot be trained directly, and each feature extraction model can be obtained by adopting a transfer learning mode.
When the basic feature extraction model is trained, SoftmaxLoss and ContrastivLoss loss functions can be adopted as the target functions for training.
Softmax optimizes the distance between classes to maximize the distance between different classes, while contentiveloss optimizes the variance within the classes to minimize the variance within the classes.
The basic feature extraction model can be constructed based on a neural network.
And for each age group, finely adjusting the basic feature extraction model according to the face data of the age group in the sample data of the face to obtain the feature extraction model of the age group.
Specifically, during the migration learning, each parameter of the basic feature extraction model is shared with the feature extraction model of each age, that is, the basic feature extraction model is used as the feature extraction model of each age before fine adjustment.
For each age group, based on the face data of the age group in the sample data of the face, the feature extraction model of the age group is finely adjusted, and the basic feature extraction model is kept, so that the trained feature extraction model of the age group not only learns stronger features from big data, but also can distribute the face data of each age group, thereby extracting the face features irrelevant to the age, reducing the influence of age change on face recognition, and improving the recognition accuracy of the cross-age face recognition.
The embodiment of the invention overcomes the problem of less face sample data of a single age group through transfer learning, respectively trains the feature extraction model aiming at each preset age group and a face picture to be recognized, and expands rare training data, thereby fusing the face features of multiple age groups to obtain stronger face features and improving the recognition accuracy rate of the cross-age face recognition.
Based on the content of the above embodiments, before inputting the face picture to be recognized into the picture generation model and obtaining the generated pictures of a plurality of age groups corresponding to the face picture to be recognized, the method further includes: and performing countermeasure training according to the sample data of the human face to obtain a discriminator and a picture generation model.
The discriminator is used for judging whether the input face picture is the sample data of the face or the face data generated by the picture generation model.
Specifically, the picture generation model is trained by adopting a confrontation training mode, a confrontation training discriminator and the picture generation model.
The input of the discriminator is sample data (namely a real sample) of a human face and a human face picture (namely a generated sample) generated by the picture generation model according to the sample data of the human face, and under the condition of inputting age groups, a judgment result of whether the input is the real sample or the generated sample is output. The labels for the generated and authentic samples may be 0 and 1, respectively.
The input of the image generation model is sample data of a human face, and under the condition of inputting an age group, the human face image of the person in the sample data in the age group is output.
The discriminator distinguishes the generated sample or the real sample as much as possible, the picture generation model generates the vivid generated sample as much as possible to deceive the discriminator, and the countermeasure mechanism enables the final discriminator and the picture generation model to achieve good performance.
According to the embodiment of the invention, the discriminator and the picture generation model are obtained through the confrontation training, so that the picture generation model with better performance can be obtained, more generated pictures with biological characteristics can be reserved, the face characteristics irrelevant to the age can be extracted according to the generated pictures, the influence of age change on face recognition can be reduced, and the recognition accuracy rate of the cross-age face recognition can be improved.
Based on the contents of the above embodiments, the picture generation model includes an encoder and a generator.
Correspondingly, the specific steps of inputting the face picture to be recognized into the picture generation model and acquiring the generated pictures of a plurality of age groups corresponding to the face picture to be recognized comprise: and inputting the face picture to be recognized into an encoder, and projecting the face picture to be recognized into a feature space corresponding to each age group according to each age group.
Specifically, the encoder projects a face in an input face picture into a hidden space (i.e., a feature space) according to the input face and a corresponding age group.
The generator up-samples the features of the face picture to be recognized in the feature space corresponding to the age group to obtain the generated picture of the age group.
Specifically, the generator performs upsampling from the hidden space and the age group, and maps the features in the hidden space into the image space to obtain a face sample of the age group as a generated picture of the age group.
It should be noted that the picture generation model and the decider form a countermeasure network.
The aim of the counternetwork learning is to locate the input real face picture at the correct position of the manifold plane M of the face feature space.
The encoder maps the input picture into a feature space. The formula of the mapping is E (x) z ∈ Rn
Wherein n represents a dimension of the feature space; z represents a feature, namely a projection result of the face picture to be recognized in a feature space; x represents the input to the countermeasure network, i.e., the true sample.
Given the feature z and the age label l, the picture generation model outputs an upsampling formula of the generated sample as
The optimization goal of the countermeasure network is to generate face samples that also share the identity and age tags of the input faces on the flow plane.
The optimization objective function may be
The identity characteristics of the human face are controlled by Dz, so that the characteristics are uniformly distributed, and the human face identity characteristics can resist and learn with an encoder.
Denote the data distribution as pdata(x) Characteristic distribution q (z | x). Assuming that p (z) is a known distribution, z x p (z) represents the random sampling process from p (z).
The E and Dz networks can be trained with a minimax optimization function.
With the same idea, under the condition of age label, DimageTraining of sum S by optimizing the following expression
The embodiment of the invention can obtain the picture generation model with better performance by the counterstudy of the encoder and the generator, thereby reserving more generated pictures with biological characteristics, extracting the face characteristics irrelevant to the age according to the generated pictures, reducing the influence of age change on face identification and improving the identification accuracy of the cross-age face identification.
Based on the content of the above embodiments, the specific steps of obtaining the feature vector according to the basic face features and the face features of each age group include: and acquiring a feature vector according to the basic face features, the face features of all age groups and preset weights.
Specifically, different weights may be given to the base face features and the age-based face features.
It can be understood that, because the face picture to be recognized is an actual face, the weight of the basic face features is the largest, and the weights of the face features of all age groups can be set according to the actual situation, but are all smaller than the weight of the basic face features.
The facial features can be fused and spliced according to the basic facial features, the facial features of all ages and preset weights, and the final facial features can be obtained by adopting methods such as weighted sum or weighted average.
The fusion and splicing results may also be normalized, for example, using L2 to obtain the final facial features.
The final face features may be in the form of vectors, i.e., feature vectors.
According to the embodiment of the invention, the characteristic vector is obtained according to the basic face characteristics, the face characteristics of all ages and the preset weight, so that stronger face characteristics can be obtained, the influence of age change on face recognition can be reduced, and the recognition accuracy of the age-crossing face recognition can be improved.
Based on the content of the above embodiments, before the face picture to be recognized is input to the generator and generated pictures of a plurality of age groups corresponding to the face picture to be recognized are obtained, the method further includes: and carrying out face key point detection and face alignment on the original picture to obtain a face picture to be recognized.
Specifically, the original picture may be subjected to face key point detection to locate a plurality of key points of the face, so as to obtain an affine transformation matrix.
It is understood that the original picture is a picture including a human face.
And carrying out face alignment on the original picture according to the affine transformation matrix, and further obtaining an aligned face picture as the face picture to be recognized.
According to the embodiment of the invention, the aligned face picture is obtained by face key point detection and face alignment, and the identification accuracy of the age-crossing face identification can be improved.
Based on the content of the above embodiments, the age group includes: children, adults, and the elderly.
Specifically, children, adults and the elderly are age groups with strong representativeness, so that generated pictures can be obtained and facial features of people in all age groups can be extracted by adopting the three age groups of the children, the adults and the elderly.
According to the embodiment of the invention, the generated pictures are obtained according to three age groups of children, adults and the elderly, and the facial features of all age groups are extracted, so that the features of the faces of different age groups can be fused, the face recognition is carried out according to the fused facial features, and the recognition accuracy rate of the cross-age face recognition can be improved.
Fig. 2 is a schematic structural diagram of a cross-age face recognition device according to an embodiment of the present invention. Based on the content of the foregoing embodiments, as shown in fig. 2, the apparatus includes a picture generation module 201, a feature extraction module 202, a feature fusion module 203, and a face recognition module 204, where:
the image generation module 201 is configured to input a face image to be recognized into an image generation model, and obtain generated images of a plurality of age groups corresponding to the face image to be recognized;
the feature extraction module 202 is configured to input the face picture to be recognized and each generated picture into the basic feature extraction model and the feature extraction models of each age group, and output the basic face features and the face features of each age group;
the feature fusion module 203 is used for acquiring feature vectors according to the basic face features and the face features of all age groups;
the face recognition module 204 is configured to obtain a recognition result of a face picture to be recognized according to the feature vector and a pre-obtained face library;
the image generation model is obtained by training according to sample data of a human face; the basic feature extraction model is obtained by training according to sample data of the human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
Specifically, the image generation module 201, the feature extraction module 202, the feature fusion module 203, and the face recognition module 204 are electrically connected in sequence.
The image generating module 201 generates a generated image of the age group corresponding to the face image to be recognized according to each age group and the face image to be recognized.
The feature extraction module 202 inputs the face picture to be recognized into the basic feature extraction model, outputs the basic face features, inputs the generated picture of each age group corresponding to the face picture to be recognized into the feature extraction model of the age group, and outputs the face features of the age group.
The feature fusion module 203 fuses or splices the basic face features and the face features of all age groups to obtain feature vectors.
The face recognition module 204 performs 1:1 matching on the feature vector and a face library acquired in advance to obtain a face most similar to the face picture to be recognized in the face library and a corresponding similarity, and judges whether the similarity is greater than a preset similarity threshold.
And if so, taking the face in the face library corresponding to the maximum similarity as a face recognition result of the face picture to be recognized, wherein the face in the face picture to be recognized is the face in the face library.
If the number of the face images is less than the preset number, the face in the face image to be recognized is not the face in the face library, and the face in the face library is not matched with the face image to be recognized.
The specific method and flow for realizing the corresponding functions of each module included in the cross-age face recognition device are described in the embodiment of the cross-age face recognition method, and are not described herein again.
The cross-age face recognition device is used for the cross-age face recognition method of the embodiments. Therefore, the description and definition in the cross-age face recognition method in the foregoing embodiments can be used for understanding the execution modules in the embodiments of the present invention.
The embodiment of the invention can improve the identification accuracy of the cross-age face identification by generating the faces of different ages, respectively extracting the features of the faces in the picture of the faces to be identified and the generated faces of different ages through a plurality of feature extraction models, fusing the features, and identifying the faces according to the fused face features. And the steps are simpler and more convenient.
Fig. 3 is a schematic physical structure diagram of an electronic device according to an embodiment of the present invention. Based on the content of the above embodiment, as shown in fig. 3, the electronic device may include: a processor (processor)301, a memory (memory)302, and a bus 303; wherein, the processor 301 and the memory 302 complete the communication with each other through the bus 303; the processor 301 is configured to invoke computer program instructions stored in the memory 302 and executable on the processor 301 to perform the cross-age face recognition methods provided by the above-described method embodiments, including, for example: inputting a face picture to be recognized into a picture generation model, and acquiring generation pictures of a plurality of age groups corresponding to the face picture to be recognized; respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group, and outputting basic human face features and human face features of each age group; acquiring a feature vector according to the basic face features and the face features of all age groups; acquiring an identification result of a face picture to be identified according to the feature vector and a face library acquired in advance; the image generation model is obtained by training according to sample data of a human face; the basic feature extraction model is obtained by training according to sample data of the human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
Another embodiment of the present invention discloses a computer program product, the computer program product includes a computer program stored on a non-transitory computer readable storage medium, the computer program includes program instructions, when the program instructions are executed by a computer, the computer can execute the cross-age face recognition method provided by the above-mentioned method embodiments, for example, the method includes: inputting a face picture to be recognized into a picture generation model, and acquiring generation pictures of a plurality of age groups corresponding to the face picture to be recognized; respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group, and outputting basic human face features and human face features of each age group; acquiring a feature vector according to the basic face features and the face features of all age groups; acquiring an identification result of a face picture to be identified according to the feature vector and a face library acquired in advance; the image generation model is obtained by training according to sample data of a human face; the basic feature extraction model is obtained by training according to sample data of the human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
Furthermore, the logic instructions in the memory 302 may be implemented in software functional units and stored in a computer readable storage medium when sold or used as a stand-alone product. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or make a contribution to the prior art, or may be implemented in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the methods of the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Another embodiment of the present invention provides a non-transitory computer-readable storage medium, which stores computer instructions, where the computer instructions cause a computer to execute the cross-age face recognition method provided by the foregoing method embodiments, for example, including: inputting a face picture to be recognized into a picture generation model, and acquiring generation pictures of a plurality of age groups corresponding to the face picture to be recognized; respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group, and outputting basic human face features and human face features of each age group; acquiring a feature vector according to the basic face features and the face features of all age groups; acquiring an identification result of a face picture to be identified according to the feature vector and a face library acquired in advance; the image generation model is obtained by training according to sample data of a human face; the basic feature extraction model is obtained by training according to sample data of the human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. It is understood that the above-described technical solutions may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method of the above-described embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A cross-age face recognition method is characterized by comprising the following steps:
inputting a face picture to be recognized into a picture generation model, and acquiring generation pictures of a plurality of age groups corresponding to the face picture to be recognized;
respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group, and outputting basic human face features and human face features of each age group;
acquiring a feature vector according to the basic face features and the face features of all age groups;
acquiring the recognition result of the face picture to be recognized according to the feature vector and a face library acquired in advance;
the image generation model is obtained by training according to sample data of a human face; the basic feature extraction model is obtained by training according to the sample data of the human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
2. The method according to claim 1, wherein before the step of inputting the face image to be recognized and each of the generated images into the basic feature extraction model and the feature extraction model of each of the age groups and outputting the basic face features and the face features of each of the age groups, the method further comprises:
training according to the sample data of the human face to obtain the basic feature extraction model;
and for each age group, finely adjusting the basic feature extraction model according to the face data of the age group in the sample data of the face to obtain the feature extraction model of the age group.
3. The cross-age face recognition method according to claim 1, wherein before the face picture to be recognized is input to a picture generation model and generated pictures of a plurality of age groups corresponding to the face picture to be recognized are obtained, the method further comprises:
performing countermeasure training according to the sample data of the human face to obtain a discriminator and a picture generation model;
the discriminator is used for judging whether the input face picture is the sample data of the face or the face data generated by the picture generation model.
4. The cross-age face recognition method of claim 1, wherein the picture generation model comprises an encoder and a generator;
correspondingly, the specific steps of inputting the face picture to be recognized into the picture generation model and acquiring the generated pictures of a plurality of age groups corresponding to the face picture to be recognized comprise:
inputting the face picture to be recognized into the encoder, and projecting the face picture to be recognized into a feature space corresponding to each age group according to each age group;
and the generator up-samples the characteristics of the face picture to be recognized in the characteristic space corresponding to the age group to obtain the generated picture of the age group.
5. The method according to claim 1, wherein the step of obtaining feature vectors according to the basic facial features and the facial features of each age group comprises:
and acquiring the feature vector according to the basic face features, the face features of all age groups and preset weights.
6. The cross-age face recognition method according to claim 1, wherein before the step of inputting the face picture to be recognized into the generator and obtaining the generated pictures of the plurality of age groups corresponding to the face picture to be recognized, the method further comprises:
and carrying out face key point detection and face alignment on the original picture to obtain the face picture to be recognized.
7. The cross-age face recognition method according to any one of claims 1 to 6, wherein the age groups include: children, adults, and the elderly.
8. An age-spanning face recognition device, comprising:
the image generation module is used for inputting the face image to be recognized into the image generation model and acquiring the generated images of a plurality of age groups corresponding to the face image to be recognized;
the feature extraction module is used for respectively inputting the human face picture to be recognized and each generated picture into a basic feature extraction model and a feature extraction model of each age group and outputting basic human face features and human face features of each age group;
the feature fusion module is used for acquiring feature vectors according to the basic face features and the face features of all age groups;
the face recognition module is used for acquiring a recognition result of the face picture to be recognized according to the feature vector and a face library acquired in advance;
the image generation model is obtained by training according to sample data of a human face; the basic feature extraction model is obtained by training according to the sample data of the human face; the feature extraction model of each age group is obtained by performing transfer learning based on the basic feature extraction model.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the steps of the cross-age face recognition method according to any one of claims 1 to 7 are implemented when the processor executes the program.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the cross-age face recognition method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010028274.7A CN111209878A (en) | 2020-01-10 | 2020-01-10 | Cross-age face recognition method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010028274.7A CN111209878A (en) | 2020-01-10 | 2020-01-10 | Cross-age face recognition method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111209878A true CN111209878A (en) | 2020-05-29 |
Family
ID=70785508
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010028274.7A Pending CN111209878A (en) | 2020-01-10 | 2020-01-10 | Cross-age face recognition method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111209878A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401339A (en) * | 2020-06-01 | 2020-07-10 | 北京金山云网络技术有限公司 | Method and device for identifying age of person in face image and electronic equipment |
| CN111860335A (en) * | 2020-07-22 | 2020-10-30 | 安徽兰臣信息科技有限公司 | Intelligent wearable device based on face recognition |
| CN111860333A (en) * | 2020-07-22 | 2020-10-30 | 安徽兰臣信息科技有限公司 | Child face recognition algorithm for cross-age recognition |
| CN111881722A (en) * | 2020-06-10 | 2020-11-03 | 广东芯盾微电子科技有限公司 | Cross-age face recognition method, system, device and storage medium |
| CN111881721A (en) * | 2020-06-10 | 2020-11-03 | 广东芯盾微电子科技有限公司 | Cross-age face recognition method, system, device and storage medium |
| CN112101204A (en) * | 2020-09-14 | 2020-12-18 | 北京百度网讯科技有限公司 | Training method of generative countermeasure network, image processing method, device and equipment |
| CN112766169A (en) * | 2021-01-21 | 2021-05-07 | 中信银行股份有限公司 | Account opening identity authentication method, device, equipment and storage medium |
| CN112906525A (en) * | 2021-02-05 | 2021-06-04 | 广州市百果园信息技术有限公司 | Age identification method and device and electronic equipment |
| CN113076833A (en) * | 2021-03-25 | 2021-07-06 | 深圳数联天下智能科技有限公司 | Training method of age identification model, face age identification method and related device |
| CN113221660A (en) * | 2021-04-14 | 2021-08-06 | 河海大学 | Cross-age face recognition method based on feature fusion |
| CN114582006A (en) * | 2022-05-06 | 2022-06-03 | 广东红橙云大数据有限公司 | Child age-crossing face recognition method and device, electronic equipment and medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107480676A (en) * | 2017-07-28 | 2017-12-15 | 浙江大华技术股份有限公司 | A kind of vehicle color identification method, device and electronic equipment |
| CN108416310A (en) * | 2018-03-14 | 2018-08-17 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN108846350A (en) * | 2018-06-08 | 2018-11-20 | 江苏大学 | Tolerate the face identification method of change of age |
| CN108875767A (en) * | 2017-12-07 | 2018-11-23 | 北京旷视科技有限公司 | Method, apparatus, system and the computer storage medium of image recognition |
| CN109255289A (en) * | 2018-07-27 | 2019-01-22 | 电子科技大学 | A kind of across aging face identification method generating model based on unified formula |
| CN110059652A (en) * | 2019-04-24 | 2019-07-26 | 腾讯科技（深圳）有限公司 | Face image processing process, device and storage medium |
| CN110197099A (en) * | 2018-02-26 | 2019-09-03 | 腾讯科技（深圳）有限公司 | The method and apparatus of across age recognition of face and its model training |
| CN110399811A (en) * | 2019-07-08 | 2019-11-01 | 厦门市美亚柏科信息股份有限公司 | A kind of face identification method, device and storage medium |
| WO2019233421A1 (en) * | 2018-06-04 | 2019-12-12 | 京东数字科技控股有限公司 | Image processing method and device, electronic apparatus, and storage medium |
| CN115147904A (en) * | 2022-07-15 | 2022-10-04 | 北京建筑大学 | Face recognition method, face recognition device, face recognition processing equipment and storage medium |
-
2020
- 2020-01-10 CN CN202010028274.7A patent/CN111209878A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107480676A (en) * | 2017-07-28 | 2017-12-15 | 浙江大华技术股份有限公司 | A kind of vehicle color identification method, device and electronic equipment |
| CN108875767A (en) * | 2017-12-07 | 2018-11-23 | 北京旷视科技有限公司 | Method, apparatus, system and the computer storage medium of image recognition |
| CN110197099A (en) * | 2018-02-26 | 2019-09-03 | 腾讯科技（深圳）有限公司 | The method and apparatus of across age recognition of face and its model training |
| CN108416310A (en) * | 2018-03-14 | 2018-08-17 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| WO2019233421A1 (en) * | 2018-06-04 | 2019-12-12 | 京东数字科技控股有限公司 | Image processing method and device, electronic apparatus, and storage medium |
| CN108846350A (en) * | 2018-06-08 | 2018-11-20 | 江苏大学 | Tolerate the face identification method of change of age |
| CN109255289A (en) * | 2018-07-27 | 2019-01-22 | 电子科技大学 | A kind of across aging face identification method generating model based on unified formula |
| CN110059652A (en) * | 2019-04-24 | 2019-07-26 | 腾讯科技（深圳）有限公司 | Face image processing process, device and storage medium |
| CN110399811A (en) * | 2019-07-08 | 2019-11-01 | 厦门市美亚柏科信息股份有限公司 | A kind of face identification method, device and storage medium |
| CN115147904A (en) * | 2022-07-15 | 2022-10-04 | 北京建筑大学 | Face recognition method, face recognition device, face recognition processing equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| HONGYU YANG等: "Learning Face Age Progression: A Pyramid Architecture of GANs" * |
| 刘玉坤;樊爱宛;廖海斌;: "分而冶之的人脸图像年龄估计" * |
| 胡俊;: "基于深度学习的跨年龄人脸识别" * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401339A (en) * | 2020-06-01 | 2020-07-10 | 北京金山云网络技术有限公司 | Method and device for identifying age of person in face image and electronic equipment |
| CN111401339B (en) * | 2020-06-01 | 2020-09-29 | 北京金山云网络技术有限公司 | Method and device for identifying age of person in face image and electronic equipment |
| CN111881722B (en) * | 2020-06-10 | 2021-08-24 | 广东芯盾微电子科技有限公司 | Cross-age face recognition method, system, device and storage medium |
| CN111881722A (en) * | 2020-06-10 | 2020-11-03 | 广东芯盾微电子科技有限公司 | Cross-age face recognition method, system, device and storage medium |
| CN111881721A (en) * | 2020-06-10 | 2020-11-03 | 广东芯盾微电子科技有限公司 | Cross-age face recognition method, system, device and storage medium |
| CN111881721B (en) * | 2020-06-10 | 2021-03-30 | 广东芯盾微电子科技有限公司 | Cross-age face recognition method, system, device and storage medium |
| CN111860335A (en) * | 2020-07-22 | 2020-10-30 | 安徽兰臣信息科技有限公司 | Intelligent wearable device based on face recognition |
| CN111860333A (en) * | 2020-07-22 | 2020-10-30 | 安徽兰臣信息科技有限公司 | Child face recognition algorithm for cross-age recognition |
| CN112101204A (en) * | 2020-09-14 | 2020-12-18 | 北京百度网讯科技有限公司 | Training method of generative countermeasure network, image processing method, device and equipment |
| CN112101204B (en) * | 2020-09-14 | 2024-01-23 | 北京百度网讯科技有限公司 | Training method, image processing method, device and equipment for generating type countermeasure network |
| CN112766169A (en) * | 2021-01-21 | 2021-05-07 | 中信银行股份有限公司 | Account opening identity authentication method, device, equipment and storage medium |
| CN112906525A (en) * | 2021-02-05 | 2021-06-04 | 广州市百果园信息技术有限公司 | Age identification method and device and electronic equipment |
| CN113076833A (en) * | 2021-03-25 | 2021-07-06 | 深圳数联天下智能科技有限公司 | Training method of age identification model, face age identification method and related device |
| CN113076833B (en) * | 2021-03-25 | 2024-05-31 | 深圳数联天下智能科技有限公司 | Training method of age identification model, face age identification method and related device |
| CN113221660A (en) * | 2021-04-14 | 2021-08-06 | 河海大学 | Cross-age face recognition method based on feature fusion |
| CN114582006A (en) * | 2022-05-06 | 2022-06-03 | 广东红橙云大数据有限公司 | Child age-crossing face recognition method and device, electronic equipment and medium |
| CN114582006B (en) * | 2022-05-06 | 2022-07-08 | 广东红橙云大数据有限公司 | Child age-crossing face recognition method and device, electronic equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111209878A (en) | Cross-age face recognition method and device | |
| CN111709409B (en) | Face living body detection method, device, equipment and medium | |
| CN111754596B (en) | Editing model generation method, device, equipment and medium for editing face image | |
| US12019718B2 (en) | Identity verification method and apparatus, computer device and storage medium | |
| US20180307947A1 (en) | Cyclic generative adversarial network for unsupervised cross-domain image generation | |
| CN111241291B (en) | Method and device for generating countermeasure sample by utilizing countermeasure generation network | |
| US11531876B2 (en) | Deep learning for characterizing unseen categories | |
| CN106022317A (en) | Face identification method and apparatus | |
| CN112188306B (en) | Label generation method, device, equipment and storage medium | |
| CN110348331A (en) | Face identification method and electronic equipment | |
| CN108509994A (en) | character image clustering method and device | |
| Li et al. | Image manipulation localization using attentional cross-domain CNN features | |
| CN113516142A (en) | Text image matching method, device, equipment and storage medium | |
| CN112101087A (en) | Facial image identity de-identification method and device and electronic equipment | |
| CN115761366A (en) | Zero sample picture classification method, system, device and medium for supplementing missing features | |
| CN115984930A (en) | Micro expression recognition method and device and micro expression recognition model training method | |
| CN116758379B (en) | Image processing method, device, equipment and storage medium | |
| CN109101984B (en) | Image identification method and device based on convolutional neural network | |
| CN116563604A (en) | End-to-end target detection model training, image target detection method and related equipment | |
| CN113947801B (en) | Face recognition method and device and electronic equipment | |
| CN113723111B (en) | Small sample intention recognition method, device, equipment and storage medium | |
| CN113379594A (en) | Face shape transformation model training, face shape transformation method and related device | |
| CN111461091B (en) | Universal fingerprint generation method and device, storage medium and electronic device | |
| CN115708135A (en) | Face recognition model processing method, face recognition method and device | |
| CN114462546A (en) | Feature distillation method, system, equipment and medium based on multi-model fusion |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200529 |
|
| RJ01 | Rejection of invention patent application after publication |