MNIST Digit Classifier

Description

This is a simple handwriting recogniser trained on the most famous and commonly used MNIST dataset. The recogniser is made possible by training a custom made deep neural network using the Tensorflow Python library, and deployed onto the AWS Lambda stack (Node.js) for inference using Tensorflow.js. Although the model is reasonably good at recognising most digits, it is not 100% accurate (just like any other ML models). So, please don't have expectations. 🙂

Examples of MNIST training dataset
Source : Wikipedia
DRAW YOUR DIGIT (0-9) HERE
Instruction
  • Try to fill the entire space 🙂.
  • Once drawn click the recognise button.
  • Re-draw after your digit is recognised.
Hold tight for few more seconds. The Lambda is getting the cold start
Error : Sorry something went wrong
Press the reset button to try again.