Classifying Hand-drawn Digits using Flask

---

• Original Paper
  • Gradient Based learning applied to document Recognition

• Learning Resources

  1. CS231N Convolutional Neural Networks
  2. CS231N Visualizing what ConvNets learn
  3. A Beginner's Guide To Understanding Convolutional Neural Networks
  4. Machine Learning is Fun! Part 3: Deep Learning and Convolutional Neural Networks
  5. Visualizing MNIST: An Exploration of Dimensionality Reduction - Chris Olag

• Code Resources

  • Convolutional Neural Network by Magnus Erik Hvass Pedersen
  • keras repositoy
  • Deep Learning with Python - Francois Chollet
  
  

• Model Training Files (2 different models trained)

  • (Basic) Initial -> mnist-keras(colab)
  • (Augmented) Used for predictions -> Mnist-data-augmented(colab)
  
  

• Run Online

  • repl.it

• Inspired by
  • Sourav Johar's Doodle Classifier

Run Locally

1. Create a seperate python 3 environment.

2. Install the required packages

    pip install -r requirements.txt

3. Run from environment

   python server.py

Outputs

• Input

• Output

• Input

• Output

• ToDo
  1. Remove scipy.misc.imresize deprecated warning. (remove this line to view the warning)

     warnings.filterwarnings("ignore")

  2. Do instant detection of number.
  3. Create a better GUI.