🧮 AI Handwritten Math Solver

Draw math. Get answers. Instantly.

---

✨ What It Does

Write any math expression by hand — just like on paper — and this AI reads it, understands it, and gives you the answer in real time.

No keyboard. No typing. Just draw.

---

🎬 How It Works

You draw on screen
        ↓
OpenCV detects each symbol
        ↓
CNN model reads what you wrote (99.63% accuracy)
        ↓
SymPy solves the expression
        ↓
Answer appears instantly

Supports:

• ➕ Addition    2 + 3 = 5
• ➖ Subtraction    9 - 4 = 5
• ✖️ Multiplication    6 * 7 = 42
• ➗ Division    10 / 2 = 5
• 🔢 Multi-digit numbers    12 + 34 = 46
• 📐 Equations    2x + 3 = 7 → x = 2

---

🚀 Quick Start

1. Clone the repo

git clone https://github.com/yourusername/handwritten-math-solver
cd handwritten-math-solver

2. Set up environment

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

3. Train the model

python train_model.py

4. Run the solver

python main.py

---

🎮 Controls

Key	Action
Draw	Use mouse to write on the canvas
S	Solve the expression
C	Clear the canvas
Q	Quit

---

🧠 How the AI Was Built

The model is a Convolutional Neural Network (CNN) trained from scratch on:

• 60,000 handwritten digit images from the MNIST dataset
• 24,000 synthetic operator images (custom generated for +, -, ×, ÷)

The architecture uses multiple Conv2D layers with BatchNormalization and Dropout for regularization — achieving 99.63% test accuracy.

---

📁 Project Structure

handwritten-math-solver/
├── model/
│   ├── math_model.keras     ← Trained CNN model
│   └── classes.npy          ← Symbol class labels
├── main.py                  ← Live camera + detection app
├── train_model.py           ← CNN training script
├── solver.py                ← Math expression evaluator
└── requirements.txt

---

📦 Requirements

tensorflow
opencv-python
numpy
sympy
cvzone
matplotlib
scikit-learn

---

🏗️ Tech Stack

Tool	Purpose
TensorFlow / Keras	CNN model training & inference
OpenCV	Real-time drawing canvas & contour detection
SymPy	Mathematical expression solving
NumPy	Data processing & array operations

---

👨‍💻 Author

Saeed Fahim BTech — Artificial Intelligence & Machine Learning Chandigarh University

---

Built as part of an AI/ML portfolio to demonstrate real-world deep learning applications.