Scribd Logo
Upload

Welcome to Scribd!

  • 13 views

    Sat - 23.Pdf - Handwritten Hindi Character Recognition Using CNN

    Uploaded by

    Vj Kumar
    The document describes a study that uses a convolutional neural network (CNN) deep learning model to recognize handwritten Hindi characters. The model achieves 92.4% recognition accuracy on a dataset of 92,000 photos of handwritten Devanagari characters separated into training and test sets. The CNN model contains four convolutional layers followed by three fully connected layers to process grayscale images of handwritten characters and output the predicted character class with the highest probability score. This model outperforms some older models in terms of performance and accuracy for recognizing handwritten Hindi characters.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Sat - 23.Pdf - Handwritten Hindi Character Recognition Using CNN

    Uploaded by

    Vj Kumar
    13 views11 pages
    The document describes a study that uses a convolutional neural network (CNN) deep learning model to recognize handwritten Hindi characters. The model achieves 92.4% recognition accuracy on a dataset of 92,000 photos of handwritten Devanagari characters separated into training and test sets. The CNN model contains four convolutional layers followed by three fully connected layers to process grayscale images of handwritten characters and output the predicted character class with the highest probability score. This model outperforms some older models in terms of performance and accuracy for recognizing handwritten Hindi characters.

    Original Title

    Sat_23.Pdf_ Handwritten Hindi Character Recognition Using Cnn

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document describes a study that uses a convolutional neural network (CNN) deep learning model to recognize handwritten Hindi characters. The model achieves 92.4% recognition accuracy on a dataset of 92,000 photos of handwritten Devanagari characters separated into training and test sets. The CNN model contains four convolutional layers followed by three fully connected layers to process grayscale images of handwritten characters and output the predicted character class with the highest probability score. This model outperforms some older models in terms of performance and accuracy for recognizing handwritten Hindi characters.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    13 views11 pages

    Sat - 23.Pdf - Handwritten Hindi Character Recognition Using CNN

    Uploaded by

    Vj Kumar
    The document describes a study that uses a convolutional neural network (CNN) deep learning model to recognize handwritten Hindi characters. The model achieves 92.4% recognition accuracy on a dataset of 92,000 photos of handwritten Devanagari characters separated into training and test sets. The CNN model contains four convolutional layers followed by three fully connected layers to process grayscale images of handwritten characters and output the predicted character class with the highest probability score. This model outperforms some older models in terms of performance and accuracy for recognizing handwritten Hindi characters.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 11

    ABSTRACT

    The goal of this study is to employ a Convolutional Neural Network (CNN)-


    based Deep Learning model to recognize handwritten Hindi characters.
    The identified characters can then be saved digitally or used for other
    purposes. The data was taken from the Kaggle dataset, which has 92,000
    photos separated into two sets: training (80 percent) and test (20 percent).
    It comprises a variety of handwritten Devanagari characters produced by
    various people that can be used to train and test handwritten text
    recognizers as well as conduct writer identification and verification tests.
    On top of a TensorFlow backend, the model is built with Keras libraries. For
    recognition, it has four CNN layers followed by three fully connected layers.
    As input, grayscale handwritten character pictures are employed. Filters
    are used to extract different features from the photos at each layer. The
    Convolution operation does this. Pooling and flattening are the other two
    major activities involved. The CNN layers' output is delivered to the fully
    connected layers. Finally, each character's chance or probability score is
    calculated, and the character with the highest probability score is displayed
    as the output. 92.4% recognition accuracy is achieved. Similar models exist
    for the purpose, but the new model outperformed and outperformed some
    of the older models in terms of performance and accuracy

    5
    TABLE OF CONTENTS

    ABSTRACT 5

    TABLE OF CONTENTS 6

    LIST OF FIGURES 7

    LIST OF TABLES 7

    ABBREVIATIONS 8

    REFERENCES 9

    KEYWORDS 10

    1. Introduction 11
    1.1. Introduction
    1.2. Aim and Objective
    1.3. Motivation
    1.4. Software Requirements Specification
    2. Literature Review 14
    2.1. Different approaches used
    2.2. Comparison of Existing methods with merits and demerits
    3. Methodology 16
    3.1. Models Used
    3.2. Architecture diagram
    4. Implementation 19

    6
    4.1. Data collection and analysis
    4.2. Data Preprocessing
    4.3. Tokenization
    4.4. Feature Extraction
    4.5. Splitting the dataset
    4.6. Model Training and verification
    5. Libraries Used 22
    5.1. Numpy
    5.2. Pandas
    5.3. Sklearn
    6. Coding and Testing 25
    7. Conclusion and Enchantment 28

    LIST OF FIGURES

    ● Architecture Diagram 18

    LIST OF TABLES

    ● Literature Review 14
    ● Accuracy Table 28

    7
    ABBREVIATIONS

    ML- Machine Learning


    np - Numpy
    pd – Pandas
    CNN – Convolutional Neural Network

    Reference :
    8
    [1] Apoorva Chaudhary, Raushan Lal Chhoker ,”Handwritten Hindi Numeric Character
    Recognition and comparison of Algorithms”, 7th International Conference on Cloud Computing,
    Data Science & Engineering – Conference (2017).
    [2] R.Parthiban,R.Ezhilarasi,D.Saravanan,”OpticalCharacter Recognition for English Handwritten
    Text Using Recurrent Neural Network”, University of Gothenburg (2020).
    [3] Ripal Patel, Palak Patel,” Handwritten Character Recognition using Neural Network”,
    International Journal of Scientific & Engineering Research Volume 2, Issue 3, March-2011.
    [4] Mahmoud M. Abu Ghosh, Ashraf Y. Maghari, ” handwritten digit recognition using neural
    networks research paper", 2017 International Conference on Promising Electronic Technologies
    (2017).
    [5] Parshuram Kamble, Ravinda, Hegadi,” Handwritten Marathi character recognition using R-
    HOG Feature", International Conference on Advanced Computing Technologies and
    Applications(2015).
    [6] M. A. Pragathi, S. Saveetha, K. Priyadarshini, A. Shavar Banu, K. O. Mohammed Aarif,
    “Handwritten Tamil Character Recognition Using Deep Learning",2019 International Conference
    on Vision Towards Emerging Trends in Communication and Networking (ViTECoN) (2019).
    [7] Naila , Adnan,” Urdu Optical Character Recognition Systems: Present Contributions and
    Future Directions",2018 IEEE.
    [8] Kaensar,” A Comparative Study on Handwritten Digit Recognition Classifier Using Neural
    Network”, 2019 IEEE.
    [9] Sujatha, Lalitha,” A Survey on Offline Handwritten Text Recognition of Popular Indian
    Scripts”, International Journal of Computer Science and Engineering 2019.
    [10] Wu, Chen.” Image Recognition Based on Deep Learning”, 2015 IEEE
    [11] Sheng ,” A Review of Gradient-Based and Edge-Based Feature Extraction Method for
    Object Detection”, 2011 IEEE

    KEYWORDS

    9
    Handwritten Hindi Characters;
    Devanagari;
    Keras;
    Convolutional Neural Network;

    1.INTRODUCTION

    10
    1.1 INTRODUCTION:

    In today's environment, handwritten text recognition technology is extremely


    useful and necessary. Physical data storage and retrieval are challenging, and
    there is a risk of harm or loss. Physical data generation is also prone to errors. If
    the text is in a digital format, however, error scanning methods and autocorrect
    tools can aid in the correct and efficient storage of data. This is one of the most
    difficult jobs in the field of pattern recognition. One of the most significant benefits
    of digitally preserving text is that it can be retrieved from any location. You do not
    need to be present at the location where the data is stored. Data storage,
    access, and analysis have all become considerably easier because of this
    technology. Optical character recognition deals with an important issue called the
    character classification and this is more challenging because of the similarities
    between them. Character classification is something that identifies the characters
    in Hindi that are trained initially and it can identify the characters written by
    different users. Hindi language is most popular and its techniques grammar are
    formulated for the recognition of a Hindi script is one of the foremost among the
    16 major national languages spoken reading and writing by the North Indian
    people. Character recognition can be online or offline. In On-line character
    recognition system, the representation of two-dimensional coordinates of
    successive points is done. It is automatic conversion of text into a digital form.
    Off-line character recognition is used to convert written text into letter codes.
    Since there are many hidden layers in deep neural network the parameters for
    the training is very huge. To prevent over fitting we require a large set of
    examples. One such magnificence of deep mastering is the Convolution Neural
    Network. Convolution Neural Network is a special type of neural network used
    effectively for image recognition and classification. By implementing these
    techniques, we recognize the handwritten Hindi characters. Hindi is one of the
    ancient languages spoken by many people.

    1.2 AIM AND OBJECTIVES

    11
    • The aim of a handwriting character recognition system is to convert
    handwritten character into machine readable formats.
    • The main objective of this work is to ensure effective and reliable
    approaches for hand-written character recognition and making machine
    readability easier.
    • Handwritten character recognition is meant for receiving and interpreting
    handwritten input in the form of pictures or paper documents.

    1.3 MOTIVATION

    Hand writing recognition of characters has been around since the 1980s.The task of
    handwritten character recognition, using a classifier, has great importance and use
    such as – online handwriting recognition on computer tablets, recognize zip codes
    on mail for postal mail sorting, processing bank check amounts, numeric entries in
    forms filled up by hand like tax forms, bank forms and so on. The general problem
    we predicted we would face in this digit classification problem was the similarity
    between the digits like 1 and 7, 5 and 6, 3 and 8, 9 and 8 etc. Also people write the
    same digit in many different ways ‐ the digit „1‟ is written as „1‟, „1‟, „1‟ or „1‟. Similarly
    7 may be written as 7, 7, or 7. Finally the uniqueness and variety in the handwriting
    of different individuals also influences the formation and appearance of the digits.

    1.4 SOFTWARE AND HARDWARE REQUIREMENTS SPECIFICATION:

    • Any modern Intel or AMD processor


    • 4 GB RAM
    • 50 GB or more storage
    • Stable Internet Connection
    • OS – Windows 7 or Above
    • Python 3.7
    • Anaconda Navigator
    2. LITERATURE REVIEW

    12
    2.1 DIFFERENT APPROACES USED:

    13
    14
    3. METHODOLOGY

    3.1 MODELS USED:

     CONVOLUTIONAL NEURAL NETWORK


    A Convolutional Neural Network is a Deep Learning algorithm which can take in
    an input image, assign importance to various aspects/objects in the image and
    be able to differentiate one from the other. The pre-processing required in a
    ConvNet is much lower as compared to other classification algorithms. While in
    primitive methods filters are hand-engineered, with enough training, ConvNets
    have the ability to learn these filters/characteristics.

    1. Convolutional Layer
    This layer is the first layer that is used to extract the various features from the input
    images. In this layer, the mathematical operation of convolution is performed
    between the input image and a filter of a particular size MxM. By sliding the filter
    over the input image, the dot product is taken between the filter and the parts of the
    input image with respect to the size of the filter (MxM). The output is termed as the
    Feature map which gives us information about the image such as the corners and
    edges. Later, this feature map is fed to other layers to learn several other features of
    the input image.

    2. Pooling Layer
    In most cases, a Convolutional Layer is followed by a Pooling Layer. The primary
    aim of this layer is to decrease the size of the convolved feature map to reduce the
    computational costs. This is performed by decreasing the connections between
    layers and independently operates on each feature map. Depending upon method
    used, there are several types of Pooling operations. In Max Pooling, the largest
    element is taken from feature map. Average Pooling calculates the average of the
    elements in a predefined sized Image section. The total sum of the elements in the
    predefined section is computed in Sum Pooling. The Pooling Layer usually serves as
    a bridge between the Convolutional Layer and the FC Layer Must Read: Neural
    Network Project Ideas

    3. Fully Connected Layer


    The Fully Connected (FC) layer consists of the weights and biases along with the
    neurons and is used to connect the neurons between two different layers. These
    layers are usually placed before the output layer and form the last few layers of a
    CNN Architecture. In this, the input image from the previous layers are flattened and
    fed to the FC layer. The flattened vector then undergoes few more FC layers where

    16

    You might also like