Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 23

WOLKITE UNIVERSITY

COLLEGE OF COMPUTING AND INFORMATICS

DEPARTMENT OF SOFTWARE ENGINEERING

SEMINAR PRESENTATION ON

Smart Card ID
WOLKITE UNIVERSITY, WOLKITE, ETHIOPIA
18 MARCH 2021
Seminar Presentation On Face Recognition By
NO NAME ID NO SECTION
1 ABDI JIRU CIR/301/11 B
2 CIR/310/10 B
3 CIR/300/10 B
4 CIR/284/10 B
5 CIR/338/10 B
6 CIR/315/10 B
Contents
01 Introduction

02 Face Recognition Processes

Face Detection
03

04 Pre-processing of Pictures

Face Recognition Algorithms


05

Applications of Face Recognition


06

Conclusion
07
“ What AI and Machine
Learning Allows You to Do is
Find the Needle in the Hay
Stick” BOB WORK
Challenges
How a machine recognizes our faces ?
Can machines replace humans ?
Why face recognition is preferred than other biometrics?
Face Recognition
Face recognition is a computer app
that is capable of detecting, It is one of the leading
tracking, identifying or biometric technologies.
verifying human faces from an Why over other biometrics?
image or video capturing using
a digital camera.
Face Recognition is
It is a technique used by non-intrusive method
computer algorithms to by identifying
identify or verify a individuals by the
person or an object feature of faces.
through images.

It is grounded in machine learning


It is a technology capable of algorithms which capture, store and
recognizing a person based analyze facial features in order to match
on their face. them with images of individuals in a
pre-existing database and, often,
information about them in that
Face Recognition Process

Face Training Classification


Images Pre-processing
Detection
Capture

The process of The process of


It has the The process of The process of
capturing the frontal predicting an
objective of standardizing the training the
images of the identity of a given
finding the faces extracted image of machine with
persons using person based on
(location and size) face. standardized
different kinds of the trained dataset.
in an image. training dataset.
cameras.
Face Detection
Before recognizing a face, it is first essential to detect and extract the faces from the original pictures. For
recognizing a face, the algorithms compare only faces. Any other element in the picture that is not part of a
face deteriorates the recognition. There are several existing algorithms for detecting faces.
1. Haar-cascade classifier
2. Histogram of Oriented Gradients (HOG)

Visualization of the HOG detector from the library


Pre-processing of Pictures
Most of the algorithms for facial recognition require the same size for the entire training set. But not all the pictures
have the same zoom on the face and have maybe not all the same size. Pre-processing includes different modifications.

First of all, the faces need The aim is to have the


to be centered in the eyes at the same level and
picture in the same way. the nose at the same
position for all images.

The aim is to find the


The location of the two appropriate size which is as
eyes and the nose is often small as possible in order to
used as a landmark for improve the computing time
centering faces. and keep enough information
for recognizing a person.
To apply these modifications, the coordinates of the landmarks are needed.

Visualization of the 22 landmarks


Facial Recognition’s Algorithms
There are several approaches for recognizing a face.

Some algorithms
can use a
convolutional
Some algorithms neural network.
Some algorithms
can use statistical
try to find a pattern
approach.
which represents a
specific person
Facial Recognition’s Algorithms
Fisher Faces Local Binary
Open Face
Patterns
Eigen Faces This algorithm is It is a face recognition library and
Histograms
a modification of it uses a deep convolutional neural
It is a method for It can represent
Eigen faces, thus network for performing a facial
performing facial local features in the
also uses Principal recognition.
recognition based images. It is more efficient for performing a
Components
on a statistical It is possible to
Analysis. facial recognition.
approach. get great results
The main
The aim of this (mainly in a
modification is that
method is to extract controlled
Fisher faces takes
the principal environment ).
into consideration
components which classes.
affect the most the
variation of the
images.

Based on level of Efficiency of the algorithms


Comparing the Algorithms
For comparing
Eigen faces the algorithms, different factors can be tested. Several tests will be run with different modifications in
the training set such as the number of subjects, the number of pictures per subject and a change in the luminosity.
Training data and test data in the same environment : 10 pictures per subject
100%
90%
80%
70%
Classification 60%
Accuracy 50% 100
92 92 96
40%
30%
20%
10%
0%
Eigen Faces Fisher Faces LBPH Open Faces

Face Recognition Algorithms


Face Recognition Algorithms
Training
Eigen faces data and test data in the different environments : 40 pictures per subject

100%
90%
80%
70%
Classification 60%
Accuracy 50% 100
40%
30% 60
50
20% 40
10%
0%
Eigen Faces Fisher Faces LBPH Open Faces

Face Recognition Algorithms


General Questions
How the Google search engine
shows the pictures that related
with our search items ?
How Amazon suggests products
to us ?
How the face book suggests
tagging person for the user ?
Applications of Face Recognition
Security and Defense
Personal Mobile Devices
Entry To Sensitive Areas
Law enforcement Healthcare
 access patient’s medical records
 monitor and diagnose certain diseases.

Retail and Marketing Face


•Prevent Shoplifting Recognition
 reduce shoplifting by over 30%
•Promote products and more
accurately target advertisements Hospitality
 suggesting skincare and cosmetic record customer preferences
products provide better security
•Collecting Data automate check-ins and check outs.
 gathering age, gender, and
ethnicity to more effectively target
their advertisements
Security and Defense
Entry To Sensitive Areas Personal Mobile Devices
Healthcare
Monitor and Diagnose Certain Diseases. Diagnosing Certain Diseases
Conclusion
Generally, Facial recognition is the process of identifying or verifying the identity of a person using their face. It
captures, analyzes, and compares patterns based on the person's facial details. Facial biometrics continues to be the
preferred biometric benchmark.

Like most innovations that have come before it, facial recognition technology has caused a stir. Applications of
facial recognition in Ethiopia constantly grow and after a few decades it might be a part of every aspect of our
life. Facial recognition in Ethiopia have been using in a many sectors such as Bank, hospital , and etc.

Like it or not, facial recognition systems are here to stay. They will keep us safer and make other parts of our
lives more interactive. Although the technology is young and not perfect , companies like FaceX look to improve
ourselves every day to serve up nothing but the best to customers and the general public.

We may not be psychic and we can’t tell what the future holds, but it looks bright and that is where we are headed.
References
4 Major Applications of Facial Recognition | History | Data Used – FaceX
A comparison of facial recognition’s algorithms Nicolas Delbiaggio
Facial Recognition in 2021 (with examples) (thalesgroup.com)
face recognition - Bing images
Seminar Presenters’ Details

Your Picture Here


And Send To Back

TIGIST ABERA ABEBAYEHU ALARO SARON DANIEL ABAS HAJI ABDILFETA MOHAMMED WAKOYA ACHILU

You might also like