Publication: Object Recognition Using Convolutional Neural Network Architecture
Date
2020-02
Authors
Prasantth a/l Subramaniam
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Evolution of current modern era demands a fast and excellent way of computing day by
day. It has dramatically transformed how we work and live. Involvement of Artificial
Intelligence (AI) has significantly show the intelligent behavior side of today’s
computers. Now machines are making intelligent decisions in recognizing objects,
people, and languages. In has massively create an impact in a world driven by
technology. My thesis illustrates the method of recognizing age and gender of a human
in real time with the help of pre-trained deep learning models. In my project work, the
age database has a total of 13,420 face images which has been separated into two
different class; child and adult. The gender database has a total of 2,000 face images
which has been separated into two different class; male and female. The main idea of
this research is to study the performance of deep learning models by training the
architectures consist of neural networks to classify images based on class and can be
use in real time classification as well. The three best deep learning models which are
VGG16, ResNet50 and MobileNet been selected based on their performance on training
and validation accuracy. Hence, some classification on test images to witness the
performance of the neural networks. My project work to be illustrated in this thesis.
Description
FYP 2 SEM 2 2019/2020
Keywords
tensorflow , deep learning , age gender classification