CN116912925A - Face recognition method, device, electronic equipment and medium - Google Patents
Face recognition method, device, electronic equipment and medium Download PDFInfo
- Publication number
- CN116912925A CN116912925A CN202311182465.9A CN202311182465A CN116912925A CN 116912925 A CN116912925 A CN 116912925A CN 202311182465 A CN202311182465 A CN 202311182465A CN 116912925 A CN116912925 A CN 116912925A
- Authority
- CN
- China
- Prior art keywords
- face
- preset
- user
- features
- target
- 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 79
- 239000011159 matrix material Substances 0.000 claims description 34
- 238000004590 computer program Methods 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 11
- 230000008521 reorganization Effects 0.000 claims description 2
- 238000005215 recombination Methods 0.000 abstract description 7
- 230000006798 recombination Effects 0.000 abstract description 7
- 238000004364 calculation method Methods 0.000 abstract description 5
- 230000001815 facial effect Effects 0.000 description 22
- 210000000887 face Anatomy 0.000 description 18
- 238000004422 calculation algorithm Methods 0.000 description 15
- 238000004891 communication Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000001680 brushing effect Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000009545 invasion Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 125000003275 alpha amino acid group Chemical group 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 210000001508 eye Anatomy 0.000 description 1
- 210000004709 eyebrow Anatomy 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000001331 nose Anatomy 0.000 description 1
- 230000036544 posture Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 210000000697 sensory organ Anatomy 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- 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
- G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical relationships
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention provides a face recognition method, a device, electronic equipment and a medium, and relates to the field of face recognition, wherein the method comprises the following steps: extracting human face features in all target human body images by the features, acquiring high-dimensional features of all human faces, dividing the high-dimensional features of all human faces, and acquiring dimension features of each human face in all groups corresponding to each human face in each human face feature; the Euclidean distance between each face dimension feature and all preset dimension features corresponding to the preset user is calculated, and each group Euclidean distance in all groups is obtained; recombining each group Euclidean distance in all groups to obtain a target Euclidean distance between the face characteristics and each preset user; and determining a target user according to the target Euclidean distance between the face characteristics and each preset user. The invention adopts a method for decomposing the high-dimensional characteristics of the face to realize Euclidean distance calculation with the characteristic data in the preset recognition library, and then utilizes characteristic recombination to realize rapid retrieval and comparison of the characteristic data, thereby finally realizing rapid and efficient face recognition.
Description
Technical Field
The present invention relates to the field of face recognition, and in particular, to a face recognition method, apparatus, electronic device, and medium.
Background
In conventional face recognition, if the face features are converted into the face high-dimensional features and are searched in a preset recognition database, the recognition speed is low due to large calculated amount, so that the recognition efficiency is reduced, and the method cannot be applied to security intrusion and face gate recognition.
Disclosure of Invention
The invention provides a face recognition method, a face recognition device, electronic equipment and a medium, which are used for solving the technical problem of low recognition speed when face recognition is carried out by adopting high-dimensional features of a face in the prior art.
In a first aspect, the present invention provides a face recognition method, including:
extracting human face features in all target human body images by the features, obtaining high-dimensional features of all human faces, dividing the high-dimensional features of all human faces by taking the number of dimension features of a preset number as grouping units, and obtaining dimension features of each human face corresponding to each human face in all groupings;
for each preset user in a preset recognition library, calculating Euclidean distances between each face dimension feature and all preset dimension features corresponding to the preset user, and obtaining each group Euclidean distance in all groups corresponding to each face feature;
For the face features in each target human body image, reorganizing the face features to correspond to each group Euclidean distance in all groups, and obtaining target Euclidean distances between the face features and each preset user;
and determining a target user according to the target Euclidean distance between the face features and each preset user, and displaying identification information corresponding to the target user in the target human body image, wherein the identification information comprises user identity and user gender.
According to the face recognition method provided by the invention, before the Euclidean distance between each face dimension feature and all preset dimension features corresponding to the preset user is calculated, the method further comprises the following steps:
dividing preset high-dimensional features corresponding to preset users by taking the number of the preset number of the dimensional features as grouping units for each preset user to obtain each preset dimensional feature in all groups;
constructing a corresponding relation between the preset user and preset identification information, and storing all preset dimension characteristics in each group corresponding to the preset user and the corresponding relation between the preset user and the preset identification information into the preset identification library;
The data size of the preset high-dimensional features is larger than the preset storage space of the storage unit in the preset identification library, and the data size of all the preset dimensional features in each group is smaller than the preset storage space of the storage unit in the preset identification library.
According to the face recognition method provided by the invention, before the face features in all target human body images are extracted by the features and all high-dimensional features of the face are acquired, the method further comprises the steps of:
decoding the video stream, and acquiring all human body images in each frame of image data corresponding to different shooting times and preset matrix diagonal coordinates corresponding to each human body image;
and determining all target human body images carrying the front face image from all human body images, and acquiring diagonal coordinates of each target matrix of all the target human body images in all frame image data.
According to the face recognition method provided by the invention, the method for displaying the recognition information corresponding to the target user in the target human body image comprises the following steps:
marking identification information and shooting time corresponding to the target user in each target matrix diagonal coordinate according to the target matrix diagonal coordinates of the target human body image in all frame image data, and obtaining all marked frame image data;
And combining all the marked frame image data, determining a marked video stream, and displaying identification information corresponding to the target user and the shooting time in the marked video stream by taking the diagonal coordinates of the target matrix as tracking objects.
According to the face recognition method provided by the invention, the determining the target user according to the target Euclidean distance between the face features and each preset user comprises the following steps:
determining a first face similarity between the face features and each preset user according to the target Euclidean distance between the face features and each preset user;
determining a preset user with the maximum similarity with the first face of the face characteristics as a first user to be identified;
and determining the first user to be identified as a target user under the condition that the first face similarity corresponding to the first user to be identified is greater than or equal to the preset similarity.
According to the face recognition method provided by the invention, after the preset user with the largest similarity with the first face of the face features is determined as the first user to be recognized, the method further comprises the steps of:
under the condition that the similarity of the first face corresponding to the first user to be identified is smaller than the preset similarity, determining the same face features corresponding to the diagonal coordinates of the target matrix in other frame image data according to the diagonal coordinates of the target matrix corresponding to the face features;
Determining a second face similarity between the same face features and each preset user according to the target Euclidean distance between the same face features and each preset user, and determining a preset user with the largest second face similarity with the same face features as a second user to be identified;
and determining the second user to be identified as a target user under the condition that the face similarity corresponding to the second user to be identified is greater than or equal to the preset similarity.
According to the face recognition method provided by the invention, after a preset user with the largest similarity with a second face with the same face characteristics is determined as a second user to be recognized, the method further comprises the steps of:
and generating an early warning instruction under the condition that the face similarity corresponding to the second user to be identified is smaller than the preset similarity, wherein the early warning instruction is used for indicating to send out an identification alarm or an intrusion alarm.
In a second aspect, there is provided a face recognition apparatus comprising:
the acquisition unit is used for extracting the human face features in all target human body images, acquiring all human face high-dimensional features, dividing the human face high-dimensional features by taking the number of the preset number of the dimension features as grouping units, and acquiring each human face feature corresponding to each human face dimension feature in all the groupings;
The computing unit is used for computing Euclidean distances between each face dimension feature and all preset dimension features corresponding to the preset users for each preset user in a preset recognition library, and acquiring each group Euclidean distance in all groups corresponding to each face feature;
the reorganization unit is used for reorganizing the face features in each target human body image, wherein the face features correspond to each group Euclidean distance in all groups, and the target Euclidean distance between the face features and each preset user is obtained;
the display unit is used for determining a target user according to the face characteristics and the target Euclidean distance between each preset user, and displaying identification information corresponding to the target user in the target human body image, wherein the identification information comprises user identity and user gender.
In a third aspect, the present invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing a face recognition method as described in any one of the above when executing the program.
In a fourth aspect, the invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a face recognition method as described in any of the above.
According to the face recognition method, the face recognition device, the electronic equipment and the medium, face features in all target human body images are extracted through the features, all face high-dimensional features are obtained, the number of the preset number of dimension features is taken as a grouping unit to divide the all face high-dimensional features, each face dimension feature in all groupings is obtained, euclidean distance between each face dimension feature and all preset dimension features corresponding to the preset user is calculated, each grouping Euclidean distance in all groupings is obtained, then each grouping Euclidean distance in all groupings is recombined, the target Euclidean distance between the face features and each preset user is obtained, finally, a target user is determined according to the target Euclidean distance between the face features and each preset user, and identification information corresponding to the target user is displayed in the target human body image. The invention adopts a method for decomposing the high-dimensional characteristics of the face to realize Euclidean distance calculation with the characteristic data in the preset recognition library, and then utilizes characteristic recombination to realize rapid retrieval and comparison of the characteristic data, thereby finally realizing rapid and efficient face recognition.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a face recognition method provided by the invention;
FIG. 2 is a second flow chart of the face recognition method according to the present invention;
fig. 3 is a third schematic flow chart of the face recognition method according to the present invention;
fig. 4 is a schematic flow chart of a face recognition method according to the present invention;
fig. 5 is a schematic flow chart of a face recognition method according to the present invention;
fig. 6 is a flowchart of a face recognition method according to the present invention;
fig. 7 is a schematic flow chart of a face recognition method according to the present invention;
fig. 8 is a schematic structural diagram of a face recognition device provided by the present invention;
fig. 9 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The face recognition technology is widely applied to the fields of security protection, intelligent travel, identity recognition and the like, and a face recognition system mainly relies on a camera to collect face images, extracts face features through a face recognition algorithm and performs list library matching and comparison aiming at the features, so that an accurate recognition result is obtained.
The face recognition technology mainly relies on video stream data acquired by a camera, decodes the video stream data into image frames, detects human body information through a face detection algorithm, extracts a face image on the basis of a human body target, realizes tracking detection of the human body information, and establishes one-to-one correlation of the face image and the personnel tracking information; face features are extracted, characterization is carried out to generate high-dimensional face feature data, comparison and identification are carried out according to the list database data, and therefore identity information of each face and human body target tracking is obtained. However, after face features are extracted and characterized to generate high-dimensional face feature data, the comparison recognition capability is reduced due to large calculation amount, and the reduction of the recognition speed seriously affects the handling of emergency situations of security protection, intelligent travel and identity recognition.
Step 101, extracting face features in all target human body images by features, obtaining all face high-dimensional features, dividing the face high-dimensional features by grouping units according to the number of the preset number of the dimension features, and obtaining each face feature corresponding to each face dimension feature in all groupings.
In step 101, the target human body image is a human body image carrying a human body face, and for each target human body image, a human face feature may be included or a plurality of human face features may be included, and for the human face features in the target human body image, all the human face features may be intercepted from the target human body image according to a preset size, where the preset size may beThe size of (2) may be +.>This may be determined according to the size of the target human body image and the resolution size.
Optionally, the feature extraction method comprises the steps of extracting the face features in all target human body images, associating each face feature with the shooting time of the current image to obtain all face high-dimensional features, and obtaining the face high-dimensional features through a preset feature extraction model, wherein the face high-dimensional features are 512-dimensional face feature data, the preset feature extraction model comprises four layers of convolution layers, each layer of convolution layers is increased to 128 feature mappings, 512-dimensional face feature data are finally produced, and the feature extraction can be executed for each face feature, so that the face high-dimensional features corresponding to each face feature are determined.
Optionally, the number of the preset number of dimension features may be 100, that is, the invention uses the number of 100 dimension features as grouping units to divide the high-dimension features of all faces, and in the case that the high-dimension features of the faces corresponding to the face features are 512 dimensions, the high-dimension features of the faces may be divided into 6 groups, where each group in the first five groups contains 100 dimension features of the faces and the last group contains 12 dimension features of the faces, so as to determine that the face features correspond to each dimension feature of the faces in all groups, and for each face feature, uses the number of 100 dimension features as grouping units to divide the high-dimension features of all faces, so as to obtain each dimension feature of each face in all groups.
Step 102, for each preset user in a preset recognition library, calculating Euclidean distances between each face dimension feature and all preset dimension features corresponding to the preset user, and obtaining Euclidean distances of each face feature corresponding to each group in all groups.
In step 102, in the present invention, when determining that each face feature corresponds to each face dimension feature in all groups, it is necessary to calculate a euclidean distance between each face feature and each preset user in a preset recognition library, by taking any face feature as an example, for each preset user in the preset recognition library, all preset dimension features corresponding to the preset user exist, and further calculate the euclidean distance between each face dimension feature and all preset dimension features corresponding to the preset user, the following formula may be referred to:
（1）
In the formula (1), the components are as follows,is->Storable features of a group and the first +.>Euclidean distance of each preset user, +.>Is->Face dimension feature of dimension,/->For presetting the identification library +.>The +.>Vector of dimensions>Is the number of dimensions in which the feature can be stored.
Step 103, recombining the face features in each target human body image to correspond to each group Euclidean distance in all groups, and obtaining the target Euclidean distance between the face features and each preset user.
In step 103, the face features are recombined to correspond to each group euclidean distance in all groups, and the target euclidean distance between the face features and each preset user is obtained, which can be referred to as the following formula:
（2）
in the formula (2), the amino acid sequence of the compound,is->Storable features of a group and the first +.>The Euclidean distance of each preset user is calculated, and the Euclidean distance recombination in the formula (2) is carried out on the calculated Euclidean distances of 6 groups, so that the face characteristics and each preset user are determinedTarget European distance->。
And (3) executing Euclidean distance recombination in the formula (2) for each face feature in each target human body image, so as to determine the target Euclidean distance between each face feature in each target human body image and each preset user.
Step 104, determining a target user according to the target Euclidean distance between the face features and each preset user, and displaying identification information corresponding to the target user in the target human body image, wherein the identification information comprises user identity and user gender.
In step 104, the present invention may determine that a preset user with the minimum distance from the face feature to the face feature is the target user according to the target euclidean distance between the face feature and each preset user, and may convert the target euclidean distance between the face feature and each preset user into a similarity value, and determine that the preset user with the maximum similarity in the similarity value is the target user according to the preset user with the maximum similarity in the similarity value, and further display identification information corresponding to the target user in the target human body image according to the correspondence between the target user and the identification information in the preset identification library, where the identification information includes a user identity and a user gender.
Optionally, in the case that a plurality of face features exist in the target human body image, the identification information corresponding to the target user identified by each face feature is displayed in the target human body image according to the target user identified by each face feature.
According to the face recognition method provided by the invention, the face features in all target human body images are extracted through the features, all the face high-dimensional features are obtained, the number of dimension features of a preset number is taken as a grouping unit to divide the all the face high-dimensional features, each face dimension feature corresponding to each face dimension feature in all the groups is obtained, euclidean distance between each face dimension feature and all the preset dimension features corresponding to the preset users is calculated, each group Euclidean distance in each face feature corresponding to all the groups is obtained, then each group Euclidean distance in all the groups is recombined, the target Euclidean distance between the face features and each preset user is obtained, finally, the target user is determined according to the target Euclidean distance between the face features and each preset user, and the identification information corresponding to the target user is displayed in the target human body image. The invention adopts a method for decomposing the high-dimensional characteristics of the face to realize Euclidean distance calculation with the characteristic data in the preset recognition library, and then utilizes characteristic recombination to realize rapid retrieval and comparison of the characteristic data, thereby finally realizing rapid and efficient face recognition.
Fig. 2 is a second flow chart of the face recognition method provided by the present invention, before calculating the euclidean distance between each face dimension feature and all preset dimension features corresponding to the preset user, the method further includes:
step 201, for each preset user, dividing preset high-dimensional features corresponding to the preset users by taking the number of dimension features in a preset number as a grouping unit, and obtaining each preset dimension feature in all the groups.
In step 201, the preset recognition library is required to be pre-constructed before the Euclidean distance between each face dimension feature and all preset dimension features corresponding to the preset user is calculated, and specifically, related users with the corresponding relationship between the preset user and the preset recognition information are stored in groups.
The person skilled in the art understands that the method can realize the storage of the feature data by adopting a high-dimensional decomposition method and realize the rapid retrieval and comparison of the feature data by utilizing a feature recombination method aiming at the defects of insufficient storage space, low retrieval speed and the like of the conventional database of the high-dimensional features of the face, and finally realize the rapid and efficient face recognition effect.
Optionally, the data size of the preset high-dimensional features is larger than the preset storage space of the preset identification library storage unit, and the data size of all the preset dimensional features in each group is smaller than the preset storage space of the preset identification library storage unit.
Optionally, the invention provides a high-dimensional face feature retrieval and storage method, which aims at realizing retrieval and identification of high-dimensional face features by utilizing dimension features stored in groups in a preset identification library while storing high-dimensional face feature data; taking 512 dimensions as an example, since the high-dimensional feature data cannot be stored in the conventional database, the high-dimensional data is decomposed into a plurality of storable dimension features within 128 dimensions of the storable conventional data, for example, the data volume of the preset dimension features is 100 dimensions of data volume, and the storable dimension features are named with different field names and are associated with the same preset identification information; in order to realize the matching retrieval of the high-dimensional features of the face corresponding to the face features and the preset high-dimensional features in the preset recognition library, the decomposed face dimension features are required to be used as input, the comparison between the retrieval algorithm and the preset dimension features in the preset recognition library is based, and the preset user with the minimum Euclidean distance or the maximum similarity is determined to be the face recognition optimal result corresponding to the face features.
Step 202, constructing a corresponding relation between the preset user and the preset identification information, and storing all preset dimension features in each group corresponding to the preset user, and the corresponding relation between the preset user and the preset identification information to the preset identification library.
In step 202, the invention can construct the corresponding relation between the preset user and the preset identification information through the prior identification matching, can also manually edit and construct the corresponding relation between the preset user and the preset identification information according to the preset requirement by calling the corresponding relation between the preset user and the preset identification information constructed in the third party data platform, and stores all preset dimension characteristics in each group corresponding to the preset user into the preset identification library.
The invention provides a face feature recognition and retrieval method, and also provides a high-dimension face feature storage method based on the face feature recognition and retrieval method, wherein the high-dimension face feature data with large information quantity is stored in a mode of decomposing and recombining the high-dimension face feature data, so that the problem that the conventional recognition library cannot store high-dimension vector data is solved, and the completeness of the face features and the accuracy of face recognition are effectively ensured under the condition that the face recognition efficiency is not influenced.
Fig. 3 is a third flow chart of the face recognition method provided by the present invention, before feature extraction of face features in all target human body images and obtaining of high-dimensional features of all faces, the method further includes:
step 301, decoding the video stream, and obtaining all human body images in each frame of image data corresponding to different shooting times and a preset matrix diagonal coordinate corresponding to each human body image.
In step 301, the video stream may be video stream information in real time monitoring acquired in security intrusion or face gate recognition, in the process of real time monitoring, the invention decodes the video stream, and acquires all human body images in each frame of image data corresponding to different shooting times, wherein the human body images include human body images with front face images and human body images without front face images.
Step 302, determining all target human body images carrying the front face image from all human body images, and acquiring diagonal coordinates of each target matrix of all the target human body images in all frame image data.
In step 302, the person is in a free state in the view of the camera, the angles of face deflection, pitching, positive side and the like are in an uncontrollable state, the face feature extraction mainly extracts the five sense organs data of eyebrows, eyes, noses and the like, the feature in the positive face state can best represent the person, and the feature ratio accuracy of the side face information is greatly reduced.
Optionally, the diagonal coordinates of the target matrix may also play a role in tracking targets in implementing real-time face recognition, and because the present invention performs association processing on the target human body image in each frame of image by acquiring the diagonal coordinates of each target matrix of all target human body images in all frame of image data, even if the target human body image in a certain human body image cannot be identified in the current frame, the target human body image in the human body image may be identified in the next frame or the next frame.
Optionally, the video image data in the video stream is used as a common input of a face recognition technology, the face recognition effect is often seriously affected by the quality of the image data, common noise such as impulse noise and Gaussian noise can generate serious interference on face feature extraction, and for face recognition in the video stream, the face feature in the image frame is usually more required to be converted into a face high-dimensional feature so as to improve the face recognition capability. According to the face recognition method, the recognition is carried out by adopting the high-dimensional features of the face, and meanwhile, the technical problem is effectively solved by adopting a mode of decomposing feature data recognition, so that the recognition speed in security intrusion and face gate recognition is improved, and finally, the recognition response speed and the recognition accuracy are improved.
Fig. 4 is a schematic flow chart of a face recognition method provided by the present invention, in which identification information corresponding to the target user is displayed in the target human body image, including:
And 401, marking identification information and shooting time corresponding to the target user in each target matrix diagonal coordinate according to the target matrix diagonal coordinates of the target human body image in all frame image data, and acquiring all marked frame image data.
In step 401, after determining the target user according to the target euclidean distance between the face feature and each preset user, the invention can continuously lock the diagonal coordinates of the target matrix in different frame image data according to the wearing colors and body type features of the upper body and the lower body of the person in the diagonal coordinates of the target matrix because the diagonal coordinates of the target matrix in all frame image data in the video stream can continuously move along with the change of time, and mark the identification information and shooting time corresponding to the target user in each diagonal coordinate of the target matrix because all diagonal coordinates of the target matrix in different frame image data are pointed to the same target user, thereby obtaining all marked frame image data.
And 402, combining all the marked frame image data, determining a marked video stream, and displaying identification information corresponding to the target user and the shooting time in the marked video stream by taking the diagonal coordinates of the target matrix as tracking objects.
In step 402, the present invention obtains all frame image data after decoding the video stream, marks the identification information and the shooting time corresponding to the target user in each diagonal coordinate of the target matrix, obtains all marked frame image data, and then reorganizes all marked frame image data according to the sequence of the shooting time, and forms a marked video stream again.
Optionally, the invention can be applied to the field of intelligent security and the field of face gate, in the intelligent security and protection field, the invention can effectively ensure the integrity of high-dimensional face feature data, the accuracy of face recognition and the storage efficiency of strange face feature data, and realize the rapid, accurate and all-weather face recognition service of an intelligent security and protection system; in the field of face gate, the face brushing security inspection by using the face verification technology at present enters a common application stage, and the face recognition accuracy can be effectively improved by applying the face brushing security inspection method based on the existing face brushing security inspection.
Fig. 5 is a schematic flow chart of a face recognition method provided by the present invention, wherein the determining a target user according to a target euclidean distance between the face feature and each preset user includes:
step 501, determining a first face similarity between the face feature and each preset user according to the target Euclidean distance between the face feature and each preset user.
In step 501, the present invention converts the face similarity to the target euclidean distance according to the face feature and the target euclidean distance between each preset user, specifically, the following formula may be referred to:
（3）
in the formula (3), theFor the first face similarity between the face features and each preset user, the +.>For the first parameter, said ++>For the second parameter, the invention can record the maximum value of Euclidean distance, and at this time, a value corresponding to the maximum percentage is set, for example, the maximum percentage is 95%, then the minimum value of Euclidean distance is recorded, and the value is setOne value corresponds to a minimum percentage, for example a minimum percentage of 5%, and then the first parameter +_ can be calculated by substituting the maximum value of Euclidean distance and the minimum value of Euclidean distance into (3) >Said second parameter +.>。
Step 502, determining a preset user with the maximum similarity with the first face of the face features as a first user to be identified.
In step 502, according to the target euclidean distance between the face feature and each preset user and formula (3), determining a first face similarity between the face feature and each preset user, selecting a preset user with the highest matching degree from all the first face similarities between the face feature and the preset users, and determining the user with the highest matching degree as a first user to be identified.
Step 503, determining that the first user to be identified is a target user when the first face similarity corresponding to the first user to be identified is greater than or equal to a preset similarity.
In step 503, if all the first facial features have low similarity with the preset user, even if the preset user having the largest first facial similarity with the facial features is selected, the first facial similarity may be only 50% similar, so in order to exclude the target user which does not match with the facial features in the preset recognition library, the first user to be recognized may be determined to be the target user only if the similarity of the first user to be recognized is 80% or more, by setting the preset similarity, for example, the preset similarity is 80%, even if the first user to be recognized is finally determined, because the similarity of the first user to be recognized is only 79%, and the first user to be recognized may be determined to be the target user only if the similarity of the first user to be recognized is 80% or more.
Fig. 6 is a flowchart of a face recognition method provided by the present invention, after determining a preset user with the greatest similarity to a first face of the face features as a first user to be recognized, the method further includes:
and 601, determining the same face feature corresponding to the diagonal coordinates of the target matrix in other frame image data according to the diagonal coordinates of the target matrix corresponding to the face feature when the first face similarity corresponding to the first user to be identified is smaller than the preset similarity.
In step 601, for example, the face similarity corresponding to the user to be identified is 65% and the preset similarity is 70%, where the face similarity corresponding to the user to be identified is smaller than the preset similarity, and at this time, according to the diagonal coordinates of the target matrix corresponding to the face feature, it is determined whether the same face feature as the face feature exists in the diagonal coordinates of the target matrix of other frame image data.
Step 602, determining a second face similarity between the same face feature and each preset user according to the target Euclidean distance between the same face feature and each preset user, and determining a preset user with the largest second face similarity with the same face feature as a second user to be identified.
In step 602, if it is determined that the same facial features as the facial features exist in the other frame image data, a second facial similarity between the same facial features and each preset user is determined according to a target euclidean distance between the same facial features and each preset user, and the same facial features are all directed to the same person, but because the same facial features are different from the target human body image in which the same facial features are located, the same facial features are also different from the facial features, so that the second facial similarity is different from the first facial similarity, and further, a preset user with the largest second facial similarity with the same facial features is determined as a second user to be identified, and because of different front view angles, different light and shade, different resolution clarity and other influencing factors of the face, the target user cannot be determined according to the first facial similarity, but the target user can be determined according to the second facial similarity, so that the second user with the same facial features is determined again as the second user to be identified.
Step 603, determining the second user to be identified as a target user when the face similarity corresponding to the second user to be identified is greater than or equal to a preset similarity.
In step 603, the present invention determines the second user to be identified by searching for the same face feature corresponding to the diagonal coordinates of the target matrix in the other frame image data again when the first face similarity corresponding to the first user to be identified is smaller than the preset similarity, and further determines the second user to be identified as the target user when the face similarity corresponding to the second user to be identified is greater than or equal to the preset similarity.
Optionally, after determining the preset user with the largest similarity to the second face with the same face feature as the second user to be identified, the method further includes:
and generating an early warning instruction under the condition that the face similarity corresponding to the second user to be identified is smaller than the preset similarity, wherein the early warning instruction is used for indicating to send out an identification alarm or an intrusion alarm.
Optionally, the method traverses all other frame image data corresponding to the face features to determine the target user according to the corresponding same face features in the other frame image data, but if the face similarity corresponding to the second user to be identified is still determined to be smaller than the preset similarity after traversing all the same face features, the current face features in the current video stream are considered not to exist in the preset identification library, if the method is applied to the intelligent security field and the face gate field, illegal invasion, unregistered, unidentified and other abnormal conditions can be considered, reporting processing can be performed according to actual conditions at this time, specifically, an early warning instruction is generated, and the early warning instruction is used for indicating to send out an identification alarm or an invasion alarm.
FIG. 7 is a schematic flow chart of the face recognition method provided by the invention, as shown in FIG. 7, the face recognition of the invention comprises a basic algorithm engine block, a high-dimensional feature storage and retrieval algorithm block and a feature retrieval block, in the basic algorithm engine block, the invention tracks a target according to a target detection algorithm on one hand, tracks the feature by adopting a tracking algorithm, specifically, sends relevant data of a human target into the tracking algorithm, inputs data comprising frame image data, diagonal coordinates of a target matrix, shooting time and the like into the tracking algorithm, tracks and files multiple frames of data with the same target feature into the same person, and counts the tracking information of the person under video stream data for further processing; on the other hand, according to the target face, a feature extraction algorithm is adopted to extract high-dimensional features, specifically, the target face and face image coordinates are input into the feature extraction algorithm, and optionally, the input information comprises frame image data, target human body images, target matrix diagonal coordinates, shooting time, user gender and other data; in the high-dimensional feature storage and retrieval algorithm plate, decomposing the high-dimensional feature to determine a plurality of groups of storable dimensional features, optionally, 6 groups of data are adopted, and in other embodiments, 8 groups, 10 groups or more are adopted, so that the database storage of the plurality of groups of storable dimensional features is realized according to an application program interface (Application Program Interface, API); in the feature retrieval plate, the invention calculates the Euclidean distance between storable dimension features in each group and preset dimension features in a preset recognition library respectively, after the Euclidean distance is extracted, the Euclidean distance in all groups is recombined, personnel information extraction is realized according to similarity conversion and similarity discrimination, as tracking algorithm tracking features are adopted in the basic algorithm engine plate, and the recognition information is associated with each group of storage fields in advance, the recognition information can be not only user identity and user gender, but also feature identification ID, and finally the tracking information is associated with the extracted personnel information, the personnel information with the same human features can be deduced based on image data such as the front face, the side face, the background or the background is not present under different time, and finally the interface display of the personnel information with all postures is realized, including the recognition, tracking and early warning of all target users.
Fig. 8 is a schematic structural diagram of a face recognition device provided by the present invention, where the face recognition device includes an obtaining unit 1, where the obtaining unit is configured to extract face features in all target human body images by using features, obtain all high-dimensional features of the faces, divide the all high-dimensional features of the faces by using a preset number of dimension feature numbers as grouping units, and obtain each face feature corresponding to each face dimension feature in all groupings, and the working principle of the obtaining unit 1 may refer to the foregoing step 101 and will not be repeated herein.
The face recognition device further includes a calculating unit 2, where the calculating unit is configured to calculate, for each preset user in the preset recognition library, a euclidean distance between each face dimension feature and all preset dimension features corresponding to the preset user, and obtain each group euclidean distance in all groups corresponding to each face feature, and the working principle of the calculating unit 2 may refer to the foregoing step 102 and will not be described herein.
The face recognition device further includes a reorganizing unit 3, where the reorganizing unit is configured to reorganize, for a face feature in each target human body image, the face feature corresponds to each group euclidean distance in all groups, and obtain a target euclidean distance between the face feature and each preset user, and the working principle of the reorganizing unit 3 may refer to the foregoing step 103 and will not be repeated herein.
The face recognition device further includes a display unit 4, where the display unit is configured to determine a target user according to a target euclidean distance between the face feature and each preset user, and display, in the target human body image, identification information corresponding to the target user, where the identification information includes a user identity and a user gender, and a working principle of the display unit 4 may refer to the foregoing step 104 and is not described herein.
According to the face recognition method, the face recognition device, the electronic equipment and the medium, face features in all target human body images are extracted through the features, all face high-dimensional features are obtained, the number of the preset number of dimension features is taken as a grouping unit to divide the all face high-dimensional features, each face dimension feature in all groupings is obtained, euclidean distance between each face dimension feature and all preset dimension features corresponding to the preset user is calculated, each grouping Euclidean distance in all groupings is obtained, then each grouping Euclidean distance in all groupings is recombined, the target Euclidean distance between the face features and each preset user is obtained, finally, a target user is determined according to the target Euclidean distance between the face features and each preset user, and identification information corresponding to the target user is displayed in the target human body image. The invention adopts a method for decomposing the high-dimensional characteristics of the face to realize Euclidean distance calculation with the characteristic data in the preset recognition library, and then utilizes characteristic recombination to realize rapid retrieval and comparison of the characteristic data, thereby finally realizing rapid and efficient face recognition.
Fig. 9 is a schematic structural diagram of an electronic device provided by the present invention. As shown in fig. 9, the electronic device may include: processor 910, communication interface (Communications Interface), memory 930, and communication bus 940, wherein processor 910, communication interface 920, and memory 930 communicate with each other via communication bus 940. Processor 910 may invoke logic instructions in memory 930 to perform a face recognition method comprising: extracting human face features in all target human body images by the features, obtaining high-dimensional features of all human faces, dividing the high-dimensional features of all human faces by taking the number of dimension features of a preset number as grouping units, and obtaining dimension features of each human face corresponding to each human face in all groupings; for each preset user in a preset recognition library, calculating Euclidean distances between each face dimension feature and all preset dimension features corresponding to the preset user, and obtaining each group Euclidean distance in all groups corresponding to each face feature; for the face features in each target human body image, reorganizing the face features to correspond to each group Euclidean distance in all groups, and obtaining target Euclidean distances between the face features and each preset user; and determining a target user according to the target Euclidean distance between the face features and each preset user, and displaying identification information corresponding to the target user in the target human body image, wherein the identification information comprises user identity and user gender.
Further, the logic instructions in the memory 930 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to 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, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product comprising a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of performing a face recognition method provided by the above methods, the method comprising: extracting human face features in all target human body images by the features, obtaining high-dimensional features of all human faces, dividing the high-dimensional features of all human faces by taking the number of dimension features of a preset number as grouping units, and obtaining dimension features of each human face corresponding to each human face in all groupings; for each preset user in a preset recognition library, calculating Euclidean distances between each face dimension feature and all preset dimension features corresponding to the preset user, and obtaining each group Euclidean distance in all groups corresponding to each face feature; for the face features in each target human body image, reorganizing the face features to correspond to each group Euclidean distance in all groups, and obtaining target Euclidean distances between the face features and each preset user; and determining a target user according to the target Euclidean distance between the face features and each preset user, and displaying identification information corresponding to the target user in the target human body image, wherein the identification information comprises user identity and user gender.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the face recognition method provided by the above methods, the method comprising: extracting human face features in all target human body images by the features, obtaining high-dimensional features of all human faces, dividing the high-dimensional features of all human faces by taking the number of dimension features of a preset number as grouping units, and obtaining dimension features of each human face corresponding to each human face in all groupings; for each preset user in a preset recognition library, calculating Euclidean distances between each face dimension feature and all preset dimension features corresponding to the preset user, and obtaining each group Euclidean distance in all groups corresponding to each face feature; for the face features in each target human body image, reorganizing the face features to correspond to each group Euclidean distance in all groups, and obtaining target Euclidean distances between the face features and each preset user; and determining a target user according to the target Euclidean distance between the face features and each preset user, and displaying identification information corresponding to the target user in the target human body image, wherein the identification information comprises user identity and user gender.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the 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 scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A face recognition method, comprising:
extracting human face features in all target human body images by the features, obtaining high-dimensional features of all human faces, dividing the high-dimensional features of all human faces by taking the number of dimension features of a preset number as grouping units, and obtaining dimension features of each human face corresponding to each human face in all groupings;
for each preset user in a preset recognition library, calculating Euclidean distances between each face dimension feature and all preset dimension features corresponding to the preset user, and obtaining each group Euclidean distance in all groups corresponding to each face feature;
for the face features in each target human body image, reorganizing the face features to correspond to each group Euclidean distance in all groups, and obtaining target Euclidean distances between the face features and each preset user;
And determining a target user according to the target Euclidean distance between the face features and each preset user, and displaying identification information corresponding to the target user in the target human body image, wherein the identification information comprises user identity and user gender.
2. The face recognition method according to claim 1, wherein before calculating the euclidean distance between each of the face dimension features and all of the preset dimension features corresponding to the preset user, the method further comprises:
dividing preset high-dimensional features corresponding to preset users by taking the number of the preset number of the dimensional features as grouping units for each preset user to obtain each preset dimensional feature in all groups;
constructing a corresponding relation between the preset user and preset identification information, and storing all preset dimension characteristics in each group corresponding to the preset user and the corresponding relation between the preset user and the preset identification information into the preset identification library;
the data size of the preset high-dimensional features is larger than the preset storage space of the storage unit in the preset identification library, and the data size of all the preset dimensional features in each group is smaller than the preset storage space of the storage unit in the preset identification library.
3. The face recognition method according to claim 1, wherein before feature extraction of face features in all target human body images and acquisition of all high-dimensional features of faces, the method further comprises:
decoding the video stream, and acquiring all human body images in each frame of image data corresponding to different shooting times and preset matrix diagonal coordinates corresponding to each human body image;
and determining all target human body images carrying the front face image from all human body images, and acquiring diagonal coordinates of each target matrix of all the target human body images in all frame image data.
4. A face recognition method according to claim 3, wherein displaying the identification information corresponding to the target user in the target human body image includes:
marking identification information and shooting time corresponding to the target user in each target matrix diagonal coordinate according to the target matrix diagonal coordinates of the target human body image in all frame image data, and obtaining all marked frame image data;
and combining all the marked frame image data, determining a marked video stream, and displaying identification information corresponding to the target user and the shooting time in the marked video stream by taking the diagonal coordinates of the target matrix as tracking objects.
5. The method of claim 4, wherein determining the target user based on the target euclidean distance between the face feature and each preset user comprises:
determining a first face similarity between the face features and each preset user according to the target Euclidean distance between the face features and each preset user;
determining a preset user with the maximum similarity with the first face of the face characteristics as a first user to be identified;
and determining the first user to be identified as a target user under the condition that the first face similarity corresponding to the first user to be identified is greater than or equal to the preset similarity.
6. The face recognition method according to claim 5, wherein after determining the preset user having the greatest similarity to the first face of the face features as the first user to be recognized, the method further comprises:
under the condition that the similarity of the first face corresponding to the first user to be identified is smaller than the preset similarity, determining the same face features corresponding to the diagonal coordinates of the target matrix in other frame image data according to the diagonal coordinates of the target matrix corresponding to the face features;
Determining a second face similarity between the same face features and each preset user according to the target Euclidean distance between the same face features and each preset user, and determining a preset user with the largest second face similarity with the same face features as a second user to be identified;
and determining the second user to be identified as a target user under the condition that the face similarity corresponding to the second user to be identified is greater than or equal to the preset similarity.
7. The face recognition method according to claim 6, wherein after determining a preset user having the greatest similarity to a second face of the same face features as the second user to be recognized, the method further comprises:
and generating an early warning instruction under the condition that the face similarity corresponding to the second user to be identified is smaller than the preset similarity, wherein the early warning instruction is used for indicating to send out an identification alarm or an intrusion alarm.
8. A face recognition device, comprising:
the acquisition unit is used for extracting the human face features in all target human body images, acquiring all human face high-dimensional features, dividing the human face high-dimensional features by taking the number of the preset number of the dimension features as grouping units, and acquiring each human face feature corresponding to each human face dimension feature in all the groupings;
The computing unit is used for computing Euclidean distances between each face dimension feature and all preset dimension features corresponding to the preset users for each preset user in a preset recognition library, and acquiring each group Euclidean distance in all groups corresponding to each face feature;
the reorganization unit is used for reorganizing the face features in each target human body image, wherein the face features correspond to each group Euclidean distance in all groups, and the target Euclidean distance between the face features and each preset user is obtained;
the display unit is used for determining a target user according to the face characteristics and the target Euclidean distance between each preset user, and displaying identification information corresponding to the target user in the target human body image, wherein the identification information comprises user identity and user gender.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the face recognition method of any one of claims 1-7 when the program is executed by the processor.
10. A non-transitory computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the face recognition method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311182465.9A CN116912925A (en) | 2023-09-14 | 2023-09-14 | Face recognition method, device, electronic equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311182465.9A CN116912925A (en) | 2023-09-14 | 2023-09-14 | Face recognition method, device, electronic equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116912925A true CN116912925A (en) | 2023-10-20 |
Family
ID=88353527
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311182465.9A Pending CN116912925A (en) | 2023-09-14 | 2023-09-14 | Face recognition method, device, electronic equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116912925A (en) |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105488478A (en) * | 2015-12-02 | 2016-04-13 | 深圳市商汤科技有限公司 | Face recognition system and method |
| CN106203367A (en) * | 2016-07-18 | 2016-12-07 | 北京红马传媒文化发展有限公司 | A kind of method of identification, system and terminal |
| CN108629345A (en) * | 2017-03-17 | 2018-10-09 | 北京京东尚科信息技术有限公司 | Dimensional images feature matching method and device |
| CN109426781A (en) * | 2017-08-29 | 2019-03-05 | 阿里巴巴集团控股有限公司 | Construction method, face identification method, device and the equipment of face recognition database |
| CN112329602A (en) * | 2020-11-02 | 2021-02-05 | 平安科技（深圳）有限公司 | Method and device for acquiring face annotation image, electronic equipment and storage medium |
| CN112990093A (en) * | 2021-04-09 | 2021-06-18 | 京东方科技集团股份有限公司 | Personnel monitoring method, device and system and computer readable storage medium |
| CN113255828A (en) * | 2021-06-17 | 2021-08-13 | 长沙海信智能系统研究院有限公司 | Feature retrieval method, device, equipment and computer storage medium |
| CN113537107A (en) * | 2021-07-23 | 2021-10-22 | 山东浪潮通软信息科技有限公司 | Face recognition and tracking method, device and equipment based on deep learning |
| CN114911958A (en) * | 2022-06-09 | 2022-08-16 | 电子科技大学 | Semantic preference-based rapid image retrieval method |
| CN114998955A (en) * | 2022-04-29 | 2022-09-02 | 北京精鸿软件科技有限公司 | Face recognition method and device, storage medium and electronic equipment |
| CN115132370A (en) * | 2022-07-07 | 2022-09-30 | 浙江大学 | Flow adjustment auxiliary method and device based on machine vision and deep learning |
| CN116071798A (en) * | 2022-12-30 | 2023-05-05 | 新石器慧通（北京）科技有限公司 | Identification authentication method and device, storage medium and unmanned vehicle |
| CN116342968A (en) * | 2023-01-18 | 2023-06-27 | 北京六律科技有限责任公司 | Dual-channel face recognition method and device |
| CN116403267A (en) * | 2023-04-10 | 2023-07-07 | 杭州海康威视数字技术股份有限公司 | Personnel file construction method, device, equipment and storage medium |
| CN116405227A (en) * | 2023-04-26 | 2023-07-07 | 中国工商银行股份有限公司 | Identity authentication method and device, storage medium and electronic equipment |
-
2023
- 2023-09-14 CN CN202311182465.9A patent/CN116912925A/en active Pending
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105488478A (en) * | 2015-12-02 | 2016-04-13 | 深圳市商汤科技有限公司 | Face recognition system and method |
| CN106203367A (en) * | 2016-07-18 | 2016-12-07 | 北京红马传媒文化发展有限公司 | A kind of method of identification, system and terminal |
| CN108629345A (en) * | 2017-03-17 | 2018-10-09 | 北京京东尚科信息技术有限公司 | Dimensional images feature matching method and device |
| CN109426781A (en) * | 2017-08-29 | 2019-03-05 | 阿里巴巴集团控股有限公司 | Construction method, face identification method, device and the equipment of face recognition database |
| CN112329602A (en) * | 2020-11-02 | 2021-02-05 | 平安科技（深圳）有限公司 | Method and device for acquiring face annotation image, electronic equipment and storage medium |
| CN112990093A (en) * | 2021-04-09 | 2021-06-18 | 京东方科技集团股份有限公司 | Personnel monitoring method, device and system and computer readable storage medium |
| CN113255828A (en) * | 2021-06-17 | 2021-08-13 | 长沙海信智能系统研究院有限公司 | Feature retrieval method, device, equipment and computer storage medium |
| CN113537107A (en) * | 2021-07-23 | 2021-10-22 | 山东浪潮通软信息科技有限公司 | Face recognition and tracking method, device and equipment based on deep learning |
| CN114998955A (en) * | 2022-04-29 | 2022-09-02 | 北京精鸿软件科技有限公司 | Face recognition method and device, storage medium and electronic equipment |
| CN114911958A (en) * | 2022-06-09 | 2022-08-16 | 电子科技大学 | Semantic preference-based rapid image retrieval method |
| CN115132370A (en) * | 2022-07-07 | 2022-09-30 | 浙江大学 | Flow adjustment auxiliary method and device based on machine vision and deep learning |
| CN116071798A (en) * | 2022-12-30 | 2023-05-05 | 新石器慧通（北京）科技有限公司 | Identification authentication method and device, storage medium and unmanned vehicle |
| CN116342968A (en) * | 2023-01-18 | 2023-06-27 | 北京六律科技有限责任公司 | Dual-channel face recognition method and device |
| CN116403267A (en) * | 2023-04-10 | 2023-07-07 | 杭州海康威视数字技术股份有限公司 | Personnel file construction method, device, equipment and storage medium |
| CN116405227A (en) * | 2023-04-26 | 2023-07-07 | 中国工商银行股份有限公司 | Identity authentication method and device, storage medium and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110418114B (en) | Object tracking method and device, electronic equipment and storage medium | |
| WO2021017882A1 (en) | Image coordinate system conversion method and apparatus, device and storage medium | |
| CN109919977B (en) | Video motion person tracking and identity recognition method based on time characteristics | |
| CN106355154B (en) | Method for detecting frequent passing of people in surveillance video | |
| CN108282460B (en) | Evidence chain generation method and device for network security event | |
| CN114332744B (en) | Transformer substation self-adaptive security method and system based on machine vision | |
| CN109214324A (en) | Most face image output method and output system based on polyphaser array | |
| CN110827432A (en) | Class attendance checking method and system based on face recognition | |
| CN116503399A (en) | Insulator pollution flashover detection method based on YOLO-AFPS | |
| Yang et al. | A method of pedestrians counting based on deep learning | |
| CN110321808B (en) | Method, apparatus and storage medium for detecting carry-over and stolen object | |
| CN109359530B (en) | Intelligent video monitoring method and device | |
| CN111738059A (en) | Non-sensory scene-oriented face recognition method | |
| CN113705393A (en) | 3D face model-based depression angle face recognition method and system | |
| CN114494744A (en) | Method and device for obtaining object track similarity, electronic equipment and storage medium | |
| CN112232269B (en) | Ship identity intelligent recognition method and system based on twin network | |
| CN112001280B (en) | Real-time and online optimized face recognition system and method | |
| CN116912925A (en) | Face recognition method, device, electronic equipment and medium | |
| CN115984178A (en) | Counterfeit image detection method, electronic device, and computer-readable storage medium | |
| Huang et al. | Robust facial expression recognition using revised canonical correlation | |
| CN112257666B (en) | Target image content aggregation method, device, equipment and readable storage medium | |
| CN108280468A (en) | A kind of image-recognizing method based on grid | |
| CN116935253A (en) | Human face tampering detection method based on residual error network combined with space-time attention mechanism | |
| CN113947789A (en) | Scene face recognition technology for events | |
| CN115719469A (en) | Target identification method and device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20231020 |
|
| RJ01 | Rejection of invention patent application after publication |