CN110413811A - A kind of across channel polymerization in passerby library based on recognition of face - Google Patents
A kind of across channel polymerization in passerby library based on recognition of face Download PDFInfo
- Publication number
- CN110413811A CN110413811A CN201910715542.XA CN201910715542A CN110413811A CN 110413811 A CN110413811 A CN 110413811A CN 201910715542 A CN201910715542 A CN 201910715542A CN 110413811 A CN110413811 A CN 110413811A
- Authority
- CN
- China
- Prior art keywords
- archives
- personnel
- candid photograph
- channel
- polymerization
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/51—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
- G06T2207/30201—Face
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Library & Information Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of across the channel polymerization in passerby library based on recognition of face, comprising the following steps: S1, a person ownership is retained by the polymerization of single channel in short-term capture in figure quality highest one;S11, the information cache queue of candid photograph figure is cached;S12, when another candid photograph figure caching is come in, compared with candid photograph figure before;S13, when reach set cache-time after, enable caching candid photograph figure enter across channel converging operation;S2, it will be captured and schemed in storage route people library by across channel polymerization;S21, in each personnel's archives in passerby library, select representative figure;S22, setting compare threshold value, are compared the candid photograph figure across channel converging operation is entered in step S13 with the representative figure of all personnel's archives in passerby library, find personnel's archives similar with the candid photograph figure;S23, personnel's archives similar with the candid photograph figure are found, which is stored into this personnel's archives.
Description
Technical field
The present invention relates to field of image processing more particularly to a kind of across channel polymerization sides in passerby library based on recognition of face
Method.
Background technique
Artificial intelligence technology is in security industry using more and more extensive, and wherein one of core is face recognition technology.With
Continuous landing of the face recognition application in security protection, have accumulated a large amount of data, one of them is passerby library.
So-called passerby library, be exactly security protection application access camera, the headend equipments such as face snap machine, by video decoding,
The technologies such as recognition of face identify the face in video flowing, and store storage and formed.
The retrieving in passerby library expends a large amount of GPU, cpu resource, and result inaccuracy, it is therefore desirable to passerby library into
Row polymerization filing.And traditional face aggregation scheme that there is polymerization effects is bad, polymerization efficiency is low, picture quality influence it is big
The problem of.Therefore, problem above is only solved, preferably precision data could be provided for upper layer application and support.
Summary of the invention
Object of the present invention is in view of the above-mentioned problems, improve it is a kind of improve polymerization effect the passerby library based on recognition of face across
Channel polymerization.
To achieve the goals above, the technical scheme is that
A kind of across channel polymerization in passerby library based on recognition of face, comprising the following steps:
S1, retain in a person ownership candid photograph figure quality highest one by the polymerization of single channel in short-term;
S11, the information cache queue of candid photograph figure is cached;
S12, when another candid photograph figure caching is come in, compared with candid photograph figure before, before its quality is higher than
Candid photograph figure when, by the candid photograph figure replace it is former capture figure, and the information cache queue of former candid photograph figure is replaced;When its quality
When lower than candid photograph figure before, the candid photograph figure is abandoned;When no similar candid photograph figure compares, new information cache is generated
Queue simultaneously caches;
S13, when reach set cache-time after, enable the candid photograph figure of caching enter across channel converging operation, while deleting slow
It deposits;
S2, it will be captured and schemed in storage route people library by across channel polymerization;
S21, in each personnel's archives in passerby library, select representative figure;
S22, setting compare threshold value, own entering in candid photograph figure and passerby library across channel converging operation in step S13
The representative figure of personnel's archives is compared, and finds personnel's archives similar with the candid photograph figure;
S23, when not finding personnel's archives similar with candid photograph figure, then this candid photograph figure is established as a new archives;
When finding personnel's archives similar with the candid photograph figure, then the candid photograph figure is stored into this personnel's archives;It is multiple when finding
Personnel's archives similar with the candid photograph figure then merge archives operation.
Further, in the step S11 information cache queue of candid photograph figure include the PID of candid photograph figure, feature, quality with
And enter the time data of caching;Wherein PID indicates personnel ID, for distinguishing different personnel.
Further, in the step S12 when capturing figure replacement original candid photograph figure, the PID of the former candid photograph figure of reservation grabs original
Feature, quality and the time data into caching of bat figure are replaced.
Further, personnel's archives in the step S21 are made of multiple capture pictures of a people, and have personnel
Archives ID.
Further, the selection mode of the representative figure in the step S21 are as follows: calculate the whole of same personnel's archives and capture
The representative score of figure represents the smallest M of score representative figures of the photo as personnel's archives.
Further, the representative is scored at the candid photograph figure and other candid photograph figures in personnel's archives where the candid photograph figure
Distance average: i-th is captured the distance calculation formula that figure captures figure with jth are as follows:
Wherein, sij∈ { 0~1 } is i-th and captures the similarity score that figure captures figure with jth, weight w0;mi, 1
∈ { 0,1 } is the quality of human face image of i-th candid photograph figure, weight w1;mi,2∈ { 0,1,2,3,4,5 } is i-th candid photograph figure
Personnel's age, weight w2;mi,3∈ { 0,1 } is personnel's gender of i-th candid photograph figure, weight w3;mi,4∈ { 0,1 } is
The personnel of i-th candid photograph figure are national, weight w4;dijTo capture figure i and capturing at a distance from figure j.
Further, a threshold value is set up, candid photograph figure of the score lower than the threshold value will be represented in personnel's archives from the personnel
It is rejected in archives.
Further, archives operation is merged in the step S23 are as follows: the representative figure of multiple personnel's archives is compared,
It determines whether its similarity reaches preset threshold value, when its similarity reaches threshold value, is determined as same personnel archives simultaneously
Merge into an archives;When its similarity does not reach threshold value, all candid photograph figures in multiple personnel's archives are compared;
When all comparison similarities for capturing figure are greater than threshold value, it is determined as same personnel archives and merges into an archives;When all
When capturing the comparison similarity of figure less than threshold value, merge without archives.
Further, in the step S23, when finding multiple personnel's archives similar with the candid photograph figure, and merge shelves
When case operation is determined as merging without archives, the highest personnel's archives of similarity in the candid photograph figure and multiple personnel's archives are selected
It is stored.
Further, in merging archives operation, after multiple personnel's archives merge as personnel's archives, weight
Newly select the representative figure of personnel's archives.
Compared with prior art, the advantages and positive effects of the present invention are:
(1) polymerization effect is significantly improved;The present invention captures figure replacement low-quality using high quality in the polymerization of single channel in short-term
Amount captures the technical solution of figure, so that it, which polymerize accuracy rate, is higher than 90%, effectively increases the polymerization that passerby library polymerize across channel
Effect.
(2) polymerization efficiency is improved;It is basically reaching in region after the number upper limit, the scale of personnel's archives reaches in passerby library
To basicly stable, retrieval radix is not further added by;It can be further decreased simultaneously by capture pictures only more a small amount of than in same archives
Retrieve radix, hence it is evident that improve the polymerization efficiency that passerby library polymerize across channel.
(3) picture of poor quality is rejected;The present invention can in such a way that threshold value is set, quality fuzzy to some images
Low capture pictures are rejected, and the polymerization effect that passerby library polymerize across channel is further improved.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention without any creative labor, may be used also for those of ordinary skill in the art
To obtain other drawings based on these drawings.
Fig. 1 is block flow diagram of the invention;
Fig. 2 is passerby library clean-up process figure;
Fig. 3 is the block flow diagram of single channel polymerization in short-term;
Fig. 4 is the periodic cleaning flow chart of single channel polymerization in short-term;
Fig. 5 is to merge archives operational flowchart.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.
As shown in Figures 1 to 5, technical solution of the present invention is divided into two steps, as shown in Figure 1:
1. single channel polymerize in short-term
One people will form multiple candid photograph figures in a channel, and angle and posture are continuous, preferably passes through polymerization in short-term and protects
Quality highest one is stayed, single channel polymerization process is as shown in Figure 3 in short-term:
Retain a buffer queue for capturing figure information, saves the PID for capturing figure, feature, quality, and enter caching
The related datas such as time.Wherein PID indicates personnel ID, while recording personnel and first entering caching team for distinguishing different people
The time of column.
One candid photograph figure C comes in, and compares first with all candid photograph figures in caching, attempts acquisition score highest and is more than to set
Determine threshold value history and capture figure, is denoted as herein.As existed, then picture quality is compared to each other.If new candid photograph plot quality is higher, use
New candid photograph figure replacement, the PID adopted, more new feature, quality and the time for entering caching.Otherwise this candid photograph figure is abandoned.If not
In the presence of then generating new PID and identify this candid photograph figure, record first enters the time of caching, saves and enters buffer queue.
Buffer queue is inspected periodically and clears up, as shown in Figure 4:
A. the time for first entering caching is checked to guarantee timeliness for the personnel for resting on camera shooting head region always,
More than the time of setting, it polymerize into across channel, while deletes caching.
B. for the personnel for leaving camera shooting head region, the time for entering caching is checked, more than the time of setting, into across logical
Road polymerization, while deleting caching.
2. across the full library polymerization in channel
Picture is captured after single channel polymerization, into across channel polymerization process, basic procedure is as follows:
(1) certain comparison threshold value is set, candid photograph figure is compared with all personnel's archives in passerby library.To improve efficiency, one
All candid photograph figures in a personnel's archives only represent figure and participate in comparison process.
Multiple capture pictures of one people form personnel's archives, and have archives ID.
By election mechanism, M representatives of the photo as archives in personnel's archives are selected, in the mistake of document retrieval
Cheng Zhong only represents figure and participates in comparing.
By certain algorithm, the representative score of whole candid photograph figures of same archives is calculated, the smallest M of score photos are made
For the representative figure of personnel's archives.It is as follows wherein to represent score computation method:
I-th candid photograph is schemed, following index is obtained:
I-th is captured the similarity score that figure captures figure with jth, is denoted as: sij∈ { 0~1 }, weight is denoted as: w0;
Face quality, is denoted as: mi, 1∈ { 0,1 }, weight is denoted as: w1;
Age is divided into six classes (children, juvenile, teenager are young, the middle age, old), is denoted as: mi,2∈{0,1,2,3,4,
5 }, weight is denoted as: w2;
Gender is divided into two classes (male, female), is denoted as: mi,3∈ { 0,1 } weight is denoted as: w3;
Ethnic group is divided into two classes (Han nationality, Uygurs), is denoted as: mi,4∈ { 0,1 } weight is denoted as: w4;
It captures figure i and captures the distance definition of figure j are as follows:
Whole candid photograph figures of same personnel's archives are mutually compared, and the representative score of every candid photograph figure is that it is grabbed with every other
Clap the average value of the distance of figure.
(2) if comparison result does not retrieve the archives of the condition of satisfaction, this, which is captured, forms new personnel's archives.
If only retrieving personnel's archives, this, which is captured, enters this personnel's archives.If retrieving multiple personnel's archives, these
The triggering of personnel's archives merges archives operation.
Assuming that certain threshold value T, it is as shown in Figure 5 to close shelves process: judgement is compared to two personnel's archives, if met
Shelves condition is closed, then is combined into personnel's archives, while re-electing representative figure.
(3) if closing shelves success, this, which is captured, enters this personnel's archives.If it is unsuccessful to close shelves, this, which is captured, enters similarity
Highest personnel's archives.
(4) it captures and enters any one personnel's archives, all triggerings re-elect representative figure.
A timed task is set, carries out archives cleaning operation, as shown in Figure 2.The purpose of cleaning is further rejecting matter
Measure bad candid photograph figure.By election mechanism, all representative scores for capturing figure in available personnel's archives.Setting one
A threshold value, the candid photograph lower than this threshold value are rejected from the archives, complete cleaning operation.
Beneficial effects of the present invention:
(1) polymerization effect is significantly improved;The present invention has in 2,400,000 capture pictures in actual candid photograph scene
It polymerize accuracy rate and is higher than 90%.
Accuracy rate calculation method: the candid photograph quantity of P is most in an archives, then the artificial P of Archiveownership.If there is multiple archives
Ownership people be all P, then the few archives of the candid photograph amount of P be calculated as without ownership people.Any one is captured, the personnel's shelves being distributed
The ownership people of case is really me, is calculated as polymerization correctly, is otherwise calculated as mistake.Assuming that always capturing number is calculated as n, polymerization is correctly grabbed
Umber of beats is c, then accuracy are as follows:
E=c/n*100%.
(2) polymerization efficiency is improved;It is basically reaching in region after the number upper limit, the scale of passerby library archives reaches basic
Stablize, retrieval radix is not further added by, while can further decrease retrieval base by representative figure only more a small amount of than in same archives
Number improves polymerization efficiency.
(3) picture of poor quality is rejected;It can find side face with large scale, obscure and the candid photograph of poor qualities is waited to shine
Piece, and rejected.
Claims (10)
1. a kind of across channel polymerization in passerby library based on recognition of face, it is characterised in that: the following steps are included:
S1, retain in a person ownership candid photograph figure quality highest one by the polymerization of single channel in short-term;
S11, the information cache queue of candid photograph figure is cached;
S12, when another candid photograph figure caching is come in, compared with candid photograph figure before, grabbing before its quality is higher than
When clapping figure, former candid photograph figure is replaced by the candid photograph figure, and be replaced to the information cache queue of former candid photograph figure;When its quality is lower than
When candid photograph figure before, the candid photograph figure is abandoned;When no similar candid photograph figure compares, new information cache queue is generated
And it caches;
S13, when reach set cache-time after, enable the candid photograph figure of caching enter across channel converging operation, while deleting caching;
S2, it will be captured and schemed in storage route people library by across channel polymerization;
S21, in each personnel's archives in passerby library, select representative figure;
S22, setting compare threshold value, and all personnel in candid photograph figure and passerby library across channel converging operation will be entered in step S13
The representative figure of archives is compared, and finds personnel's archives similar with the candid photograph figure;
S23, when not finding personnel's archives similar with candid photograph figure, then this candid photograph figure is established as a new archives;When looking for
To personnel's archives similar with the candid photograph figure, then the candid photograph figure is stored into this personnel's archives;When find it is multiple with should
It captures and schemes similar personnel's archives, then merge archives operation.
2. across the channel polymerization in passerby library based on recognition of face as described in claim 1, it is characterised in that: the step
The information cache queue of candid photograph figure includes PID, feature, quality and the time data for entering caching of candid photograph figure in S11;Wherein
PID indicates personnel ID, for distinguishing different personnel.
3. across the channel polymerization in passerby library based on recognition of face as claimed in claim 2, it is characterised in that: the step
In S12 when the former candid photograph of replacement is schemed in candid photograph to be schemed, retain the former PID for capturing figure, by the feature of original candid photograph figure, quality and enters slow
The time data deposited are replaced.
4. across the channel polymerization in passerby library based on recognition of face as described in claim 1, it is characterised in that: the step
Personnel's archives in S21 are made of multiple capture pictures of a people, and have personnel's archives ID.
5. across the channel polymerization in passerby library based on recognition of face as described in claim 1, it is characterised in that: the step
The selection mode of representative figure in S21 are as follows: the representative score for calculating whole candid photograph figures of same personnel's archives represents score minimum
M representative figures of the photo as personnel's archives.
6. across the channel polymerization in passerby library based on recognition of face as claimed in claim 5, it is characterised in that: the representative
Be scored at the distance average of other candid photograph figures in the candid photograph figure and personnel's archives where the candid photograph figure: i-th candids photograph figure and
Jth captures the distance calculation formula of figure are as follows:
Wherein, sij∈ { 0~1 } is i-th and captures the similarity score that figure captures figure with jth, weight w0;mI, 1∈{0,
It 1 } is the quality of human face image of i-th candid photograph figure, weight w1;mi,2∈ { 0,1,2,3,4,5 } is the people of i-th candid photograph figure
Member's age, weight w2;mi,3∈ { 0,1 } is personnel's gender of i-th candid photograph figure, weight w3;mi,4∈ { 0,1 } is i-th
The personnel for capturing figure are national, weight w4;dijTo capture figure i and capturing at a distance from figure j.
7. across the channel polymerization in passerby library based on recognition of face as claimed in claim 6, it is characterised in that: set up one
Threshold value will represent candid photograph figure of the score lower than the threshold value and reject from personnel's archives in personnel's archives.
8. across the channel polymerization in passerby library based on recognition of face as described in claim 1, it is characterised in that: the step
Merge archives operation in S23 are as follows: the representative figure of multiple personnel's archives is compared, determines whether its similarity reaches and sets in advance
Fixed threshold value is determined as same personnel archives and merges into an archives when its similarity reaches threshold value;When its similarity does not have
Have when reaching threshold value, all candid photograph figures in multiple personnel's archives are compared;When all comparison similarities for capturing figure are big
When threshold value, it is determined as same personnel archives and merges into an archives;When all comparison similarities for capturing figure are less than threshold value
When, merge without archives.
9. across the channel polymerization in passerby library based on recognition of face as claimed in claim 8, it is characterised in that: the step
In S23, when finding multiple personnel's archives similar with the candid photograph figure, and merges archives operation and be determined as merging without archives
When, select the candid photograph figure to store with the highest personnel's archives of similarity in multiple personnel's archives.
10. across the channel polymerization in passerby library based on recognition of face as claimed in claim 8, it is characterised in that: the conjunction
And in archives operation, after multiple personnel's archives merge as personnel's archives, the representative of personnel's archives is reselected
Figure.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910715542.XA CN110413811A (en) | 2019-08-05 | 2019-08-05 | A kind of across channel polymerization in passerby library based on recognition of face |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910715542.XA CN110413811A (en) | 2019-08-05 | 2019-08-05 | A kind of across channel polymerization in passerby library based on recognition of face |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110413811A true CN110413811A (en) | 2019-11-05 |
Family
ID=68365985
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910715542.XA Pending CN110413811A (en) | 2019-08-05 | 2019-08-05 | A kind of across channel polymerization in passerby library based on recognition of face |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110413811A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111160200A (en) * | 2019-12-23 | 2020-05-15 | 浙江大华技术股份有限公司 | Method and device for establishing passerby library |
CN111522980A (en) * | 2020-04-26 | 2020-08-11 | 深圳惟远智能技术有限公司 | File-based quick search method under mass data |
CN113269022A (en) * | 2021-03-24 | 2021-08-17 | 七台河市公安局 | System and method for predicting recessive specific personnel |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160335482A1 (en) * | 2014-11-07 | 2016-11-17 | Noblis, Inc. | Vector-based face recognition algorithm and image search system |
CN107566736A (en) * | 2017-09-30 | 2018-01-09 | 努比亚技术有限公司 | A kind of grasp shoot method and mobile terminal |
CN107808152A (en) * | 2017-11-23 | 2018-03-16 | 高域(北京)智能科技研究院有限公司 | Lift the method and face identification system of face recognition accuracy rate |
CN108052925A (en) * | 2017-12-28 | 2018-05-18 | 江西高创保安服务技术有限公司 | A kind of cell personnel archives intelligent management |
CN108733819A (en) * | 2018-05-22 | 2018-11-02 | 深圳云天励飞技术有限公司 | A kind of personnel's archives method for building up and device |
CN108875778A (en) * | 2018-05-04 | 2018-11-23 | 北京旷视科技有限公司 | Face cluster method, apparatus, system and storage medium |
CN109145844A (en) * | 2018-08-29 | 2019-01-04 | 北京旷视科技有限公司 | Archive management method, device and electronic equipment for city safety monitoring |
CN109241378A (en) * | 2018-08-29 | 2019-01-18 | 北京旷视科技有限公司 | Archives method for building up, device, equipment and storage medium |
CN109740004A (en) * | 2018-12-28 | 2019-05-10 | 上海依图网络科技有限公司 | A kind of archiving method and device |
CN109815369A (en) * | 2018-12-28 | 2019-05-28 | 上海依图网络科技有限公司 | A kind of archiving method and device |
CN109840446A (en) * | 2017-11-24 | 2019-06-04 | 华平信息技术股份有限公司 | Management control method, the apparatus and system of plurality of human faces identification engine |
CN110059657A (en) * | 2019-04-25 | 2019-07-26 | 北京旷视科技有限公司 | Records handling method, apparatus, electronic equipment and computer readable storage medium |
-
2019
- 2019-08-05 CN CN201910715542.XA patent/CN110413811A/en active Pending
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160335482A1 (en) * | 2014-11-07 | 2016-11-17 | Noblis, Inc. | Vector-based face recognition algorithm and image search system |
CN107566736A (en) * | 2017-09-30 | 2018-01-09 | 努比亚技术有限公司 | A kind of grasp shoot method and mobile terminal |
CN107808152A (en) * | 2017-11-23 | 2018-03-16 | 高域(北京)智能科技研究院有限公司 | Lift the method and face identification system of face recognition accuracy rate |
CN109840446A (en) * | 2017-11-24 | 2019-06-04 | 华平信息技术股份有限公司 | Management control method, the apparatus and system of plurality of human faces identification engine |
CN108052925A (en) * | 2017-12-28 | 2018-05-18 | 江西高创保安服务技术有限公司 | A kind of cell personnel archives intelligent management |
CN108875778A (en) * | 2018-05-04 | 2018-11-23 | 北京旷视科技有限公司 | Face cluster method, apparatus, system and storage medium |
CN108733819A (en) * | 2018-05-22 | 2018-11-02 | 深圳云天励飞技术有限公司 | A kind of personnel's archives method for building up and device |
CN109145844A (en) * | 2018-08-29 | 2019-01-04 | 北京旷视科技有限公司 | Archive management method, device and electronic equipment for city safety monitoring |
CN109241378A (en) * | 2018-08-29 | 2019-01-18 | 北京旷视科技有限公司 | Archives method for building up, device, equipment and storage medium |
CN109740004A (en) * | 2018-12-28 | 2019-05-10 | 上海依图网络科技有限公司 | A kind of archiving method and device |
CN109815369A (en) * | 2018-12-28 | 2019-05-28 | 上海依图网络科技有限公司 | A kind of archiving method and device |
CN110059657A (en) * | 2019-04-25 | 2019-07-26 | 北京旷视科技有限公司 | Records handling method, apparatus, electronic equipment and computer readable storage medium |
Non-Patent Citations (2)
Title |
---|
TIANYI ZHAO,ET AL: "Semi-automatic Data Annotation Tool for Person Re-identification Across Multi Cameras", 《2018 IEEE INTERNATIONAL CONFERENCE ON BIG DATA》 * |
肖轶涛,张义泰: "基于人像识别技术在"一人一档"系统中的应用与探索", 《数字技术与应用》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111160200A (en) * | 2019-12-23 | 2020-05-15 | 浙江大华技术股份有限公司 | Method and device for establishing passerby library |
CN111160200B (en) * | 2019-12-23 | 2023-06-16 | 浙江大华技术股份有限公司 | Method and device for establishing passerby library |
CN111522980A (en) * | 2020-04-26 | 2020-08-11 | 深圳惟远智能技术有限公司 | File-based quick search method under mass data |
CN113269022A (en) * | 2021-03-24 | 2021-08-17 | 七台河市公安局 | System and method for predicting recessive specific personnel |
CN113269022B (en) * | 2021-03-24 | 2022-02-08 | 七台河市公安局 | System and method for predicting recessive specific personnel |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110413811A (en) | A kind of across channel polymerization in passerby library based on recognition of face | |
US8869198B2 (en) | Producing video bits for space time video summary | |
CN105100894B (en) | Face automatic labeling method and system | |
Özyer et al. | Human action recognition approaches with video datasets—A survey | |
KR20170128771A (en) | Entity-based temporal segmentation of video streams | |
US10104345B2 (en) | Data-enhanced video viewing system and methods for computer vision processing | |
CN110516586A (en) | A kind of facial image clustering method, system, product and medium | |
CN108564052A (en) | Multi-cam dynamic human face recognition system based on MTCNN and method | |
JP2011505601A (en) | Video processing method and video processing apparatus | |
CN105138953B (en) | A method of action recognition in the video based on continuous more case-based learnings | |
CN109145844A (en) | Archive management method, device and electronic equipment for city safety monitoring | |
US20140093174A1 (en) | Systems and methods for image management | |
Xu et al. | Fusion of AV features and external information sources for event detection in team sports video | |
CN109934104A (en) | The pedestrian retrieval method and system across camera lens identified again based on pedestrian | |
CN109977782A (en) | Across shop operation detection method based on target position information reasoning | |
Xu et al. | Fine-grained instance-level sketch-based video retrieval | |
CN108427713A (en) | A kind of video summarization method and system for homemade video | |
CN114973317A (en) | Pedestrian re-identification method based on multi-scale adjacent interaction features | |
CN110750663B (en) | Cross-modal image retrieval method for life records | |
WO1999005865A1 (en) | Content-based video access | |
US11893084B2 (en) | Object detection systems and methods including an object detection model using a tailored training dataset | |
CN110263744A (en) | The method for improving noninductive face identification rate | |
Zhang et al. | Human pose estimation based on parallel atrous convolution and body structure constraints | |
Sanabria et al. | Profiling actions for sport video summarization: An attention signal analysis | |
Shirahama et al. | Kobe University and Kindai University at TRECVID 2018 AVS Task. |
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 | ||
AD01 | Patent right deemed abandoned | ||
AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20230224 |