You are on page 1of 5

METHODOLOGY

Project Name: Handwritten Digit recognition

Members:
Sanjeeva Kumar -1AT17CS072
Seenakula Ravi Shankar -1AT17CS077
Shashank M V -1AT17CS078
Vandana K C -1AT17CS092
ABSTRACT:
A handwritten digit recognition system was used in a
demonstration project to visualize artificial neural networks,
in particular Korhonen’s self-organizing feature map. The
purpose of this project was to introduce neural networks
through a relatively easy-to-understand application to the
general public. This paper describes several techniques used
for pre-processing the handwritten digits, as well as a number
of ways in which neural networks were used for the
recognition task. Whereas the main goal was a purely
educational one, a moderate recognition rate of 98% was
reached on a test set.
OBJECTIVES:
The main purpose of this thesis is to find out the recognition
performance of the methods. In order to analyse the
performance of the methods, data is needed to be used for
training using machine learning methods. Then digit data is
tested on the desired machine learning technique.

In this thesis, the following methods are performed:


● Implementation of HOG Feature extraction method with
SVM.
● Implementation of HOG Feature extraction method with
ANN.
● Implementation of Deep Learning methods with CNN.
BLOCK DIAGRAM:
Block diagram of methodology used in handwritten digit
recognition. We propose schemes based upon contributed
decision posted by combining multiple base ensemble
recognition neural network classifiers for Handwritten digit
dataset. Figure Shows the methodology used in proposed
work, that is, to employ recognition as neural network base
classifier for the ensemble.
INPUT:
Real-time user input(digits)
We chose MNIST data sets to demonstrate that dropout is a
general technique for improving neural nets and is not specific
to any particular application domain. The MNIST data set
consists of 60,000 training and 10,000 test examples each
representing a 28×28 digit image. These data sets include
different image types and training set sizes. Models which
achieve state-of-the art results on all of these data sets use
dropout.
OUTPUT: We have built and trained the Convolutional
neural network which is very effective for image classification
purposes. Later on, we build the GUI where we draw a digit
on the canvas then we classify the digit and show the results.
ALGORITHMS:
1. Compute the distance metric between the test data point
and all labelled data points.
2. Order the labelled data points in increasing order of
distance metric. 3. Select the top K labelled data points and
look at class labels.
4. Look for the class labels that majority of these K labelled
data points have and assign it to test data points.
PLATFORM:
Desktop/Laptop
Languages:
PYTHON
Library:
Keras and Tkinter

Estimated time for following activities:


Data collection – 2 days
Acquiring essential skills – 10 days
Understanding the algorithms – 10days
Allocate work among team – 1 day
Code – 7 days
Unit test – 3 days
Integration – 2 days
Analyse the results – 1 day
Demonstration – 1 day
Report writing – 5 days

You might also like