AL-SAEED FACULTY OF ENGINEERING &

INFORMATION TECHNOLOGY

DEPARTMENT OF MECHATRONICS
TAIZ UNIVERSITY

FACE TRACKING AND AUTOMATIC ATTENDANCE MANAGEMENT

SYSTEM USING FACE RECOGNITION TECHNIQUES

BY:
01 Amer Al-basser
02 Amr Al-sabaeei
03 Hesham Al-khateeb
Supervisor:
04 Mohammed Moghalies Dr/ Ghazi Alnowaini
05 Alaa Alshargaby
 Outline
01 Introduction
Problem statement, Objectives, Importance

02 Related Work
Review of The Previous Studies, Research Gap

03 Method and Material

Dataset, Pre-processing, ML & DL Models

04 Results & Discussion

Machine & Deep Learning Results, discussion

05 Conclusion
What did We Conclude from Results

06 Future Work
Recommendation and Future Work
3. Method and Material
3.1 Steps of Our Project

01 FACE DETECTION

02 DATA GATHERING

04 TRAINING DATA

FACE TRACKING, RECOGNITION

05 AND ATTENDACE
3. Method and Material
FACE DETECTION
Access the Attendance Management System
Website
The attendance taking session can be started after
the lecturer selected the related date and timetable
ID for the current class session
Face Detection
The captured image will first undergo a face
detection algorithm to ensure the system can
identify a face in every portrait.
DATA GATHERING
Portrait Pre-processing
Captured image with a confirmed face detected in it
will then undergo cropping, color conversion before
actually being stored into the files.
Creation of CSV file
Then, by using the path of the stored image, a csv
file is created with each image being labelled
accordingly. The csv file is created to aid the
process of list insertion of the portraits with their
corresponding labels.
3. Method and Material
TRAINING DATA
Training the Recognizer
The images in the created list retrieved from the csv will
then be pumped into a recognizer which is the LPBH
Recognizer (a library provided by OpenCV) to do the
training.
Save the Trained Data
After the training process is done, the trained sets of
data will be stored into a (.yml) file which will be
retrieved during the recognition process to ensure the
training process are only done for the minimum time.
FACE TRACKING, RECOGNITION
AND ATTENDACE
Tracking and Recognition
Face Tracking generally works when the camera catches
the person face. After tracking a face or faces, the
haarCascade recognizer uses data from the
(Trainner.yml) file to predict whose face it is. The
LBPHFaceRecognizer in OpenCV takes a captured
portion of the face to be analyzed as a parameter and
returns its likely owner, including its id and how
confident the recognizer is in this match.
Taking Attendance
After it recognize the face, the date and time when the
face first appeared will be registered in a database and
an excel sheet.
3. Method and Material

3.2 Method ( Algorithm )

Local Binary Patterns Histogram (LBP) is defined as an ordered

collection of binary comparisons of pixel intensities between the
center pixel and its surrounding pixels, as introduced by Ojala et al.
[1]. LBP essentially recreates the local structure of a picture by
matching each pixel to its immediate surroundings. Take a pixel as
the center and set a threshold for its neighbors.
3. Method and Material

3.2 Extracting the Histograms

As we have a facial image in grayscale, each histogram (from each grid) will contain only
256 positions (0~255) representing the occurrences of each pixel intensity.

Then, we need to focus each histogram so that we can create a new histogram. Supposing
we’ve 8x8 grids, we will have 8x8x256=16.384 positions in the final histogram. The final
histogram represents the better characteristics of the initial image.
3. Method and Material

CNN Model

Output 1

Pool Pool Pool

Layer Layer Conv Layer
Conv Layer Output 2
Fully Connected
Conv Layer Layer
Layer
Input Image