Project: Handwritten Digit Recognition (Real-Time Air Drawing)
This project demonstrates a real-time handwritten digit recognition system using a webcam. It tracks your index finger in the air (via MediaPipe), draws the path onto a virtual canvas, and uses a Convolutional Neural Network (CNN) to classify the drawn digit.
🎯 Features
- 📷 Real-time finger tracking using MediaPipe
- 🧠 Digit recognition using a trained CNN (on MNIST dataset)
- 🖼️ Live canvas drawing with OpenCV
- 🧪 Keyboard shortcuts to predict, clear, or quit
🔧 Tech Stack
- Python 3.10+
- TensorFlow / Keras
- OpenCV
- MediaPipe
- NumPy