🍚 Rice Type Classification using PyTorch

A deep learning project to classify rice types (Jasmine vs Gonen) using a Neural Network built with PyTorch.

📋 Project Overview

This project implements a Binary Classification model to distinguish between two types of rice:

• Jasmine Rice (Class 1)
• Gonen Rice (Class 0)

The model is trained on physical characteristics of rice grains like Area, Perimeter, Eccentricity, etc.

🎯 Results

Metric	Value
Test Accuracy	~92-95%
Training Epochs	50
Batch Size	32

📊 Dataset

• Source: Kaggle - Rice Type Classification
• Total Samples: 18,185
• Features: 11 (Area, Perimeter, MajorAxisLength, etc.)
• Target: Binary (0 or 1)

Data Split:

Set	Percentage	Samples
Training	70%	~12,729
Validation	15%	~2,728
Testing	15%	~2,728

🧠 Model Architecture

Input (11 features)
      ↓
Linear(11 → 64) → ReLU → Dropout(0.2)
      ↓
Linear(64 → 32) → ReLU → Dropout(0.2)
      ↓
Linear(32 → 1) → Sigmoid
      ↓
Output (0 to 1 probability)

🛠️ Tech Stack

• Python 3.8+
• PyTorch - Deep Learning Framework
• Pandas - Data Manipulation
• NumPy - Numerical Operations
• Matplotlib - Visualization
• Scikit-learn - Train/Test Split & Metrics

📁 Project Structure

rice_type_classification/
│
├── code.ipynb              # Main Jupyter Notebook
├── rice_classifier.pth     # Saved Model Weights
├── README.md               # Project Documentation
└── rice-type-classification/
    └── riceClassification.csv  # Dataset

🚀 How to Run

1. Clone the Repository

git clone https://github.com/KumarKhailendra/rice_type_classification.git
cd rice_type_classification

2. Install Dependencies

pip install torch pandas numpy matplotlib scikit-learn opendatasets

3. Run the Notebook

jupyter notebook code.ipynb

4. Or Run in VS Code

• Open code.ipynb in VS Code
• Run all cells

💡 Key Learnings

• Custom Dataset class in PyTorch
• DataLoader for batch processing
• Neural Network with Dropout for regularization
• Binary Cross Entropy Loss for classification
• Adam Optimizer for training
• Model saving and loading

🔮 Future Improvements

• Add more rice types (Multiclass Classification)
• Implement CNN for image-based classification
• Create a Flask/Streamlit web app
• Add confusion matrix visualization
• Hyperparameter tuning

👨‍💻 Author

Khailendra Prasad

📄 License

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

⭐ Show Your Support

Give a ⭐ if this project helped you!

---

Made with ❤️ by Khailendra Prasad