MNIST Digit Classifier

Handwritten digit recognition using MLP & CNN neural networks built from scratch

This project demonstrates handwritten digit recognition using neural networks built from scratch for MNIST classification. The demo below lets you draw a single digit from 0 to 9, send it to a Python backend, and receive a model prediction with confidence.For best results, draw one large digit in the center of the box and avoid drawing multiple numbers at once.⏳ Checking backend status...

Draw a Digit

Brush Size:18px

⌨️ Keyboard shortcuts: C Clear • Enter Predict

Prediction Result

✍️

Draw a digit and click Predict

Best results with centered, large digits

💡 Tips for Best Results

1. Center Your Digit

Draw the digit in the middle of the canvas for optimal recognition

2. Use Large, Thick Strokes

Adjust brush size for clear, bold digits that fill the space

3. One Digit Only

Draw a single digit, not multiple numbers, for accurate predictions