Handwritten Text Recognition and Digitization System

EasyChair Preprint 12663

Abstract

The project's objective is to enhance the performance of handwritten character recognition using machine learning techniques. The proposed solution involves employing a convolutional neural network (CNN) model, which is well-suited for handling complex data patterns. The dataset utilized for testing comprises handwritten alphabetical characters and digits. Increasing the size of the dataset can lead to improved accuracy. Unlike traditional neural networks where inputs pass through fully connected layers with numerous neurons, CNNs are preferred for Computer Vision tasks involving image inputs. This preference is due to CNNs' ability to manage image data efficiently, unlike fully connected networks that struggle with the increasing number of parameters as layers are added. Thus, CNNs are chosen for their superior accuracy in handwritten recognition compared to other neural network architectures.

Keyphrases: Convolutional Neural Network, machine learning, morphological processing, text detection

@booklet{EasyChair:12663,
  author    = {Venna Sri Ajay Vigneshwar Reddy and Kishori Shekokar},
  title     = {Handwritten Text Recognition and  Digitization System},
  howpublished = {EasyChair Preprint 12663},
  year      = {EasyChair, 2024}}