AUTOMATED ATTENDANCE MACHINE USING...
Transcript of AUTOMATED ATTENDANCE MACHINE USING...
Automated Attendance Machine; FaceDetection and Recognition.
Presented by:
Kitili Jackson Mwendwa.F17/1437/2011
INTRODUCTIONTraditional methods of monitoring student attendances are tedious,time consuming and prone to inaccuracies as some students often signfor their absent colleagues.
This project proposes the design and implementation of an automatedattendance management system using face detection and recognition.
This biometric system will consist of the following processes:• Enrollment process• Identification/Verification
SYSTEM DESIGNThe attendance management system:• Desktop module• Mobile module
The Desktop Module system is divided in to two related sub-systems, the training set manager and the face recognizer.
Training set manager components.• Image acquisition component• Face detection component.
Face recognizer Sub system• Training component• Image acquisition• Face detection and recognition• Attendance component
Functions of the two sub systems.
Database of
Faces (This
contains the
training set)
Image Acquisition
(Gets the input
image with the
human face )
Face Recognizer
Recognizes the
detected faces from the
trained data
Trains the recognizer
on the training set
Loads the training set
Shows the calculated
average face and the
eigenfaces
Connects to the faces
database
TRAINING SET
MANAGER
Connects to faces
Database
Loads the training
set to display present
faces
Deletes a face from
the training set
Updates a face in the
training set
Adds faces with
labels to the training
set
Face Detector
Component
(detects faces and
extracts them)
SYSTEM DESIGN
Face detection
Viola jones object detection frame work based on Haar Cascades.
Improvements:-
• scale increase rates,• minimum detection scale,• Canny pruning flag and• Minimum neighbors threshold.
An input variable array for group photos
SYSTEM DESIGN
SYSTEM DESIGNFace Recognition
Principal Component Analysis(PCA) based Eigen faces method wasused.
PCA uses an orthogonal transformation to convert a set of correlatedM faces in to a set of K faces of uncorrelated variables calledprincipal components or eigen faces.
The weight of the incoming unknown image was found and thencompared to the weights of those already in the system. If the inputimage's weight is over a given threshold it is considered to beunknown.
Input image is identified by finding the image in the database whoseweights are the closest to the weights of the input image. Theimage in the database with the closest weight will be returned as ahit to the user of the system.
Face Recognition Algorithm flow chart
Imageprocessing
andcomparison
module
Database
Recognise the faces
Compares withDatabase and fills upthe attendance
Cont’d …
Since all faces of the same person are heterogeneous,several images of the same person were chosen withdifferent facial expressions and under differentillumination conditions.
Tools
Mobile tools• OpenCV for Android• Android studio/Eclipse IDE
Desktop tools• Development tools Visual studio, MS Access• EmguCV library• Programming language C#
RESULTSFace Detection• Single and group frontal photos had 100% face detection
rate• Live camera feed – track and recognize.
RESULTSFace Recognition
Yale and Local faces yielded an average percentagerecognition rate of about 80%.
First few eigen faces and average face obtained wereenough to fully represent all the faces in the trainingset – image compression.
RESULTS• The labels of the returned/recognized faces
were utilized in populating the databasefields.
FUTURE SCOPE Anti spoofing techniques.
Additional module for verification-foolproof.
Automated attendance report generation.
Well structured attendance registers for each class.
THANK YOU