CN109191341A - A kind of classroom video point based on recognition of face and Bayesian learning is to method - Google Patents

A kind of classroom video point based on recognition of face and Bayesian learning is to method Download PDF

Info

Publication number
CN109191341A
CN109191341A CN201810996596.3A CN201810996596A CN109191341A CN 109191341 A CN109191341 A CN 109191341A CN 201810996596 A CN201810996596 A CN 201810996596A CN 109191341 A CN109191341 A CN 109191341A
Authority
CN
China
Prior art keywords
face
picture
probability
classroom
bayesian
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.)
Granted
Application number
CN201810996596.3A
Other languages
Chinese (zh)
Other versions
CN109191341B (en
Inventor
温浩
陈江豪
石君
陈兰
万珺
周曦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangzhou Kaifeng Technology Co Ltd
Original Assignee
Guangzhou Kaifeng Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangzhou Kaifeng Technology Co Ltd filed Critical Guangzhou Kaifeng Technology Co Ltd
Priority to CN201810996596.3A priority Critical patent/CN109191341B/en
Publication of CN109191341A publication Critical patent/CN109191341A/en
Application granted granted Critical
Publication of CN109191341B publication Critical patent/CN109191341B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/20—Education
    • G06Q50/205—Education administration or guidance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Multimedia (AREA)
  • Educational Administration (AREA)
  • Educational Technology (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • General Business, Economics & Management (AREA)
  • Image Analysis (AREA)

Abstract

A kind of classroom video point based on recognition of face and Bayesian learning is to method, using following steps, step 1: processing module carries out head and shoulder detection to t moment individual picture in camera video stream, and demographics model the position coordinates of classroom location distribution map and everyone；If the student that attends class collection is combined into N, the number that should arrive is n=| N |, the collection that head and shoulder detects is combined into M (t), and number is m (t), then m (t)≤n；Step 2: feature extraction is carried out to the face in picture, and list of attending class n is retrieved according to face characteristic, confirm current collection k (t), it is iterated with previous pair picture confirming face set K (t-1), obtains confirmation set K (t)=k (t) ∪ K (t-1) of t moment；Without being transformed to hardware device, camera, existing major part camera can be met the requirements, and reduce user cost；Improve accuracy rate and speed that original recognition of face dynamic point arrives.Using the present invention, be able to solve at present to classroom student quickly, high-efficiency point to the problem of.

Description

A kind of classroom video point based on recognition of face and Bayesian learning is to method
Technical field
The present invention relates to field of face identification, and in particular to a kind of classroom video based on recognition of face and Bayesian learning Point arrives method.
Background technique
Traditional classroom is attended class, teacher and Educational Affairs Office all think it is fast and accurate it is true know which student does not come to class, thus Implement subsequent teaching management.If by teacher, student calls the roll one by one, to waste many times, the feelings of 50 people of number of student Under condition, 3-5 minutes time is substantially expended, and not necessarily accurate.If signed by student oneself, it may appear that student's allograph Situation.If swiped the card by installation, perhaps student also will be lined up before the class or after class and go to register fingerprint if face attendance recorder, It is very troublesome, and add additional cost.The recognition of face point based on video monitoring has been had already appeared in the market to method, with sea Kang Wei the monitoring manufacturers such as regards as representative, carries out feature extraction and student's list to the face that monitored picture real-time grasp shoot arrives on backstage Face database carries out 1:n retrieval, confirms one by one, final to realize that unaware point arrives.But this method depends critically upon whether student faces When front, especially student are being bowed always or facial angle is more inclined, it may appear that the case where leakage is grabbed, the point of individual picture To accuracy rate only 80% or so.Even according to prolonged monitored picture (student, which bows at random, to be occurred) come repeatedly Iteration recognition result also can only achieve 95% or so, and there are problems that taking long time at this time, generally want 15 minutes or so (100 Secondary iteration was calculated according to each iteration interval 10 seconds) result could be gone out.
Summary of the invention
In view of the deficiencies of the prior art, the present invention proposes a kind of classroom video point based on recognition of face and Bayesian learning To method, specific technical solution is as follows: a kind of classroom video point based on recognition of face and Bayesian learning to method, feature It is:
Using following steps,
Step 1: processing module carries out head and shoulder detection, demographics, modeling to t moment individual picture in camera video stream The position coordinates of classroom location distribution map and everyone；
If the student that attends class collection is combined into N, the number that should arrive is n=| N |, the collection that head and shoulder detects is combined into M (t), and number is m (t), Then m (t)≤n；
Step 2: feature extraction being carried out to the face in picture, and retrieves list of attending class n according to face characteristic, confirmation is worked as Preceding set k (t) and previous pair picture confirming face set K (t-1) are iterated, and obtain confirmation set K (t)=k of t moment (t)∪K(t-1)；
Step 3: judging whether K (t) is equal to N and otherwise, enters step 4 if it is, entering step 11；
Step 4: if K (t) < N, then M (t)-K (t) member object gathered being subjected to Bayesian analysis one by one；
Judge in classroom with the presence or absence of isolated island position, the definition of the isolated island position is, the position all around nobody, such as Fruit is to enter step 5, otherwise enters step 6；
Step 5: probability P (j) being attended according to the history of religion indoor occupant, M (t)-K (t) set personnel are ranked up；
Step 6: being the member having determined in target members position to be determined all around, if i is that adjacent seat has confirmed that The face of personnel, i=i1, i2..., i4∈K(t)；
J is for target members to be confirmed, j=j1..., js∈ M (t)-K (t), and adjacent probability is not zero, according to pattra leaves This new probability formula
P (A | B)=P (B | A) * P (A)/P (B), it obtains in the case where i is attended, the estimated probability that j is also attended:
Pn(j1|i1)=Pn (i1|j1)*P(j1)/P(i1)
Pn(j1|i1i2)=Pn (i1|j1)*P(j1)*P(i2|j1,i1)/P(i1)*P(i2|i1) iterate to calculate,
And so on obtain Pn (j1|i1i2i3i4), it is assumed that jsFor maximum Bayesian probability, s ∈ { 1,2,3,4 }, jsThen For the Bayesian Estimation of target members to be confirmed, estimation set B (t) is obtained；
Step 7: setting reanalyses new picture at the t+1 moment, is repeated step 1 to step by the number of iterations at least n times Rapid 6 at least n times；
Step 8: after the n times picture of iteration setting, if not reaching termination condition, stopping iteration, will finally estimate Count js, adjacent personnel be input in Bayesian network and learn and update, enter step 9, otherwise, enter step 10；
Step 9: turn out for work list M (t), K (t), B (t), M (t)=K (t)+B (t), and wherein K (t) is confirmation list, B It (t) is estimation list, which includes personnel and corresponding probability；
Step 10: illustrating that whole students have been acknowledged and finish, terminating epicycle point and arriving, output result is student attendance list N (t)=K (t)=M (t), and the personnel adjacent all around of turn out for work list and each student are input to the corresponding shellfish of the course Learn in this network of leaf, updates each student subject and attend probability P (i) and other students probability that have position adjacent Pn (i | j), i, j ∈ K terminate epicycle point and arrive.
Further: using based on deep learning CNN network in step 1 to t moment individual picture in camera video stream Carry out identifying processing.
The invention has the benefit that first, without being transformed to hardware device, camera, existing most of camera shooting Head can be met the requirements, and reduce user cost；
Second, improve accuracy rate and speed that original recognition of face dynamic point arrives.Using the present invention, it is able to solve at present To classroom student quickly, high-efficiency point to the problem of.
Detailed description of the invention
Fig. 1 is work flow diagram of the invention.
Specific embodiment
The preferred embodiments of the present invention will be described in detail with reference to the accompanying drawing, so that advantages and features of the invention energy It is easier to be readily appreciated by one skilled in the art, so as to make a clearer definition of the protection scope of the present invention.
A kind of classroom video point based on recognition of face and Bayesian learning is to method as shown in Figure 1:, it is characterised in that:
Using following steps,
Step 1: processing module carries out head and shoulder detection, demographics, modeling to t moment individual picture in camera video stream The position coordinates of classroom location distribution map and everyone；
If the student that attends class collection is combined into N, the number that should arrive is n=| N |, the collection that head and shoulder detects is combined into M (t), and number is m (t), Then m (t)≤n；
Step 2: feature extraction being carried out to the face in picture, and retrieves list of attending class n according to face characteristic, confirmation is worked as Preceding set k (t) and previous pair picture confirming face set K (t-1) are iterated, and obtain confirmation set K (t)=k of t moment (t)∪K(t-1)；
Step 3: judging whether K (t) is equal to N and otherwise, enters step 4 if it is, entering step 11；
Step 4: if K (t) < N, M (t)-K (t) member object gathered being subjected to Bayesian analysis one by one, the M (t)-K (t) is that M (t) removes the part of K (t), or is expressed as supplementary set CuK (t) of the K (t) in complete or collected works M (t)；
Judge in classroom with the presence or absence of isolated island position, the definition of the isolated island position is, the position all around nobody, such as Fruit is to enter step 5, otherwise enters step 6；
Step 5: probability P (j) being attended according to the history of religion indoor occupant, M (t)-K (t) set personnel are ranked up；
Step 6: being the member having determined in target members position to be determined all around, if i is that adjacent seat has confirmed that The face of personnel, i=i1, i2..., i4∈K(t)；
J is for target members to be confirmed, j=j1..., js∈ M (t)-K (t), and adjacent probability is not zero, according to pattra leaves This new probability formula
P (A | B)=P (B | A) * P (A)/P (B), it obtains in the case where i is attended, the estimated probability that j is also attended:
Pn(j1|i1)=Pn (i1|j1)*P(j1)/P(i1)
Pn(j1|i1i2)=Pn (i1|j1)*P(j1)*P(i2|j1,i1)/P(i1)*P(i2|i1) iterate to calculate,
And so on obtain Pn (j1|i1i2i3i4), it is assumed that jsFor maximum Bayesian probability, s ∈ { 1,2,3,4 }, jsThen For the Bayesian Estimation of target members to be confirmed, estimation set B (t) is obtained；
Step 7: setting reanalyses new picture at the t+1 moment, is repeated step 1 to step by the number of iterations at least n times Rapid 6 at least n times；
Step 8: after the n times picture of iteration setting, if not reaching termination condition, stopping iteration, will finally estimate Count js, adjacent personnel be input in Bayesian network and learn and update, enter step 9, otherwise, enter step 10；
Step 9: turn out for work list M (t), K (t), B (t), M (t)=K (t)+B (t), and wherein K (t) is confirmation list, B It (t) is estimation list, which includes personnel and corresponding probability；
Step 10: illustrating that whole students have been acknowledged and finish, terminating epicycle point and arriving, output result is student attendance list N (t)=K (t)=M (t), and the personnel adjacent all around of turn out for work list and each student are input to the corresponding shellfish of the course Learn in this network of leaf, updates each student subject and attend probability P (i) and other students probability that have position adjacent Pn (i | j), i, j ∈ K terminate epicycle point and arrive.
T moment individual picture in camera video stream is carried out at identification using based on deep learning CNN network in step 1 Reason.

Claims (2)

1. a kind of classroom video point based on recognition of face and Bayesian learning is to method, it is characterised in that:
Using following steps,
Step 1: processing module carries out head and shoulder detection to t moment individual picture in camera video stream, and demographics model classroom The position coordinates of location map and everyone；
If the student that attends class collection is combined into N, the number that should arrive is n=| N |, the collection that head and shoulder detects is combined into M (t), and number is m (t), then m (t)≤n；
Step 2: feature extraction being carried out to the face in picture, and retrieves list of attending class n according to face characteristic, confirms current collection It closes k (t) and previous pair picture confirming face set K (t-1) is iterated, obtain confirmation set K (t)=k (t) ∪ of t moment K(t-1)；
Step 3: judging whether K (t) is equal to N and otherwise, enters step 4 if it is, entering step 10；
Step 4: if K (t) < N, then M (t)-K (t) member object gathered being subjected to Bayesian analysis one by one；
Judge in classroom with the presence or absence of isolated island position, the definition of the isolated island position is, the position all around nobody, if it is 5 are then entered step, otherwise enters step 6；
Step 5: probability P (j) being attended according to the history of religion indoor occupant, M (t)-K (t) set personnel are ranked up；
Step 6: being the member having determined in target members position to be determined all around, if i is that adjacent seat has confirmed that personnel Face, i=i1, i2..., i4∈K(t)；
J is for target members to be confirmed, j=j1..., js∈ M (t)-K (t), and adjacent probability is not zero, it is general according to Bayes Rate formula
P (A | B)=P (B | A) * P (A)/P (B), it obtains in the case where i is attended, the estimated probability that j is also attended:
Pn(j1|i1)=Pn (i1|j1)*P(j1)/P(i1)
Pn(j1|i1i2)=Pn (i1|j1)*P(j1)*P(i2|j1,i1)/P(i1)*P(i2|i1) iterate to calculate,
And so on obtain Pn (j1|i1i2i3i4), it is assumed that jsFor maximum Bayesian probability, s ∈ { 1,2,3,4 }, jsThen for true The Bayesian Estimation of the target members recognized obtains estimation set B (t)；
Step 7: setting reanalyses new picture at the t+1 moment, is repeated step 1 to step 6 by the number of iterations at least n times At least n times；
Step 8: after the n times picture of iteration setting, if not reaching termination condition, stopping iteration, will finally estimate js、 Adjacent personnel, which are input in Bayesian network, to be learnt and updates, and is entered step 9, otherwise, is entered step 10；
Step 9: the list M (t) that turns out for work, confirmation list K (t), estimation list B (t), M (t)=K (t)+B (t), the estimation name Single B (t) includes personnel and corresponding probability；
Step 10: illustrate that whole students have been acknowledged and finish, terminate epicycle point arrive, output result for student attendance list N (t)= K (t)=M (t), and the personnel adjacent all around of turn out for work list and each student are input to the corresponding Bayesian network of the course Learn in network, update each student subject attend probability P (i) and other students had the adjacent probability P n in position (i | J), i, j ∈ K terminate epicycle point and arrive.
2. a kind of classroom video point based on recognition of face and Bayesian learning is to method, it is characterised in that: use base in step 1 Identifying processing is carried out to t moment individual picture in camera video stream in deep learning CNN network.
CN201810996596.3A 2018-08-29 2018-08-29 Classroom video frequency point arrival method based on face recognition and Bayesian learning Active CN109191341B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810996596.3A CN109191341B (en) 2018-08-29 2018-08-29 Classroom video frequency point arrival method based on face recognition and Bayesian learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810996596.3A CN109191341B (en) 2018-08-29 2018-08-29 Classroom video frequency point arrival method based on face recognition and Bayesian learning

Publications (2)

Publication Number Publication Date
CN109191341A true CN109191341A (en) 2019-01-11
CN109191341B CN109191341B (en) 2020-10-23

Family

ID=64916549

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810996596.3A Active CN109191341B (en) 2018-08-29 2018-08-29 Classroom video frequency point arrival method based on face recognition and Bayesian learning

Country Status (1)

Country Link
CN (1) CN109191341B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110533788A (en) * 2019-08-14 2019-12-03 合肥智圣新创信息技术有限公司 A kind of college student attendance checking system based on recognition of face and online real time data
CN110827432A (en) * 2019-11-11 2020-02-21 深圳算子科技有限公司 Class attendance checking method and system based on face recognition
CN110889672A (en) * 2019-11-19 2020-03-17 哈尔滨理工大学 Student card punching and class taking state detection system based on deep learning

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102831412A (en) * 2012-09-11 2012-12-19 魏骁勇 Teaching attendance checking method and device based on face recognition
CN104376611A (en) * 2014-10-20 2015-02-25 胡昔兵 Method and device for attendance of persons descending well on basis of face recognition
US20170287302A1 (en) * 2010-02-26 2017-10-05 Thl Holding Company, Llc Wireless device and methods for use in determining classroom attendance

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170287302A1 (en) * 2010-02-26 2017-10-05 Thl Holding Company, Llc Wireless device and methods for use in determining classroom attendance
CN102831412A (en) * 2012-09-11 2012-12-19 魏骁勇 Teaching attendance checking method and device based on face recognition
CN104376611A (en) * 2014-10-20 2015-02-25 胡昔兵 Method and device for attendance of persons descending well on basis of face recognition

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110533788A (en) * 2019-08-14 2019-12-03 合肥智圣新创信息技术有限公司 A kind of college student attendance checking system based on recognition of face and online real time data
CN110827432A (en) * 2019-11-11 2020-02-21 深圳算子科技有限公司 Class attendance checking method and system based on face recognition
CN110889672A (en) * 2019-11-19 2020-03-17 哈尔滨理工大学 Student card punching and class taking state detection system based on deep learning
CN110889672B (en) * 2019-11-19 2022-04-12 哈尔滨理工大学 Student card punching and class taking state detection system based on deep learning

Also Published As

Publication number Publication date
CN109191341B (en) 2020-10-23

Similar Documents

Publication Publication Date Title
Patil et al. Implementation of classroom attendance system based on face recognition in class
CN109191341A (en) A kind of classroom video point based on recognition of face and Bayesian learning is to method
CN102214291B (en) Method for quickly and accurately detecting and tracking human face based on video sequence
CN110135249B (en) Human behavior identification method based on time attention mechanism and LSTM (least Square TM)
CN110910549A (en) Campus personnel safety management system based on deep learning and face recognition features
CN109145717A (en) A kind of face identification method of on-line study
CN104992167A (en) Convolution neural network based face detection method and apparatus
CN105205646A (en) Automatic roll call system and realization method thereof
CN113903081A (en) Visual identification artificial intelligence alarm method and device for images of hydraulic power plant
CN109214263A (en) A kind of face identification method based on feature multiplexing
CN108921038A (en) A kind of classroom based on deep learning face recognition technology is quickly called the roll method of registering
CN110827432B (en) Class attendance checking method and system based on face recognition
CN109345427A (en) The classroom video point of a kind of combination recognition of face and pedestrian&#39;s identification technology is to method
CN104183029A (en) Portable quick crowd attendance method
CN112115796A (en) Attention mechanism-based three-dimensional convolution micro-expression recognition algorithm
CN107741996A (en) Family&#39;s map construction method and device based on recognition of face, computing device
CN111079720A (en) Face recognition method based on cluster analysis and autonomous relearning
CN111582195B (en) Construction method of Chinese lip language monosyllabic recognition classifier
CN112580778A (en) Job worker mobile phone use detection method based on YOLOv5 and Pose-animation
WO2012071677A1 (en) Method and system for face recognition
WO2020234737A1 (en) Localised, loop-based self-learning for recognising individuals at locations
Abusham Face verification using local graph stucture (LGS)
CN111444877B (en) Classroom people number identification method based on video photos
Autade et al. Automated Multi Face Recognition and Identification using Facenet and VGG-16 on Real-World Dataset for Attendance Monitoring System
CN103886304B (en) It is a kind of that recognition methodss are laughed at based on the true and false of space-time local description

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
CB02 Change of applicant information

Address after: 511458 room 1009, No.26, Jinlong Road, Nansha District, Guangzhou City, Guangdong Province (only for office use)

Applicant after: Guangzhou yuncongkaifeng Technology Co., Ltd

Address before: 511457 Room 1009, No. 26 Jinlong Road, Nansha District, Guangzhou City, Guangdong Province

Applicant before: GUANGZHOU KAIFENG TECHNOLOGY Co.,Ltd.

CB02 Change of applicant information
GR01 Patent grant
GR01 Patent grant