Research Paper
Leveraging a convolutional neural network for developing a hand-written text recognition system
The explication revolving around the digital outlook of Hand-written Text has been carried out widely with the upsurge of Technological Advancements in the field of Artificial Intelligence. Automation in daily chores has made human life more consistent and yet stable. Machine Learning and Deep Learning modus operandi has been actuated for contemplating the automated outlook. Hand-written Text Recognition has been a well-scrutinized domain wherein the incorporation of different paradigms of learning has been elucidated. The escalation into the digitization of almost everything necessitates the prevalence of the same in assuaging the Text Recognition System as well. In this articulation, we’ve consummated Convolutional Neural Networks (CNNs) i.e., 03-Layered, 04-Layered, and 05-Layered Convolutional Neural Networks (CNNs) which are trained over Hand-written Characters which include Alphabets (A-Z), Digits (0-9) and widely utilized Symbols and thereby detecting the Text from a Dataset of diverse Hand-written words consisting of Names and Surnames. The proposed Hand-written Text Recognition System is structured in the Offline method, as the scarcity of proficient outcomes revolves around it when compared to Online prospects. The proposition also gave us a comparative insight into the impact of layers in models. The efficacy we achieved was phenomenal, which gave us a potent resolution for contemplating it in real-time.
Published by: Likith Pulluri, Yashwanth Konakanchi, Divya Prabha, Asima Nayak, Pradeep Kumar Mishra
Author: Likith Pulluri
Paper ID: V8I2-1319
Paper Status: published
Published: April 23, 2022
Full Details