Recognition of Handwritten Digits

Project completed in October 2021, with the aim of implementing a handwritten digit recognition system using an interactive interface, and trained on the famous MNIST database.

❓ Issue

How can we enable a user to draw a digit by hand and have a system recognise it in real-time with a high level of accuracy?

🛠️ Implemented Solution

• 📥 Training a CNN (Convolutional Neural Network) model on the MNIST dataset

• 🖼️ Image preprocessing to normalise, resize, and format user inputs

• 🧠 The model predicts in real-time the digit drawn by the user

• 🎨 Creation of a simple graphical interface with Tkinter, allowing the user to draw and submit their image

• ⚙️ Backend processing in Python, immediate display of the prediction

Link to the project's GitHub repository: Handwritten Digit Recognition

⚙️ Technical Stack

• Language: Python

• Libraries: Keras · TensorFlow · Pillow (image handling)

• Dataset: MNIST

• Interface: Tkinter (drawing, prediction)

• Environment: Jupyter Notebook

• Skills: Deep Learning · Computer Vision · User Interfaces