Scribd
Upload
37 views17 pages

Minor Presentation - II

Uploaded by

rehbarnaqvi0510
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
37 views17 pages

Minor Presentation - II

Uploaded by

rehbarnaqvi0510
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 17

ORIENTAL COLLEGE OF TECHNOLOGY

BHOPAL (M.P.)

“COMPUTER SCIENCE ENGINEERING (CYBER SECURITY)”


MINOR PROJECT PROGRESS PRESENTATION-II
ON
“FACE RECOGNITION SOFTWARE”
21/11/2022

GUIDED BY: PRESENTED BY:


PROF. ROOPALI SONI AYUSHI TOMAR
HOD (CSE-CYBER SECURITY) 0126CY201020
HARSHI VERMA
0126CY201027
Contents
 Introduction
 Literature Review
 Gap Analysis
 Objective
 Technology used
 Flow Chart & Description
 Proposed Methodology
 Work Plan
 Development stages Photographs
 Results & Discussions
 Conclusions
 References
Introduction

 Face recognition refers to the technology capable


of identifying or verifying a subject through an image,
video, or any audiovisual element of his face. Generally,
this identification is used to access an application,
system, or service and it works like a face scanner.
 It is a method of biometric identification that uses that
body measures, in this case, face and head, to verify the
identity of a person through its facial biometric pattern
and data. The technology collects a set of unique
biometric data of each person associated with their face
and facial expression to identify, verify and/or
authenticate a person.
Literature Review
 In 2016, in order to increase the recognition rate, genetic algorithms (GA) were
employed. With a recognition rate of 96.507%, the model was assessed by three
databases: ORL, UMIST, and Indbase. Additionally, the recognition of thermal faces
was improved using a multi-feature fusion method. Four steps made up the model:
feature extraction, local classification, computation of the feature weight vector, and
residual computation, dictionary creation, and sparse representation as a classifier for
each individual feature. The calculated result was 91.5%. Another approach was put out
to extract features from the images using high rank classes using dynamic low rank
representations and K-NN as the classifier. Three databases, Extended Yale B, ORL,
and AR, totaling 6814 pictures, tested the model.
 The typical rate of recognition was 92.7%. Another approach for face recognition
was Sparse Boosting Representation Classification (SBRC). Five databases—AR,
Extended Yale B, ORL, FERET, and LFW—were used to assess the model. For all
databases combined, the average face recognition rate was 76.296%.
Literature Review
In year 2017, Han Yui made a dataset, which included over 4 million photos.
Human annotators have, however, assigned gender designations and emotional
component labels to these photos. 5.76 was the new mean absolute error. When the
photos were annotated, the age categorization accuracy was 96.2%, compared to 61.3%
when the data were not annotated. Another model for recognizing newborn faces uses
deep convolution neural networks. There were certain restrictions, though. The
database has not been subjected to any separate pre-processing. Even though each
parameter has been altered and tested numerous times, it is possible that some
parameter combinations were not examined. Additionally, the model was examined
independently for posture and lighting issues. For facial recognition based on the
Bayesian framework (FRAB), a different technique known as age-variational was
employed to enhance performance by automatically estimating age. The experimental
FG-NET performance results of 0.41 were attained. Three databases, JAFFE, ORL, and
FERET, totaling 11968 pictures, were used to test the model. A 96.9766% success rate
for recognition was achieved.
Literature Review

In 2019, Roshan Shantaran Tavhare put up a technique


for "Real time/sleep/Drowsiness Detection." In this study, the
researchers first extract image frames and convert them to
grayscale. For the face and eye region, they use a facial
landmark detector. The calculation of the Eye State Ratio
(EAR) and comparison with a threshold determines whether
the eye is open or closed. It will sound an alarm if the driver or
other person is drowsy.
GAP ANALYSIS

 I’ve read the articles and research papers and found some
gap or missing points like:
 Low resolution
 Different expressions
 Scars or facial marks
 But still these papers have a great research scope as I also
took reference from these papers.
Objective

 To find a series of data of the same face in a set of training


images in a database.
 To carry-out the process in real-time.
 To recognize a person through the software.
Technology Used

 Visual Studio Code


 Machine Learning
 Python
Flow Chart & Description
Proposed Methodology
 LBPH-based Algorithm: The first step is to extract the image pattern with
the LBPH (Local Binary Patterns Histogram) algorithm. Then, two thresholds
are set to calculate the probability of face in the image pattern. After that, the
sliding window applied to identify the faces in given images and recognize
those faces.
Work Plan
S.NO Particulars Duration Outcome

1 Module1(Literature 2 weeks Found some gaps.


survey)

2 Module2(Planning) 1 week Came up with the idea for facial


recognition

3 Module3(Design) 5 days Made a design and figure out


methodology for the project

4 Module4(Implementa 4 weeks Implementation of the methodology


tion) (approx.) and completion of the project.
Results & Discussions

 On the basis our experiments, the results we got are that when the
software processes the image, it gives output as the identity of the
person whose image was given as an input.
 Person name is determined with respect to output of face recognition.
 This face recognition system algorithm performs fast and accurate
person name identification. Performance of skin segmentation is
improved with white balance correction and facial feature extraction
performance is improved with LoG filter.
 Accuracy of classification is improved with decreasing gradient value
of performance value.
 In the discussion we found that most of the people found this very
helpful. They agreed to the approach, they all agreed to its usage and
processing speed. Some people disagreed to its recognition method
and says not so useful.
Conclusion

 Face recognition system is very important in our daily life.


It possesses a really great advantage.
 It was very costly but now new technologies have evolved
and the cost of equipment is going down dramatically due to
the integration and the increasing processing power.
 Certain applications of face recognition technology are now
cost effective, reliable and highly accurate.
Applications

• Helps find missing people.


• Protects businesses against theft.
• Improves medical treatment.
• Strengthens security measures.
• Makes shopping more efficient.
• Reduces the number of touchpoints.
• Improves photo organization.
References

 Face recognition using Histograms of Oriented Gradients,


by G. Bueno & J. Salido in year 2009.
 Face recognition for authentication on mobile devices, by
Daniel Gonzalez & Esteban V. in year 2016.
 Face Detection: A Survey, by B.K. Low & Erik H. in year
2000.
Thank You

You might also like