CN112418067A - Simple and convenient face recognition online learning method based on deep learning model - Google Patents
Simple and convenient face recognition online learning method based on deep learning model Download PDFInfo
- Publication number
- CN112418067A CN112418067A CN202011306290.4A CN202011306290A CN112418067A CN 112418067 A CN112418067 A CN 112418067A CN 202011306290 A CN202011306290 A CN 202011306290A CN 112418067 A CN112418067 A CN 112418067A
- Authority
- CN
- China
- Prior art keywords
- local
- face feature
- face
- feature vector
- deep learning
- 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 40
- 238000013136 deep learning model Methods 0.000 title claims abstract description 36
- 239000013598 vector Substances 0.000 claims abstract description 127
- 238000001514 detection method Methods 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 4
- 238000012217 deletion Methods 0.000 abstract description 4
- 230000037430 deletion Effects 0.000 abstract description 4
- 230000000694 effects Effects 0.000 abstract description 4
- 230000001815 facial effect Effects 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 210000000887 face Anatomy 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 210000000554 iris Anatomy 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005070 sampling 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
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention is suitable for the technical field of computer vision, and provides a simple and convenient human face recognition online learning method based on a deep learning model, which uses files such as hdf5 and the like which can be added with data types to store human face characteristic vectors and identity label information, the human face characteristic vectors and corresponding identity labels are added into a local human face characteristic vector library by the same index, the addition and deletion of the members are more convenient, the trouble of re-collecting the sample and retraining by the traditional method is solved, the method has the advantages of simple operation, easy operation and obvious effect, the local face feature vector library is updated without extracting the face features of all local members again, only the face features of newly added personnel are extracted, the face feature vectors and the identity label information of the newly added members are added to the local face feature vector library, and the local face feature vector library is greatly convenient to update.
Description
Technical Field
The invention belongs to the technical field of computer vision, and particularly relates to a simple and convenient face recognition online learning method based on a deep learning model.
Background
Face recognition refers to a biometric technique that uniquely identifies or verifies a person by comparing and analyzing patterns based on the person's facial profile. After decades of development, the face recognition technology is mature day by day, and the application range is wide day by day, however, because the face is different from the biological characteristics of fingerprints, irises and the like, the face recognition technology has various changes and is influenced by illumination, posture and shielding, and the face recognition accuracy is difficult to guarantee. In order to increase the robustness of the algorithm, the operation of updating the model is necessary. However, for the conventional method, model updating means that the sampling is repeated and then the model is trained, which is very troublesome and still has the defect of poor adaptability. Therefore, an online learning method capable of realizing face recognition is highly required.
In recent years, with the application of convolutional neural networks in the field of image detection and classification, the detection and identification effects are greatly improved in the field of computer vision. At present, a convolutional neural network algorithm taking deep learning as a backbone becomes a mainstream algorithm in the field of computer vision. The human face recognition is a common computer vision problem, the neural network is adopted as a human face recognition algorithm, compared with the manually designed features, the neural network has very obvious advantages in recognition accuracy, and is different from the manually designed shallow features. However, the training of the model takes a long time, the trained neural network structure is fixed, the model needs to be trained again when the model is changed, and the retrained model needs to be introduced into the face recognition device, so that the addition and deletion of the members are troublesome, and great inconvenience is brought to the user. Therefore, a simple, convenient and accurate face recognition online learning method is urgently needed.
Disclosure of Invention
In view of the above problems, the present invention aims to provide a simple and convenient online learning method for face recognition based on a deep learning model, and aims to solve the technical problem that the existing face recognition technology needs to repeatedly train the model when an error case occurs.
The simple and convenient face recognition online learning method based on the deep learning model comprises the following steps:
training a deep learning model by using the public face recognition data set so as to enable the deep learning model to realize face feature extraction;
collecting a local member photo to establish a local face data set;
carrying out face detection and feature extraction on the photos in the local face data set by using the trained deep learning model, and storing the face feature vectors and corresponding identity label information by using files capable of adding data types according to the detection sequence for the extracted face feature vectors to complete the establishment of a local face feature vector library;
inputting the picture to be detected into a trained deep learning model, and obtaining a face feature vector of the picture to be detected after face detection and face feature extraction are completed;
comparing the face feature vector of the photo to be detected with the face feature vector in the local face feature vector library, and outputting the identity label information corresponding to the closest face feature vector under the condition of meeting a comparison threshold value to finish the face identification process of the photo to be detected;
if the identification is correct, detecting the next photo to be detected;
and if the identification is wrong, directly inputting correct identity tag information, and adding the face feature vector of the photo to be detected and the input correct identity information to a local face feature vector library by using the same index.
Furthermore, the output identity label information is judged to be correct, and then the current identification is determined to be correct.
Furthermore, there are two cases of identification errors, namely, the currently output identity tag information is the identity information of another local member, that is, the to-be-detected picture of the local member is wrongly identified as another local member; and secondly, the photos to be detected of the local members do not meet the threshold condition after comparison and are identified as unknown, namely, the photos to be detected of the local members are identified as unknown by mistake.
Further, the method further comprises adding and deleting local members of the local face feature vector library, and the specific process is as follows:
when a local member is added, only identity input is carried out on the added member, and the extracted face feature vector and the corresponding identity label information are added to a local face feature vector library according to the same index;
and when the local member is deleted, deleting all the identity tags and the corresponding face feature vectors of the local member according to the index.
Further, the file of the appendable data type is an hdf5 file.
Furthermore, the similarity between the human faces is determined by comparing the Euclidean distance between the human face feature vector of the photo to be detected and the human face feature vector in the local human face feature library, and the set similarity threshold is the comparison threshold condition.
The invention has the beneficial effects that:
(1) the invention carries out face recognition based on the deep learning model, uses the deep learning neural network and uses a large amount of public data sets, so that the face feature extractor has better robustness and good face detection and identity authentication effects.
(2) The method stores the face feature vector and the identity tag information by using the file capable of adding the data type, and adds the face feature vector and the corresponding identity tag to a local face feature vector library by using the same index, so that addition and deletion of members are simpler and more convenient;
(3) the invention provides a simple and effective face recognition online learning method, solves the trouble of retraining the sample re-acquisition by the traditional method, and has the advantages of simple operation, easy operation and obvious effect; meanwhile, a simple and convenient updating method of the local face feature vector library is also provided, the local face feature vector library is updated without extracting the face features of all local members again, only by extracting the face features of newly added personnel, the face feature vectors and the identity label information of the newly added members are added to the local face feature vector library, and the local face feature vector library is greatly convenient to update.
Drawings
Fig. 1 is a flowchart of a simple face recognition online learning method based on a deep learning model according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In order to explain the technical means of the present invention, the following description will be given by way of specific examples.
Fig. 1 shows a flow of the simple and convenient face recognition online learning method based on the deep learning model according to the embodiment of the present invention, and only the parts related to the embodiment of the present invention are shown for convenience of description.
As shown in fig. 1, the simple and convenient online learning method based on deep learning model in this embodiment includes the following steps:
and step S1, training the deep learning model by using the public face recognition data set so that the deep learning model can realize face feature extraction.
In the step, a deep learning model is trained by using a network public data set, so that the deep learning model can realize the function of extracting the human face features, and a human face feature extractor is established. In this embodiment, MTCNN and Facenet deep learning models are used, face detection is completed through a trained MTCNN model, a detected face image is introduced into the trained Facenet model, the face image is mapped to a multidimensional space, and the Facenet model output is an extracted face feature vector, so that a face feature extraction function is realized.
Of course, other deep learning models may be used, not to mention them.
And step S2, collecting the local member photos to establish a local face data set.
The method comprises the steps of collecting photos of local members, including but not limited to identification photos, transferring one photo of the local member from each folder, placing the photos with different identities in different folders, naming each folder by the correct name of the local member under the folder, and finally placing all the folders under a total folder to complete the establishment of a local face data set.
And step S3, carrying out face detection and feature extraction on the photos in the local face data set by using the trained deep learning model, and storing the face feature vectors and corresponding identity label information by using files capable of adding data types according to the detection sequence for the extracted face feature vectors to complete the establishment of the local face feature vector library.
The establishment of the local face feature vector library is an important component for realizing simple and convenient face recognition online learning, the step of establishing the local face feature vector library converts the updating of the deep learning model into the updating of the local face feature vector library, and the updating of the local face feature vector library is realized through files with data types capable of being added. The establishment of the local face feature vector library can be divided into two steps: extracting the face feature vector and storing the face feature vector.
Extracting the face feature vector: the method comprises the steps of utilizing a previously trained face feature extractor, namely a deep learning model, namely a trained MTCNN (multiple-terminal connected computing network) and Facenet model, sequentially carrying out face detection and face feature extraction on photos in a local face data set, and sequentially outputting face feature vectors of local members by the deep learning model according to a detection sequence to finish extraction work of the face feature vectors.
In a specific implementation, it is assumed that the local member is composed of m persons, and the composition set P ═ P1，p2，…，pm}。
Face detection and face detection are performed on the identification photographs of each individual in the set P using trained face feature extractors, namely trained MTCNN and Facenet modelsAnd (5) feature extraction work. For a single individual, the facial feature vector extracted by the Facenet model is yi(i is more than or equal to 1 and less than or equal to m), the shape is (1, n), and n represents the dimension of the face feature vector; for the entire local face data set, there will be m n-dimensional face feature vectors.
In order to more conveniently add or delete the extracted facial feature vector, the extracted facial feature vector is converted into an array object a by using an asarray function in Numpy, the array object is in the shape of (m, 1, n), where m is the number of local members, and in particular, m is a]Because each local member under the folder of the local face data set only uses one certificate photo, the identity of the local member can be determined through the value of m; n represents the dimension of the extracted face feature vector, thereby forming a local face feature vector library S ═ y1，y2，…，ym}。
For the storage of face feature vectors: after the extraction of the face feature vector is completed, the extracted face feature vector needs to be stored. The size of the face feature vector library is determined by the number of local members, in practical use, files with large storage capacity are required to be used for ensuring long-term use, when a large number of data sets are stored, a file format aiming at processing the data sets is preferably used, and the embodiment selects files with an addable data type, such as hdf5 files.
In a specific implementation, the local facial feature vector library is updated by a file capable of adding data types, and in this embodiment, the hdf5 file is used to store the extracted facial feature vectors. An hdf5 file is a container that holds two types of objects: dataset and group. Dataset is a data set like an array, while group is a container like a folder, holding both Dataset and other groups.
The mode of the hdf5 file is set to "a", i.e., an addition mode, to ensure the data addition capability. Before storing the face feature vector, firstly initializing a dataset, and specifying the name and the shape of the dataset, wherein the name is equivalent to a key value in a dictionary, and a data set under the name can be found by specifying the name; the shape is initialized to (0, 1, n), n being the extracted face feature vector dimension. After the initialization operation is completed, the shape of the dataset is converted into (m, 1, n) by adopting a resize function, and then the extracted face feature vector is added into the dataset, so that the storage of the face feature vector is completed.
Since dataset is an array-like data set, it means that the identity of the local member can also be determined from dataset. shape [0 ].
In addition, the change situation of the local member is also considered, and in real life, the change situation of the local member is really existed, which also requires that the local face feature vector library can realize the adding and deleting operations of the local member face feature vector, which means that the file storing the face feature vector is a file with an addable data type. When the files with data types capable of being added, such as files like hdf5, are used and the face feature vectors are stored, only the members needing to be added or deleted are processed, and all the members do not need to be operated again, so that the time for adding and deleting local member information is reduced, the operation efficiency is greatly improved, and meanwhile, a simple and convenient method is provided for the subsequent face recognition online learning.
Therefore, in the embodiment, the files with data types capable of being added, such as hdf5, are used in the storage mode of the face feature vector, so that the requirement of large data storage is met, and the local face feature vector library is more convenient to update.
The method changes the updating of the deep learning model in the face recognition into the updating of the local face feature vector library, and solves the technical problem that the existing face recognition technology needs to repeatedly train the model when an error case occurs; the updating of the local face feature vector library is realized by files with data types capable of being added, the updating of the local face feature vector library is more convenient due to the data addability, the efficiency is greatly improved, and the method conforms to the conditions of practical application.
And step S4, inputting the picture to be detected into the trained deep learning model, and obtaining the face feature vector of the picture to be detected after finishing face detection and face feature extraction.
And importing the established local face feature vector library to serve as a reference feature vector to form a reference feature vector space for subsequent comparison and calculation. And loading the hdf5 file for storing the face feature vector by using the read-only model 'r', and importing the file into a local face feature vector library S as a reference feature vector space. In this embodiment, the photo to be detected is input into the trained MTCNN and Facenet models, and after face detection and face feature extraction are completed, the face feature vector of the photo to be detected is obtained.
And step S5, comparing the face feature vector of the photo to be detected with the face feature vector in the local face feature vector library, and outputting the identity label information corresponding to the closest face feature vector under the condition of meeting the comparison threshold value, thereby completing the face recognition process of the photo to be detected.
The face feature vector of the photo to be detected is obtained as
Comparing the face feature vector with the face feature vector in the local face feature vector library, in this embodiment, comparing by calculating the euclidean distance, and calculating the euclidean distance between the face feature vector and the face feature vector in the local face feature vector library, where the smaller the similarity between two face feature vectors is, the smaller the difference between the two face feature vectors is, and the larger the similarity of the face is. And under the condition of meeting the threshold, namely the similarity is smaller than the threshold, outputting the identity label information closest to the local face feature vector library, and finishing the face recognition process of the picture to be detected.
And step S6, if the identification is correct, detecting the next photo to be detected.
And if the identification is correct, judging whether the identity label information output by the model is correct, namely judging whether the identity label information is consistent with the current photo to be detected. And if the two pictures are consistent, detecting the next picture to be detected. If not, the identification is wrong.
And step S7, if the identification is wrong, directly inputting correct identity label information, and adding the face feature vector of the photo to be detected and the input correct identity information to a local face feature vector library by the same index.
There are two cases of identification error, one is that the currently output identity tag information is the identity information of another local member, that is, the case of identifying the to-be-detected photo of the local member as another local member by error under the identification error condition in the step S6; and secondly, the photos to be detected of the local members do not meet the threshold condition after comparison and are identified as unknown, namely, the photos to be detected of the local members are identified as unknown by mistake.
The subsequent processing steps of the recognition error are that the face feature vector extracted from the photo to be detected and the manually input correct identity tag information are added to the hdf5 file storing the local face feature vector, that is, the face feature vector extracted from the photo to be detected and the input correct identity tag are respectively added to the hdf5 file by the same index. The photo is again identified, so that the correct identity tag is output. And finishing the online learning process of face recognition.
Meanwhile, the invention provides a simple and convenient updating mode of the local face vector library, which comprises the steps of adding and deleting local members of the local face feature vector library, and the specific process is as follows: when a local member is added, only identity input is carried out on the added member, and the extracted face feature vector and the corresponding identity label information are added to a local face feature vector library according to the same index; and when the local member is deleted, deleting all the identity tags and the corresponding face feature vectors of the local member according to the index. In a specific implementation, the addition of the data means that a dataset is established before, and the face feature vector stored under the name can be found by specifying the name. The local face feature vector array object is a, m is a.shape [0], the face feature vector array object to be added is B, z is m + b.shape [0], the shape of the dataset is converted from (m, 1, n) to (z, 1, n) by using a resize function, and then the face feature vector to be added is added to the dataset [ m: z ], addition of data is completed.
When a local member leaves, all information of the local member needs to be deleted. Because the identity tag and the face feature vector are both stored in the array, the data deletion is completed by inquiring all indexes of the member identity tag in the identity tag array and deleting the face feature vector in the dataset through the indexes.
The face feature vector is stored by using files such as hdf5 and the like which can be added, so that face features do not need to be extracted repeatedly for all members, only face features need to be extracted for newly added members, and the method is suitable for updating a local face feature vector library which has more local members and large personnel mobility.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (6)
1. A simple and convenient face recognition online learning method based on a deep learning model is characterized by comprising the following steps:
training a deep learning model by using the public face recognition data set so as to enable the deep learning model to realize face feature extraction;
collecting a local member photo to establish a local face data set;
carrying out face detection and feature extraction on the photos in the local face data set by using the trained deep learning model, and storing the face feature vectors and corresponding identity label information by using files capable of adding data types according to the detection sequence for the extracted face feature vectors to complete the establishment of a local face feature vector library;
inputting the picture to be detected into a trained deep learning model, and obtaining a face feature vector of the picture to be detected after face detection and face feature extraction are completed;
comparing the face feature vector of the photo to be detected with the face feature vector in the local face feature vector library, and outputting the identity label information corresponding to the closest face feature vector under the condition of meeting a comparison threshold value to finish the face identification process of the photo to be detected;
if the identification is correct, detecting the next photo to be detected;
and if the identification is wrong, directly inputting correct identity tag information, and adding the face feature vector of the photo to be detected and the input correct identity information into a local face feature vector library by using the same index.
2. The simple and convenient online learning method for face recognition based on the deep learning model as claimed in claim 1, wherein the output identity tag information is judged to be correct, and then the current recognition is determined to be correct.
3. The simple and convenient online learning method for face recognition based on the deep learning model as claimed in claim 1, wherein there are two cases of recognition error, one is that the currently output identity tag information is the identity information of another local member, i.e. the picture to be tested of the local member is wrongly recognized as another local member; and secondly, the photos to be detected of the local members do not meet the threshold condition after comparison and are identified as unknown, namely, the photos to be detected of the local members are identified as unknown by mistake.
4. The simple online learning method for face recognition based on deep learning model as claimed in any one of claims 1-3, wherein the method further comprises adding and deleting local members of local face feature vector library, and the specific process is as follows:
when a local member is added, only identity input is carried out on the added member, and the extracted face feature vector and the corresponding identity label information are added to a local face feature vector library according to the same index;
and when the local member is deleted, deleting all the identity tags and the corresponding face feature vectors of the local member according to the index.
5. The easy human face recognition online learning method based on the deep learning model as claimed in claim 4, wherein the file of the appendable data type is hdf5 file.
6. The simple and convenient online learning method for face recognition based on the deep learning model as claimed in claim 5, wherein the similarity between faces is determined by comparing the Euclidean distance between the face feature vector of the photo to be tested and the face feature vector in the local face feature library, and the set similarity threshold is the comparison threshold condition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011306290.4A CN112418067A (en) | 2020-11-20 | 2020-11-20 | Simple and convenient face recognition online learning method based on deep learning model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011306290.4A CN112418067A (en) | 2020-11-20 | 2020-11-20 | Simple and convenient face recognition online learning method based on deep learning model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112418067A true CN112418067A (en) | 2021-02-26 |
Family
ID=74774697
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011306290.4A Pending CN112418067A (en) | 2020-11-20 | 2020-11-20 | Simple and convenient face recognition online learning method based on deep learning model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112418067A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113486745A (en) * | 2021-06-24 | 2021-10-08 | 哈尔滨工业大学（深圳） | Microscopic structure image recognition method and device based on self-supervision neural network |
| CN116129227A (en) * | 2023-04-12 | 2023-05-16 | 合肥的卢深视科技有限公司 | Model training method, device, electronic equipment and computer readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446597A (en) * | 2016-09-06 | 2017-02-22 | 清华大学 | Multi-species feature selection and unknown gene identification methods |
| CN109145717A (en) * | 2018-06-30 | 2019-01-04 | 东南大学 | A kind of face identification method of on-line study |
-
2020
- 2020-11-20 CN CN202011306290.4A patent/CN112418067A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446597A (en) * | 2016-09-06 | 2017-02-22 | 清华大学 | Multi-species feature selection and unknown gene identification methods |
| CN109145717A (en) * | 2018-06-30 | 2019-01-04 | 东南大学 | A kind of face identification method of on-line study |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113486745A (en) * | 2021-06-24 | 2021-10-08 | 哈尔滨工业大学（深圳） | Microscopic structure image recognition method and device based on self-supervision neural network |
| CN116129227A (en) * | 2023-04-12 | 2023-05-16 | 合肥的卢深视科技有限公司 | Model training method, device, electronic equipment and computer readable storage medium |
| CN116129227B (en) * | 2023-04-12 | 2023-09-01 | 合肥的卢深视科技有限公司 | Model training method, device, electronic equipment and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111126360B (en) | Cross-domain pedestrian re-identification method based on unsupervised combined multi-loss model | |
| CN109800648B (en) | Face detection and recognition method and device based on face key point correction | |
| CN109993236B (en) | One-shot Simese convolutional neural network-based small-sample Manchu matching method | |
| CN109145717B (en) | Face recognition method for online learning | |
| CN109241317B (en) | Pedestrian Hash retrieval method based on measurement loss in deep learning network | |
| CN110263659B (en) | Finger vein recognition method and system based on triplet loss and lightweight network | |
| Zhao et al. | Unconstrained face recognition using a set-to-set distance measure on deep learned features | |
| CN106845358B (en) | Method and system for recognizing image features of handwritten characters | |
| WO2020164278A1 (en) | Image processing method and device, electronic equipment and readable storage medium | |
| Liu et al. | Towards optimal binary code learning via ordinal embedding | |
| CN110942091A (en) | Semi-supervised few-sample image classification method for searching reliable abnormal data center | |
| CN112633382A (en) | Mutual-neighbor-based few-sample image classification method and system | |
| US10423817B2 (en) | Latent fingerprint ridge flow map improvement | |
| CN112418067A (en) | Simple and convenient face recognition online learning method based on deep learning model | |
| CN106575280A (en) | System and methods for analysis of user-associated images to generate non-user generated labels and utilization of the generated labels | |
| CN110516533A (en) | A kind of pedestrian based on depth measure discrimination method again | |
| CN109670423A (en) | A kind of image identification system based on deep learning, method and medium | |
| CN114565807A (en) | Method and device for training target image retrieval model | |
| CN109886065A (en) | A kind of online increment type winding detection method | |
| CN100365645C (en) | Identity recognition method based on eyebrow recognition | |
| CN114579794A (en) | Multi-scale fusion landmark image retrieval method and system based on feature consistency suggestion | |
| Paul et al. | Mining images for image annotation using SURF detection technique | |
| CN117609583A (en) | Customs import and export commodity classification method based on image text combination retrieval | |
| CN108537855B (en) | Ceramic stained paper pattern generation method and device with consistent sketch | |
| Leng | Co‐Metric Learning for Person Re‐Identification |
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: 20210226 |
|
| RJ01 | Rejection of invention patent application after publication |