This document discusses hand gesture recognition using an artificial neural network. It aims to classify hand gestures into five categories (pointing one to five fingers) using a supervised feed-forward neural network and backpropagation algorithm. The objective is to facilitate communication for deaf people by automatically translating hand gestures into text. The system requires software like Pandas, Numpy and Matplotlib as well as hardware with a quad core processor and 16GB RAM. It explains key concepts of neural networks like neurons, weights, biases, activation functions and their advantages in handling large datasets and inferring unseen relationships.
2. Contents
■ Introduction
■ Gesture Recognition
■ Hand Gesture Recognition
■ Objective
■ Software Requirements
■ Hardware Requirements
■ Artificial Neural Network
■ A Simple Neuron
■ Activation Function
■ Advantages of using ANN
■ Conclusion
■ Future Scope
■ References
3. Introduction:-
■ Pattern recognition and Gesture recognition are the growing fields of
research. Being a significant part in non verbal communication, hand
gestures are playing vital role in our daily life.
■ In this project we proposed a method for recognizing hand gestures. We
have designed a system which can identify specific hand gestures and use
them to convey information.
■ We have used supervised feed-forward neural net-based training and back
propagation algorithm for classifying hand gestures into five categories:
hand pointing one, two, three, four or five fingers, and gives the output as
number of fingers user was showing.
4. Hand Gesture Recognition
■ Hand gesture recognition is one obvious way to create a useful, highly
adaptive interface between machines and their users.
■ Hand gesture recognition technology would allow for the operation of
complex machines using only a series of finger and hand movements,
eliminating the need for physical contact between operator and machine.
5. Objective
■ Touch screen technology is not cheap enough to be implemented in
desktop systems.
■ Sign language is the primary way, more than 70 million deaf people
communicate with non-deaf people. Being able to automatically translate
hand gesture to text can facilitate deaf to non-deaf communication.
6. Software Requirements
Pandas, Numpy, MatPlotlib
Web browser (Internet explorer)
Anaconda/ Jupyter Notebook
Hardware Requirements
2.5 GH quad core processor with 16 Gb of RAM
7. Artificial Neural Network (ANN)
Background
Artificial neural networks (ANN) are computing systems that are inspired by, but not
necessarily identical to, the biological neural networks that constitute animal brains.
Such systems "learn" to perform tasks by considering examples, generally without being
programmed with any task-specific rules.
8. A Simple Neuron
■ Each input is multiplied by its corresponding weights. Weights refers to amplitude or
strength of connection between two nodes.
■ The weighted inputs are all summed up inside computing unit (artificial neuron).
■ Bias is added to make output non-zero or to scale up to the system response.
9. Activation Function
■ The Activation function is a set of
transfer function used to get desired
output.
■ Activation functions are used to
manipulate the output of the neural
network layer to a more suitable form.
■ There are various types of activation
functions i.e. sigmoid function, ReLu
activation function, Softmax Activation
Function.
■ ReLu: It stands for Rectified Linear Units.
It is the most widely used activation
function. It is non-linear in nature.
It’s just R(x)= max(0,x) i.e. if x<0,
R(x)=0 and if x>=0, R(x)= x.
10. ■ Softmax Function: The softmax function is handy when we are trying to
handle classification problems. The softmax function is ideally used in
the output layer of the classifier where we are actually trying to attain the
probabilities to define the class of each input.
11. Advantages of using ANN
■ ANNs can generalize- After learning from initial inputs and their
relationships it can infer unseen relationships on unseen data as well.
Thus, making the model more generalized and predict on unseen data.
■ The main advantage of using ANN includes, it can handle large amount
of datasets.
15. Conclusion
■ Gesture Recognition provides the most important means for non-verbal
interaction among people specially for impaired people (i.e. deaf and
dumb).
■ ANN is one of the most effective techniques of software computing for
hand gestures and recognition problem.
■ Neural Network is efficient as long as we have a vast train data set.
16. Future Scope
■ The detection capability of the system could be expanded to body
gestures as well.
■ Area of Hand gesture-based computer human interaction is very vast.
This project recognizes hand gesture off-line so work can be done to do it
for real time purpose.
17. References
■ MACHINE LEARNING YEARNING (Technical Strategy for AI Engineers,
In the Era of Deep Learning) by “Andrew NG”.
■ HANDS-ON MACHINE LEARNING WITH SCIKIT-LEARN & TENSOR
FLOW (Concepts tools and Techniques to Build Intelligent Systems ) by
“Aurelien Geron”.
■ RULES OF MACHINE LEARNING by “Martin Zinkevich”.
■ DATA MINING CONCEPTS AND TECHNIQUES by “Jiawei Han”
“Micheline Kamber” “Jian Pei”
■ https://www.analyticsvidhya.com/blog/2016/08/evolution-core-concepts-
deeplearning-neural-networks/