Yoga Pose Classifier

A deep learning project that classifies yoga poses using Convolutional Neural Networks (CNN). The model can identify and classify 5 different yoga poses: Downdog, Goddess, Plank, Tree, and Warrior2.

🧘‍♀️ Project Overview

This project uses TensorFlow and Keras to build a CNN model that can accurately classify yoga poses from images. The model achieves high accuracy in distinguishing between the 5 different yoga poses included in the dataset.

🎯 Supported Yoga Poses

1. Downdog - Downward-Facing Dog pose
2. Goddess - Goddess pose
3. Plank - Plank pose
4. Tree - Tree pose
5. Warrior2 - Warrior II pose

🚀 Features

• CNN-based image classification
• Data augmentation for better generalization
• Model training with validation
• Real-time pose prediction
• Pre-trained model included (model.h5)

📁 Project Structure

yoga-pose-classifier/
├── DATASET/
│   ├── TRAIN/          # Training images organized by pose
│   │   ├── downdog/
│   │   ├── goddess/
│   │   ├── plank/
│   │   ├── tree/
│   │   └── warrior2/
│   └── TEST/           # Test images organized by pose
│       ├── downdog/
│       ├── goddess/
│       ├── plank/
│       ├── tree/
│       └── warrior2/
├── yoga_pose12s.ipynb  # Main training notebook
├── Main.py             # Inference script
├── model.h5            # Pre-trained model
├── training_logs.csv   # Training history
└── README.md

🛠️ Installation

1. Clone the repository

   git clone https://github.com/your-username/yoga-pose-classifier.git
   cd yoga-pose-classifier

2. Install dependencies

   pip install -r requirements.txt

3. Verify installation

   python Main.py

📊 Dataset

The project uses a yoga poses dataset with:

• Training set: 1,081 images across 5 classes
• Test set: 470 images across 5 classes
• Image size: 150x150 pixels
• Classes: 5 different yoga poses

Dataset source: Yoga Poses Dataset on Kaggle

🔥 Usage

Training the Model

1. Open the Jupyter notebook:

   jupyter notebook yoga_pose12s.ipynb

2. Run all cells to train the model from scratch

Making Predictions

1. Using the inference script:

   python Main.py

2. Using the model programmatically:

   import tensorflow as tf
   import cv2
   import numpy as np
   
   # Load the model
   model = tf.keras.models.load_model('model.h5')
   
   # Load and preprocess image
   image = cv2.imread('path/to/your/image.jpg')
   image = cv2.resize(image, (150, 150))
   image = np.array(image).reshape(1, 150, 150, 3)
   
   # Make prediction
   prediction = model.predict(image)[0]
   
   # Convert to class name
   poses = {1: 'Downdog', 2: 'Goddess', 3: 'Plank', 4: 'Tree', 5: 'Warrior2'}
   predicted_pose = poses[np.argmax(prediction) + 1]
   print(f'Predicted pose: {predicted_pose}')

🏗️ Model Architecture

The CNN model consists of:

• Input Layer: 150x150x3 (RGB images)
• Convolutional Blocks: 4 blocks with increasing filter sizes (64, 128, 256, 256)
• Pooling: MaxPooling after each block
• Dropout: 0.2 dropout rate for regularization
• Dense Layers: Fully connected layers for classification
• Output: 5 classes with softmax activation

📈 Model Performance

• Training Accuracy: High accuracy achieved during training
• Validation Accuracy: Good generalization on test set
• Training History: Available in training_logs.csv

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

👨‍💻 Author

Anuj Dev Singh

• Project Creator & Lead Developer

🙏 Acknowledgments

• Dataset provided by Niharika41298 on Kaggle
• Built with TensorFlow and Keras
• OpenCV for image processing

📞 Contact

If you have any questions or suggestions, feel free to reach out!

---

Happy Coding! 🧘‍♂️✨