📄 README.md

# Devanagari Character Recognition 📝🔤

A deep learning-based project to recognize handwritten Devanagari characters using Convolutional Neural Networks (CNNs).

---

📂 Project Structure

devanagari-character-recognition/ │ ├── DevanagariHandwrittenCharacterDataset/ # Dataset (excluded from Git) ├── saved_models/ # Trained models and logs (excluded from Git) ├── src/ # Source code │ ├── data_loader.py │ ├── evaluate.py │ ├── model_builder.py │ ├── predict.py │ ├── train.py │ └── visualize.py ├── run_all.py # Main runner script ├── requirements.txt # Python dependencies ├── structure.txt # Folder/code structure ├── .gitignore # Git ignore rules └── README.md # You are here

---

🚀 Getting Started

🔧 Prerequisites

- Python 3.10 or later
- pip

🛠️ Setup

1. Clone the repo:
   ```bash
   git clone https://github.com/Tejas-Indrale/devanagari-character-recognition.git
   cd devanagari-character-recognition

2. Create and activate a virtual environment:

   python -m venv env
   .\env\Scripts\activate    # On Windows

3. Install required packages:

   pip install -r requirements.txt

4. Download and extract the dataset manually into:

   devanagari-character-recognition/DevanagariHandwrittenCharacterDataset/
   

---

🧠 Training

To train the model, simply run:

python run_all.py

It will:

• Load the dataset
• Build the CNN
• Train the model
• Save results in saved_models/

---

📈 Outputs

• ✅ Trained model: saved_models/devnagri_character_model.h5
• 📊 Training metrics: training_history.png
• 🔍 Evaluation reports and TensorBoard logs inside saved_models/logs/

---

📌 Notes

• env/, saved_models/, and dataset folders are ignored via .gitignore.
• This repo is for educational/research use. Dataset should be downloaded manually due to size.

---

📜 License

This project is open-source and available under the MIT License

---

👨‍💻 Author

Tejas Indrale 🔗 GitHub Profile