This repository contains a comprehensive analysis of Devanagari Handwritten Character Classification, in accordance with the Devanagari Alphabet. It encompasses all the requisite stages of a Machine Learning approach to classification problems, culminating in the selection of the optimal final model. A variety of non-linear models, meticulously chosen for this specific problem, have been assessed through a rigorous validation process to ensure their effectiveness in addressing this particular challenge.
Link to the website offering this public database: Devanagari Handwritten Characters Database
The dataset have been processed and prepared in a specific file format .npz, easier to work with, located in the folder DATA of this repository.
The main natural language of this project is SPANISH.
The project has been made with python and specific machine learning libraries like scikit-learn.
The project is entirely located in the file Devanagari_Classifier in this repository. This file consists of a jupyter notebook containing the results of the machine learning algorithms and the python code used in the project.