CN114373212A - Face recognition model construction method, face recognition method and related equipment - Google Patents
Face recognition model construction method, face recognition method and related equipment Download PDFInfo
- Publication number
- CN114373212A CN114373212A CN202210022384.1A CN202210022384A CN114373212A CN 114373212 A CN114373212 A CN 114373212A CN 202210022384 A CN202210022384 A CN 202210022384A CN 114373212 A CN114373212 A CN 114373212A
- Authority
- CN
- China
- Prior art keywords
- face
- cluster
- clustering
- data subset
- face recognition
- 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 53
- 238000010276 construction Methods 0.000 title claims abstract description 31
- 238000012549 training Methods 0.000 claims abstract description 76
- 238000012545 processing Methods 0.000 claims abstract description 54
- 238000003062 neural network model Methods 0.000 claims description 11
- 238000007781 pre-processing Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 8
- 238000012216 screening Methods 0.000 abstract description 2
- 238000000605 extraction Methods 0.000 description 8
- 238000013527 convolutional neural network Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 230000001815 facial effect Effects 0.000 description 4
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 210000004709 eyebrow Anatomy 0.000 description 1
- 210000000887 face Anatomy 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The application provides a face recognition model construction method, a face recognition method and related equipment, wherein the method comprises the following steps: the method comprises the steps of obtaining a training data set, wherein the training data set comprises a plurality of data subsets, each data subset comprises a plurality of face images belonging to the same category, performing cluster clustering processing on each face image included by each data subset to obtain a plurality of initial clusters corresponding to each data subset, screening out noise data, performing cluster clustering processing among the initial clusters to obtain a plurality of target clusters, merging and stripping the screened noise data, and accordingly correctly classifying all the face images.
Description
Technical Field
The present application relates to the field of face recognition technologies, and in particular, to a face recognition model construction method, a face recognition method, and a related device.
Background
With the development of face recognition technology, the face recognition human-computer interaction method has penetrated into various aspects of life, such as access control systems, face payment, smart communities, intelligent security, social entertainment and the like.
Currently, face recognition is implemented by a face recognition model, which is trained from a labeled training data set, however, there is often noisy data resulting from misclassification in the training data set.
In the prior art, noise data is removed through some models, namely training data with wrong classification are removed, but the reduction of the training data in a training data set is caused, and further the training effect of a face recognition model is poor.
Disclosure of Invention
The application provides a face recognition model construction method, a face recognition method and related equipment, and aims to solve the problem that noise data are removed to reduce training data and further cause poor training effect of a face recognition model.
In order to achieve the above object, the present application provides the following technical solutions:
the first aspect of the application discloses a face recognition model construction method, which comprises the following steps:
acquiring a training data set; the training data set comprises a plurality of data subsets, and each data subset comprises a plurality of face images belonging to the same category;
for each data subset, performing cluster-like clustering processing on each face image included in the data subset to obtain a plurality of initial cluster-like clusters corresponding to the data subset;
performing inter-class clustering processing on each initial class cluster to obtain a plurality of target class clusters;
and performing model training on the pre-constructed neural network model based on each target class cluster to obtain a face recognition model.
A second aspect of the present application discloses a face recognition method, including:
acquiring a face image to be recognized;
processing the face image to be recognized by utilizing a pre-constructed face recognition model to obtain a target face characteristic of the face image to be recognized; the face recognition model is obtained based on a target training data set, and the target training data set is obtained by performing cluster clustering processing and cluster clustering processing between clusters on each data subset included in the training data set;
similarity calculation is carried out on the target face features and face features stored in advance;
and taking the face information corresponding to the face features corresponding to the maximum similarity as the face recognition result of the face image to be recognized.
A third aspect of the present application discloses a face recognition model construction apparatus, including:
a first acquisition unit for acquiring a training data set; the training data set comprises a plurality of data subsets, and each data subset comprises a plurality of face images belonging to the same category;
the first clustering unit is used for carrying out cluster clustering on each face image included in each data subset aiming at each data subset to obtain a plurality of initial cluster clusters corresponding to the data subsets;
the second clustering unit is used for clustering between clusters of each initial cluster to obtain a plurality of target clusters;
and the training unit is used for carrying out model training on the pre-constructed neural network model based on each target class cluster to obtain the face recognition model.
The fourth aspect of the present application discloses a face recognition apparatus, comprising:
the second acquisition unit is used for acquiring a face image to be recognized;
the processing unit is used for processing the face image to be recognized by utilizing a pre-constructed face recognition model to obtain the target face characteristics of the face image to be recognized; the face recognition model is obtained based on a target training data set, and the target training data set is obtained by performing cluster clustering processing and cluster clustering processing between clusters on each data subset included in the training data set;
the calculating unit is used for calculating the similarity of the target face features and the face features stored in advance;
and the determining unit is used for taking the face information corresponding to the face features corresponding to the maximum similarity as the face recognition result of the face image to be recognized.
A third aspect of the present application discloses a storage medium storing a set of instructions, wherein the set of instructions, when executed by a processor, implement the face recognition model construction method as described above and the face recognition method as described above.
A fourth aspect of the present application discloses an electronic device, comprising:
a memory for storing at least one set of instructions;
and the processor is used for executing the instruction set stored in the memory, and implementing the face recognition model construction method as described above and the face recognition method as described above by executing the instruction set.
Compared with the prior art, the method has the following advantages:
the application provides a face recognition model construction method, a face recognition method and related equipment, wherein the method comprises the following steps: the method comprises the steps of obtaining a training data set, wherein the training data set comprises a plurality of data subsets, each data subset comprises a plurality of face images belonging to the same category, performing cluster clustering processing on each face image included by each data subset to obtain a plurality of initial clusters corresponding to each data subset, screening out noise data, performing cluster clustering processing among the initial clusters to obtain a plurality of target clusters, merging and stripping the screened noise data, and accordingly correctly classifying all the face images.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a flowchart of a method for constructing a face recognition model according to the present application;
fig. 2 is a flowchart of another method of a face recognition model construction method provided in the present application;
FIG. 3 is a flowchart of another method of a face recognition model construction method according to the present application;
FIG. 4 is a flowchart of another method of a face recognition model construction method according to the present application;
FIG. 5 is an exemplary diagram of a face recognition model construction method provided in the present application;
fig. 6 is a flowchart of a method of a face recognition method according to the present application;
fig. 7 is a schematic structural diagram of a face recognition model construction apparatus provided in the present application;
fig. 8 is a schematic structural diagram of a face recognition apparatus provided in the present application;
fig. 9 is a schematic structural diagram of an electronic device provided in the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. 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 application.
The term "include" and variations thereof as used herein are open-ended, i.e., "including but not limited to". The term "based on" is "based, at least in part, on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Relevant definitions for other terms will be given in the following description.
It should be noted that the terms "first", "second", and the like in the disclosure of the present application are only used for distinguishing different devices, modules or units, and are not used for limiting the order or interdependence relationship of the functions performed by the devices, modules or units.
It is noted that references to "a", "an", and "the" modifications in the disclosure herein are exemplary rather than limiting, and those skilled in the art will understand that "one or more" will be understood unless the context clearly dictates otherwise.
The application is operational with numerous general purpose or special purpose computing device environments or configurations. For example: personal computers, server computers, hand-held or portable devices, tablet-type devices, multi-processor apparatus, distributed computing environments that include any of the above devices or equipment, and the like.
The embodiment of the application provides a face recognition model construction method, which can be applied to a plurality of system platforms, wherein an execution subject of the method can be a computer terminal or a processor of various mobile devices, and a flow chart of the method is shown in fig. 1 and specifically comprises the following steps:
s101, acquiring a training data set.
In this embodiment, a training data set is obtained, where the training data set includes a plurality of data subsets, and each data subset includes a plurality of face images belonging to the same category.
It should be noted that different data subsets correspond to different categories.
And S102, performing cluster clustering on each face image included in the data subsets aiming at each data subset to obtain a plurality of initial cluster clusters corresponding to the data subsets.
In this embodiment, the similar clustering processing is performed on each face image included in each data subset to obtain a plurality of initial class clusters corresponding to each data subset, so that the noise data is screened out.
The inventor finds through research that the noise data is usually generated by misclassification, and the form of misclassification has two possibilities, namely that the noise data should belong to other data subsets except the current data subset, and should be screened out and correctly merged; the other is that the noise data is deployed in any data subset in the training data set, and should be screened out to form a new set.
Referring to fig. 2, the process of performing cluster-like clustering on each face image included in each data subset to obtain a plurality of initial cluster-like clusters corresponding to the data subsets specifically includes the following steps:
s201, extracting the features of each face image included in the data subset to obtain the face features of each face image.
In this embodiment, feature extraction is performed on each face image included in the data subset to obtain a face feature of each face image, specifically, feature extraction is performed on each face image included in the data subset by using a pre-trained feature extraction model, and optionally, the feature extraction model may be obtained by training a convolutional neural network model based on a classification loss function.
S202, based on the face features of each face image, performing cluster clustering processing on each face image in the data subset by using a first clustering algorithm to obtain a plurality of initial clusters corresponding to the data subset.
In this embodiment, Based on the facial features of each facial image, a first Clustering algorithm is used to perform cluster Clustering on each facial image included in the data subset, so as to obtain a plurality of initial clusters corresponding to the data subset, and optionally, the first Clustering algorithm may be a DBSCAN (Density-Based Spatial Clustering of Applications with Noise) Clustering algorithm.
In this embodiment, the original class is expanded into N classes by performing class clustering on each face image included in the data subset, where N is greater than or equal to 1.
In this embodiment, the initial cluster with the largest number of face images is determined from the plurality of initial clusters, the initial cluster with the largest number of face images is determined as the class to which the corresponding data subset belongs, and the remaining other initial clusters are determined as classes based on the respective initial clusters.
It should be noted that, of the multiple initial clusters corresponding to the data subsets, the initial cluster with the largest number of face images is the sample of the current class, where the current class is the class to which the corresponding data subset belongs, and the face images in other initial clusters are regarded as face images classified incorrectly in the data subsets, that is, noise data in the data subsets.
S103, clustering between the clusters of each initial cluster to obtain a plurality of target clusters.
In this embodiment, a plurality of target clusters are obtained by performing inter-cluster clustering on each initial cluster, and merging and stripping are performed on the filtered noise data, so that all face images are correctly classified.
Referring to fig. 3, the process of performing inter-class clustering on each initial class cluster to obtain a plurality of target class clusters specifically includes the following steps:
s301, aiming at each initial cluster, calculating the average face features of the initial clusters based on the face features of all face images in the initial clusters.
In this embodiment, the average face feature of each initial cluster is calculated based on the face features of each face image included in each initial cluster, specifically, for each initial cluster, the face features of each face image included in the initial cluster are summed to obtain a summation result, and the summation result is divided by the number of face images included in the initial cluster.
In this embodiment, the average face feature of each initial cluster is used as the feature representation of the initial cluster.
S302, based on the average face features of each initial cluster, clustering processing is carried out on each initial cluster by using a second clustering algorithm to obtain a plurality of target clusters.
In this embodiment, based on the average facial feature of each initial class cluster, a second clustering algorithm is used to perform inter-class clustering on each initial class cluster, so as to obtain a plurality of target class clusters, where the second clustering algorithm may be a DBSCAN clustering algorithm.
It should be noted that the neighborhood radius of the second clustering algorithm is larger than the neighborhood radius of the first clustering algorithm. That is to say, the clustering constraint of the second clustering algorithm is greater than that of the first clustering algorithm, so that the new classes separated by the intra-class clustering processing are prevented from being re-aggregated into the original classes in the inter-class clustering processing process, and a more accurate and credible classification result is obtained.
And S104, performing model training on the pre-constructed neural network model based on each target class cluster to obtain a face recognition model.
In this embodiment, model training is performed on a pre-constructed neural network model based on each target class cluster, where the neural network model may be a convolutional neural network, and specifically, based on each target class cluster, the convolutional neural network model is trained by using a classification loss function, and the trained convolutional neural network model is determined as a face recognition model.
The method for constructing the face recognition model provided by the embodiment of the application obtains a training data set, the training data set comprises a plurality of data subsets, each data subset comprises a plurality of face images belonging to the same category, the noise data is screened out by carrying out similar clustering processing on each face image included in each data subset to obtain a plurality of initial class clusters corresponding to each data subset, and a plurality of target clusters are obtained by clustering between clusters of each initial cluster, so that the selected noise data is merged and stripped, thereby realizing the correct classification of all face images and based on each correctly classified target cluster, model training is carried out on the pre-constructed neural network model to obtain the face recognition model, and the training effect of the face recognition model is improved under the condition that the training data volume is not reduced.
In the method for constructing a face recognition model provided in the embodiment of the present application, before step S102, the method may further include:
and carrying out image preprocessing on each face image included in each data subset.
In this embodiment, before performing cluster classification on each face image included in each data subset, image preprocessing may be performed on each face image included in each data subset, specifically, referring to fig. 4, a process of performing image preprocessing on each face image included in each data subset includes the following steps:
s401, carrying out face key point detection on each face image included in each data subset.
In this embodiment, the face keypoints of each face image included in each data subset are detected, so as to obtain the face keypoints included in each face image, where the face keypoints include, but are not limited to, eyebrows, eyes, a nose, a mouth, and a face contour.
S402, carrying out face alignment processing on each face image based on the result obtained by face key point detection.
In this embodiment, face alignment processing is performed on each face image based on a result obtained by detecting the face keypoints, specifically, for each face, the face is aligned by using rotation and/or scaling and/or translation operations according to the result obtained by detecting the face keypoints, and the face image is cut into a face image with a fixed size (e.g., 112 × 112).
In this embodiment, the face images of the fixed size processed by the respective face images are realized by performing image preprocessing on the respective face images included in each data subset.
Referring to fig. 5, an implementation process of the above-mentioned face recognition model construction method is illustrated as follows:
a training data set is obtained, the training data set comprises a plurality of data subsets, the data subsets comprise noise data, and each image in the data subsets carries tag data.
Stage 1, pre-learning.
The face feature extraction is performed on each face image in the data subset including the noise data, specifically, the face feature extraction is performed on each face image in the data subset by using a feature extraction model, wherein the feature extraction model is obtained by training a convolutional neural network model based on a classification loss function.
Stage 2, cleaning in class.
And aiming at each data subset, based on the face features of each face image included in the data subset, performing class aggregation on each face image in the data subset by using a DBSCAN clustering technology to realize class extension, wherein the face features of the face images are obtained by processing the face features based on a trained convolutional neural network.
Stage 3, inter-class merging.
And calculating the average characteristic of each class cluster obtained by clustering the classes to obtain the characteristic prototype vector of each class. And taking the feature prototype vector as the feature representation of the corresponding class cluster, constructing a feature prototype set based on each class cluster, and clustering the feature prototype set by using a DBSCAN clustering technology. In order to avoid the image stripped from the cluster is merged into the original cluster again, cluster constraint is strengthened at the stage, and cluster recovery is not achieved. By the clustering operation, the small data which possibly belong to the same type can be merged, and the inter-class merging is realized. And the clustering constraint is that the neighborhood radius of the clustering technology corresponding to the inter-class clustering is larger than the neighborhood radius of the clustering technology corresponding to the intra-class clustering.
And 4, final learning of characteristics.
And processing the convolutional neural network by utilizing a classification loss function based on each class cluster obtained by combining the classes to obtain a face recognition model.
Referring to fig. 6, an embodiment of the present application further provides a face recognition method, which specifically includes the following steps:
s601, obtaining a face image to be recognized.
S602, processing the face image to be recognized by using the pre-constructed face recognition model to obtain the target face characteristics of the face image to be recognized.
In this embodiment, a pre-constructed face recognition model is used to process a face image to be recognized, specifically, the face image to be recognized is input into the pre-constructed face recognition model, and the face image to be recognized is processed by the face recognition model, so as to obtain a target face feature of the face image to be recognized, where the face recognition model is obtained based on a target training data set, and the target training data set is obtained by performing cluster clustering processing and inter-cluster processing on each data subset included in the training data set, that is, the face recognition model is constructed by using the face recognition model construction method disclosed in any one of the above embodiments.
And S603, calculating the similarity of the target face features and the face features stored in advance.
In this embodiment, similarity calculation is performed on the target face features and the face features stored in advance, specifically, similarity calculation is performed on the target face features and each face feature stored in advance through a similarity calculation formula to obtain similarity corresponding to each face feature, where the similarity calculation formula includes, but is not limited to, a cosine similarity calculation formula.
And S604, taking the face information corresponding to the face features corresponding to the maximum similarity as a face recognition result of the face image to be recognized.
In this embodiment, the maximum similarity is determined from the similarities corresponding to each face feature, and the face information corresponding to the face feature corresponding to the maximum similarity is used as the face recognition result of the face image to be recognized.
According to the face recognition method provided by the embodiment of the application, the training effect of the constructed face recognition model is good, so that the face features extracted by the face recognition model are accurate, and the accuracy of face recognition is improved.
It should be noted that while operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous.
It should be understood that the various steps recited in the method embodiments disclosed herein may be performed in a different order and/or performed in parallel. Moreover, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the disclosure is not limited in this respect.
Corresponding to the method described in fig. 1, an embodiment of the present application further provides a face recognition model construction device, which is used for implementing the method in fig. 1 specifically, and a schematic structural diagram of the face recognition model construction device is shown in fig. 7, and specifically includes:
a first obtaining unit 701, configured to obtain a training data set; the training data set comprises a plurality of data subsets, and each data subset comprises a plurality of face images belonging to the same category;
a first clustering unit 702, configured to perform cluster-like clustering on each face image included in each data subset to obtain multiple initial cluster-like clusters corresponding to the data subset;
a second clustering unit 703, configured to perform inter-class clustering on each initial class cluster to obtain multiple target class clusters;
and the training unit 704 is used for performing model training on the pre-constructed neural network model based on each target class cluster to obtain a face recognition model.
The face recognition model construction device provided by the embodiment of the application acquires a training data set, wherein the training data set comprises a plurality of data subsets, each data subset comprises a plurality of face images belonging to the same category, the noise data is screened out by carrying out similar clustering processing on each face image included in each data subset to obtain a plurality of initial class clusters corresponding to each data subset, and a plurality of target clusters are obtained by clustering between clusters of each initial cluster, so that the selected noise data is merged and stripped, thereby realizing the correct classification of all face images and based on each correctly classified target cluster, model training is carried out on the pre-constructed neural network model to obtain the face recognition model, and the training effect of the face recognition model is improved under the condition that the training data volume is not reduced.
In an embodiment of the present application, based on the foregoing scheme, the first clustering unit 702 is specifically configured to:
extracting the features of each face image included in the data subset to obtain the face features of each face image;
and based on the face features of each face image, performing cluster clustering processing on each face image in the data subset by using a first clustering algorithm to obtain a plurality of initial clusters corresponding to the data subset.
In one embodiment of the present application, based on the foregoing scheme, the second classification unit 703 is specifically configured to:
aiming at each initial cluster, calculating the average face characteristics of the initial clusters based on the face characteristics of all face images in the initial clusters;
based on the average face features of each initial cluster, performing inter-cluster clustering processing on each initial cluster by using a second clustering algorithm to obtain a plurality of target clusters; wherein a neighborhood radius of the second clustering algorithm is larger than a neighborhood radius of the first clustering algorithm.
In an embodiment of the present application, based on the foregoing scheme, the method may further include:
and the preprocessing unit is used for preprocessing the images of the human faces included in each data subset.
In an embodiment of the present application, based on the foregoing scheme, the preprocessing unit is specifically configured to:
performing face key point detection on each face image included in each data subset;
and carrying out face alignment processing on each face image based on a result obtained by face key point detection.
Corresponding to the method described in fig. 6, an embodiment of the present application further provides a face recognition apparatus, which is used for specifically implementing the method in fig. 6, and a schematic structural diagram of the face recognition apparatus is shown in fig. 8, and specifically includes:
a second acquiring unit 801, configured to acquire a face image to be recognized;
the processing unit 802 is configured to process the face image to be recognized by using a pre-constructed face recognition model, so as to obtain a target face feature of the face image to be recognized; the face recognition model is obtained based on a target training data set, and the target training data set is obtained by performing cluster clustering processing and cluster clustering processing between clusters on each data subset included in the training data set;
a calculating unit 803, configured to perform similarity calculation between the target face feature and a face feature stored in advance;
the determining unit 804 is configured to use the face information corresponding to the face feature corresponding to the maximum similarity as a face recognition result of the face image to be recognized.
The face recognition device provided by the embodiment of the application has the advantages that the training effect of the constructed face recognition model is good, so that the face features extracted through the face recognition model are accurate, and the accuracy of face recognition is improved.
The embodiment of the application further provides a storage medium, wherein the storage medium stores an instruction set, and when the instruction set runs, the face recognition model construction method and the face recognition method disclosed in any one of the above embodiments are executed.
An electronic device is further provided in the embodiments of the present application, and a schematic structural diagram of the electronic device is shown in fig. 9, and specifically includes a memory 901 for storing at least one set of instruction sets; a processor 902, configured to execute the instruction set stored in the memory, and implement the face recognition model construction method and the face recognition method disclosed in any of the above embodiments by executing the instruction set.
In the detailed description section, this application will repeat, in part, all of the claims as issued:
according to one or more embodiments disclosed in the present application, fig. 1 provides a face recognition model construction method, including: acquiring a training data set; the training data set comprises a plurality of data subsets, and each data subset comprises a plurality of face images belonging to the same category; for each data subset, performing cluster-like clustering processing on each face image included in the data subset to obtain a plurality of initial cluster-like clusters corresponding to the data subset; performing inter-class clustering processing on each initial class cluster to obtain a plurality of target class clusters; and performing model training on the pre-constructed neural network model based on each target class cluster to obtain a face recognition model.
According to one or more embodiments disclosed in the present application, fig. 2 provides another face recognition model construction method, including: extracting the features of each face image included in the data subset to obtain the face features of each face image; and based on the face features of each face image, performing cluster clustering processing on each face image in the data subset by using a first clustering algorithm to obtain a plurality of initial clusters corresponding to the data subset.
According to one or more embodiments disclosed in the present application, fig. 3 provides another face recognition model construction method, including: aiming at each initial cluster, calculating the average face characteristics of the initial clusters based on the face characteristics of all face images in the initial clusters; based on the average face features of each initial cluster, performing inter-cluster clustering processing on each initial cluster by using a second clustering algorithm to obtain a plurality of target clusters; wherein a neighborhood radius of the second clustering algorithm is larger than a neighborhood radius of the first clustering algorithm.
According to one or more embodiments disclosed in the present application, fig. 4 provides another face recognition model construction method, including: and carrying out image preprocessing on each face image included in each data subset.
Performing face key point detection on each face image included in each data subset; and carrying out face alignment processing on each face image based on a result obtained by face key point detection.
According to one or more embodiments disclosed in the present application, fig. 6 provides a face recognition method, including: acquiring a face image to be recognized; processing the face image to be recognized by utilizing a pre-constructed face recognition model to obtain a target face characteristic of the face image to be recognized; the face recognition model is obtained based on a target training data set, and the target training data set is obtained by performing cluster clustering processing and cluster clustering processing between clusters on each data subset included in the training data set; similarity calculation is carried out on the target face features and face features stored in advance; and taking the face information corresponding to the face features corresponding to the maximum similarity as the face recognition result of the face image to be recognized.
Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.
While several specific implementation details are included in the above discussion, these should not be construed as limitations on the scope of the disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination.
The foregoing description is only exemplary of the preferred embodiments disclosed herein and is illustrative of the principles of the technology employed. It will be appreciated by those skilled in the art that the scope of the disclosure herein is not limited to the particular combination of features described above, but also encompasses other arrangements formed by any combination of the above features or their equivalents without departing from the spirit of the disclosure. For example, the above features and (but not limited to) technical features having similar functions disclosed in the present disclosure are mutually replaced to form the technical solution.
Claims (10)
1. A face recognition model construction method is characterized by comprising the following steps:
acquiring a training data set; the training data set comprises a plurality of data subsets, and each data subset comprises a plurality of face images belonging to the same category;
for each data subset, performing cluster-like clustering processing on each face image included in the data subset to obtain a plurality of initial cluster-like clusters corresponding to the data subset;
performing inter-class clustering processing on each initial class cluster to obtain a plurality of target class clusters;
and performing model training on the pre-constructed neural network model based on each target class cluster to obtain a face recognition model.
2. The method according to claim 1, wherein the performing cluster-like clustering on each face image included in the data subset to obtain a plurality of initial cluster-like clusters corresponding to the data subset comprises:
extracting the features of each face image included in the data subset to obtain the face features of each face image;
and based on the face features of each face image, performing cluster clustering processing on each face image in the data subset by using a first clustering algorithm to obtain a plurality of initial clusters corresponding to the data subset.
3. The method according to claim 2, wherein the inter-class clustering of each initial class cluster to obtain a plurality of target class clusters comprises:
aiming at each initial cluster, calculating the average face characteristics of the initial clusters based on the face characteristics of all face images in the initial clusters;
based on the average face features of each initial cluster, performing inter-cluster clustering processing on each initial cluster by using a second clustering algorithm to obtain a plurality of target clusters; wherein a neighborhood radius of the second clustering algorithm is larger than a neighborhood radius of the first clustering algorithm.
4. The method according to any one of claims 1 to 3, wherein before performing cluster-like clustering on each face image included in each data subset to obtain a plurality of initial cluster-like clusters corresponding to the data subset, the method further comprises:
and carrying out image preprocessing on each face image included in each data subset.
5. The method of claim 4, wherein the image preprocessing of the respective face images included in each data subset comprises:
performing face key point detection on each face image included in each data subset;
and carrying out face alignment processing on each face image based on a result obtained by face key point detection.
6. A face recognition method, comprising:
acquiring a face image to be recognized;
processing the face image to be recognized by utilizing a pre-constructed face recognition model to obtain a target face characteristic of the face image to be recognized; the face recognition model is obtained based on a target training data set, and the target training data set is obtained by performing cluster clustering processing and cluster clustering processing between clusters on each data subset included in the training data set;
similarity calculation is carried out on the target face features and face features stored in advance;
and taking the face information corresponding to the face features corresponding to the maximum similarity as the face recognition result of the face image to be recognized.
7. A face recognition model construction device is characterized by comprising:
a first acquisition unit for acquiring a training data set; the training data set comprises a plurality of data subsets, and each data subset comprises a plurality of face images belonging to the same category;
the first clustering unit is used for carrying out cluster clustering on each face image included in each data subset aiming at each data subset to obtain a plurality of initial cluster clusters corresponding to the data subsets;
the second clustering unit is used for clustering between clusters of each initial cluster to obtain a plurality of target clusters;
and the training unit is used for carrying out model training on the pre-constructed neural network model based on each target class cluster to obtain the face recognition model.
8. A face recognition apparatus, comprising:
the second acquisition unit is used for acquiring a face image to be recognized;
the processing unit is used for processing the face image to be recognized by utilizing a pre-constructed face recognition model to obtain the target face characteristics of the face image to be recognized; the face recognition model is obtained based on a target training data set, and the target training data set is obtained by performing cluster clustering processing and cluster clustering processing between clusters on each data subset included in the training data set;
the calculating unit is used for calculating the similarity of the target face features and the face features stored in advance;
and the determining unit is used for taking the face information corresponding to the face features corresponding to the maximum similarity as the face recognition result of the face image to be recognized.
9. A storage medium storing a set of instructions, wherein the set of instructions, when executed by a processor, implement the face recognition model construction method according to any one of claims 1 to 5 and the face recognition method according to claim 6.
10. An electronic device, comprising:
a memory for storing at least one set of instructions;
a processor for executing the instruction set stored in the memory, and implementing the face recognition model construction method according to any one of claims 1 to 5 and the face recognition method according to claim 6 by executing the instruction set.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210022384.1A CN114373212A (en) | 2022-01-10 | 2022-01-10 | Face recognition model construction method, face recognition method and related equipment |
| PCT/CN2022/087635 WO2023130613A1 (en) | 2022-01-10 | 2022-04-19 | Facial recognition model construction method, facial recognition method, and related device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210022384.1A CN114373212A (en) | 2022-01-10 | 2022-01-10 | Face recognition model construction method, face recognition method and related equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114373212A true CN114373212A (en) | 2022-04-19 |
Family
ID=81144810
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210022384.1A Pending CN114373212A (en) | 2022-01-10 | 2022-01-10 | Face recognition model construction method, face recognition method and related equipment |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN114373212A (en) |
| WO (1) | WO2023130613A1 (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117079379B (en) * | 2023-10-16 | 2024-01-09 | 八维通科技有限公司 | High-tolerance face passing method and system for rail transit and computer program medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106778684A (en) * | 2017-01-12 | 2017-05-31 | 易视腾科技股份有限公司 | deep neural network training method and face identification method |
| CN109447186A (en) * | 2018-12-13 | 2019-03-08 | 深圳云天励飞技术有限公司 | Clustering method and Related product |
| CN110287835A (en) * | 2019-06-14 | 2019-09-27 | 南京云创大数据科技股份有限公司 | A kind of Asia face database Intelligent Establishment method |
| CN112446361A (en) * | 2020-12-16 | 2021-03-05 | 上海芯翌智能科技有限公司 | Method and equipment for cleaning training data |
| WO2021082426A1 (en) * | 2019-10-29 | 2021-05-06 | 平安科技（深圳）有限公司 | Human face clustering method and apparatus, computer device, and storage medium |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112232117A (en) * | 2020-09-08 | 2021-01-15 | 深圳微步信息股份有限公司 | Face recognition method, face recognition device and storage medium |
| CN112818885B (en) * | 2021-02-07 | 2024-05-28 | 中国银联股份有限公司 | Face recognition method, device, equipment and storage medium |
| CN113239907B (en) * | 2021-07-12 | 2021-12-14 | 北京远鉴信息技术有限公司 | Face recognition detection method and device, electronic equipment and storage medium |
-
2022
- 2022-01-10 CN CN202210022384.1A patent/CN114373212A/en active Pending
- 2022-04-19 WO PCT/CN2022/087635 patent/WO2023130613A1/en unknown
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106778684A (en) * | 2017-01-12 | 2017-05-31 | 易视腾科技股份有限公司 | deep neural network training method and face identification method |
| CN109447186A (en) * | 2018-12-13 | 2019-03-08 | 深圳云天励飞技术有限公司 | Clustering method and Related product |
| CN110287835A (en) * | 2019-06-14 | 2019-09-27 | 南京云创大数据科技股份有限公司 | A kind of Asia face database Intelligent Establishment method |
| WO2021082426A1 (en) * | 2019-10-29 | 2021-05-06 | 平安科技（深圳）有限公司 | Human face clustering method and apparatus, computer device, and storage medium |
| CN112446361A (en) * | 2020-12-16 | 2021-03-05 | 上海芯翌智能科技有限公司 | Method and equipment for cleaning training data |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2023130613A1 (en) | 2023-07-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Goodfellow et al. | Multi-digit number recognition from street view imagery using deep convolutional neural networks | |
| CN109583332B (en) | Face recognition method, face recognition system, medium, and electronic device | |
| Siddiqi et al. | Facial expression recognition using active contour-based face detection, facial movement-based feature extraction, and non-linear feature selection | |
| CN111931592B (en) | Object recognition method, device and storage medium | |
| US9460398B2 (en) | Apparatus and method for recognizing user activity | |
| CN111639535A (en) | Face recognition method and device based on deep learning | |
| CN111144348A (en) | Image processing method, image processing device, electronic equipment and storage medium | |
| EP4099217A1 (en) | Image processing model training method and apparatus, device, and storage medium | |
| Shirazi et al. | Clustering on Grassmann manifolds via kernel embedding with application to action analysis | |
| Dubey et al. | A review of face recognition methods using deep learning network | |
| Chanti et al. | Improving bag-of-visual-words towards effective facial expressive image classification | |
| US10423817B2 (en) | Latent fingerprint ridge flow map improvement | |
| CN113255557B (en) | Deep learning-based video crowd emotion analysis method and system | |
| CN110008365B (en) | Image processing method, device and equipment and readable storage medium | |
| Pal et al. | Off-line signature verification systems: a survey | |
| CN111898550A (en) | Method and device for establishing expression recognition model, computer equipment and storage medium | |
| CN112966626A (en) | Face recognition method and device | |
| Arlot et al. | A kernel multiple change-point algorithm via model selection | |
| CN110414622B (en) | Classifier training method and device based on semi-supervised learning | |
| Lahiani et al. | Hand pose estimation system based on Viola-Jones algorithm for android devices | |
| Qin et al. | Finger-vein quality assessment based on deep features from grayscale and binary images | |
| CN111178254A (en) | Signature identification method and device | |
| CN114373212A (en) | Face recognition model construction method, face recognition method and related equipment | |
| US9454706B1 (en) | Arabic like online alphanumeric character recognition system and method using automatic fuzzy modeling | |
| Nagendraswamy et al. | LBPV for recognition of sign language at sentence level: An approach based on symbolic representation |
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 |