/Ivlabs-Fall-Internship-Project

This repository contains code of MNIST image classification

Primary LanguageJupyter Notebook

MNIST Digit Classification

  • The project is to classify Famous MNIST Handwritten Digits Dataset using Fully Connected,Convolution and Residual Networks
  • This project is done as a part of Fall Internship at Ivlabs.The Internship is Aimed at studying the Concepts of Deep Learning
  • The Models are able to classify MNIST handwritten Digits with high Accuracy
  • To train the Model and check the Accuracy of the Model on test data,type
python test.py

in your terminal

  • All the results, loss plots, Image predictions are given in the .ipynb Notebooks too. The .ipynb notebooks are also available in this repository so that you can run them easily on platforms like Colab. You can just upload the .ipynb notebooks in your colab/jupyter notebook and simply run them to view results.

Using Residual Networks

(Every network is trained for 5 epochs)

Loss plot

Accuracy: 99.02

Prediction on a Test Image

Using Convolution Neural Networks

Loss plot

Prediction on a Test Image

Using Fully Connected Networks

Loss plot

Prediction on a Test Image