Scribd
Upload
37 views

Technical Seminar On: Face Recognition Based On Convolution Neural Network

This document summarizes a technical seminar presentation on face recognition using convolutional neural networks. The presentation outlines the objectives of establishing simulations for occluded and oriented faces. It discusses the existing methods using PCA and LDA that have disadvantages for uncontrolled images. The proposed system uses a CNN with convolutional, pooling, ReLU and fully connected layers. An experiment on the Georgia Tech database showed higher recognition rates than other algorithms with less computation time.

Uploaded by

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

Technical Seminar On: Face Recognition Based On Convolution Neural Network

This document summarizes a technical seminar presentation on face recognition using convolutional neural networks. The presentation outlines the objectives of establishing simulations for occluded and oriented faces. It discusses the existing methods using PCA and LDA that have disadvantages for uncontrolled images. The proposed system uses a CNN with convolutional, pooling, ReLU and fully connected layers. An experiment on the Georgia Tech database showed higher recognition rates than other algorithms with less computation time.

Uploaded by

Kushal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

Akshaya Institute of

Technology ,Tumkur

Technical Seminar on:


Face Recognition based
on Convolution Neural
Network

By: CHANDAN BG
USN: 1AK17CS003
Outline
• Introduction
• Objectives
• Existing System
• Disadvantages
• Proposed System
• Advantages
• Methodology
• Experimental Analysis
• Conclusion
• References
Introduction

• Face recognition is the method of recognizing the


face of a relevant person by a vision system.
• It makes use of unique facial characteristics.
• Most face recognition algorithms are designed to
work best well aligned and frontal pose face image.
• It has a great importance in real world applications
such as video surveillance, human machine
interaction ,security systems and attendance
system.
Objectives

• To establish simulation of all possible face


occlusions and face posture orientation.
• To determine a state of the art techniques dealing
with facial occlusion , face posture orientation and
face scale variation.
• To determine the real time face recognition
system by using CNN.
Existing System

• In existing method Principle Component Analysis


(PCA),Linear Discriminant Analysis(LDA) are used for
feature extraction.
• Support Vector Machine or Artificial Neural Network
are used for classifier.
• No knowledge of geometry and transparency of
faces are required.
• These models are not suitable for uncontrolled face
images.
Disadvantages

• Its recognition rate decreases for recognition


under varying pose and lighting of the image.
• It consumes more time for feature extraction and
classification of the images.
• Training large dataset is unsolved in SVM.
• While solving an image classification problem
using ANN, the first step is to convert a 2-
dimensional image into a 1-dimensional vector
prior to training the model.
Proposed System
• In proposed system Convolutional Neural
Network (CNN) recognition algorithm is used.
It is mainly carried out three steps i.e.,
1. Resizing the input image as 16x16x1,
16x16x3,32x32x1,32x32x3,64x64x1,64x64x3.
2. Build a CNN Structure of 8 layers made up of
Convolution layer and max pooling layers.
3. After extracting all features ,use softmax
classifier for classification.
Cntd…

• The block schema of the proposed CNN recognition algorithm is given in Fig.
Advantages
• CNN captures the spatial features from an
image.
• CNN follows the concept of parameter sharing.
• CNN automatically detects the important
features.
• High Accuracy for uncontrolled face images.
Methodology
• Convolutional Layer
• Pooling Layer
• Rectified Linear Unit(Relu Layer)
• Fully Connected layer
Convolutional layer
• Convolutional layer is the core building block of
CNN.
• The primary purpose of convolutional layer is
to extract features from input data which is an
image.
• The main advantage compared to its
predecessors is that it automatically detects
the important features without any human
supervision.
• The working of convolutional layer is given
below.
Pooling Layer
• A pooling layer is another building block of a
CNN.  Its function is to progressively reduce
the spatial size of the representation to reduce
the amount of parameters and computation in
the network. 
• It operates each feature map independently.
• The most common method used in pooling is
max pooling.
•Working of max pooling layer is shown below.
Relu Layer
• Relu is the most commonly used activation
function in CNN.
• It is an element wise operation that means it is
applied per pixel and reconstitutes all negative
values in the feature map by zero.
• It is defined as f(x)=max(0,x).
Fully Connected Layer
• The fully connected layer (FC) operates on a
flattened input where each input is connected
to all neurons.
• FC layers are used to detect specific global
configurations of the features detected by the
lower layers in the net.
• The goal of employing the FCL is to employ
these features for classifying the input image
into various classes based on the training
dataset.
A Traditional Convolutional Neural Networks Design
Experimental Analysis

Georgia Tech Database and a brief result of Proposed Algorithm


Conclusion
• The results showed higher recognition rates
compared to other learning algorithms.
• Less time consumed for computation.

Future Work
There are many pre-trained models.We can modify
these pre-trained models as for our project
requirements.This increases relatively higher
prediction and accuracy.
References
• S. G. Bhele and V. H. Mankar, “A Review Paper on
Face Recognition Techniques,” Int. J. Adv. Res.
Comput. Eng. Technol., vol. 1, no. 8, pp.
2278–1323, 2012.
• D. N. Parmar and B. B. Mehta, “Face Recognition
Methods & Applications,” Int. J. Comput.
Technol. Appl., vol. 4, no. 1, pp. 84–86, 2013.
• S. N. Borade, R. R. Deshmukh, and S. Ramu,
“Face recognition using fusion of PCA and LDA:
Borda count approach,” in 24th Mediterranean
Conference on Control and Automation, MED
2016, 2016, pp. 1164– 1167.
Thank You

You might also like