README.md

metadata

title: Satellite Classification Dashboard
emoji: 🛰️
colorFrom: blue
colorTo: purple
sdk: streamlit
sdk_version: 1.47.0
app_file: app.py
pinned: false
preload_from_hub:
  - Bhavi23/Custom_CNN/best_multimodal_model.keras
  - Bhavi23/MobilenetV2/multi_input_model_v1.keras
  - Bhavi23/EfficientNet_B0/efficientnet_model.keras
  - Bhavi23/DenseNet/densenet_model.keras

🛰️ Satellite Classification Dashboard

A comprehensive deep learning application for classifying satellite images using multiple state-of-the-art neural network models. This interactive dashboard allows you to compare different models, analyze their performance, and classify satellite images in real-time.

🌟 Features

🏠 Dashboard Overview

• Interactive web interface built with Streamlit
• Multi-page navigation system
• Real-time image classification
• Comprehensive model comparison
• Performance analytics and visualizations

🤖 Multiple Model Support

• Custom CNN: Tailored architecture for satellite imagery (95.2% accuracy)
• MobileNetV2: Lightweight model optimized for speed (92.8% accuracy, 18ms inference)
• EfficientNetB0: Best accuracy-efficiency balance (96.4% accuracy)
• DenseNet121: Complex pattern recognition capabilities (94.7% accuracy)

🎯 Classification Categories

The system can classify 11 different types of satellites and space objects:

• AcrimSat
• Aquarius
• Aura
• Calipso
• Cloudsat
• CubeSat
• Debris
• Jason
• Sentinel-6
• TRMM
• Terra

🚀 Quick Start

Try the Live Demo

Visit the Hugging Face Space to use the application directly in your browser.

Local Installation

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/satellite-classification-dashboard.git
cd satellite-classification-dashboard

2. Install dependencies

pip install -r requirements.txt

3. Run the application

streamlit run app.py

4. Open your browser and navigate to http://localhost:7860

📦 Dependencies

• streamlit==1.47.0
• tensorflow-cpu==2.15.0
• numpy==1.26.4
• pandas==2.2.2
• plotly==5.22.0
• Pillow==10.4.0
• requests==2.32.3
• protobuf<4.0.0

🎮 How to Use

1. Home Page

• Overview of the system capabilities
• Class distribution visualization
• Quick navigation to other features

2. Model Comparison

• Side-by-side performance metrics
• Interactive charts comparing accuracy, speed, and model size
• Model selection recommendations

3. Image Classification

• Upload satellite images (PNG, JPG, JPEG)
• Select one or multiple models for prediction
• View confidence scores and detailed probabilities
• Get model recommendations based on performance

4. Performance Analytics

• Detailed performance breakdowns
• Efficiency metrics analysis
• Comprehensive model comparison tables

5. About Models

• Technical specifications for each model
• Strengths and use cases
• Model architecture details

📊 Model Performance

Model	Accuracy	Precision	Recall	F1-Score	Inference Time	Model Size
EfficientNetB0	96.4%	96.1%	96.2%	96.1%	35ms	20.1MB
Custom CNN	95.2%	94.8%	95.1%	94.9%	45ms	25.3MB
DenseNet121	94.7%	94.2%	94.5%	94.3%	52ms	32.8MB
MobileNetV2	92.8%	92.1%	92.5%	92.3%	18ms	8.7MB

🎯 Model Selection Guide

• 🏆 Best Overall Accuracy: EfficientNetB0 (96.4%)
• ⚡ Fastest Inference: MobileNetV2 (18ms)
• 💾 Most Lightweight: MobileNetV2 (8.7MB)
• 🎯 Best Balance: EfficientNetB0 (High accuracy + efficiency)
• 📱 Mobile/Edge Deployment: MobileNetV2
• 🔬 Research/High Accuracy: EfficientNetB0 or DenseNet121

🏗️ Architecture

Model Sources

All models are hosted on Hugging Face Model Hub:

• Custom CNN: Bhavi23/Custom_CNN
• MobileNetV2: Bhavi23/MobilenetV2
• EfficientNetB0: Bhavi23/EfficientNet_B0
• DenseNet121: Bhavi23/DenseNet

Data Pipeline

• Image Upload: Supports PNG, JPG, JPEG formats
• Preprocessing: Resize to 224x224, normalize to [0,1]
• Model Inference: Multi-model prediction with timing
• Post-processing: Confidence scoring and recommendations

🔧 Technical Details

Input Requirements

• Image Format: PNG, JPG, JPEG
• Input Size: 224x224x3 (RGB)
• Preprocessing: Automatic resizing and normalization

Output Format

• Class Prediction: One of 11 satellite categories
• Confidence Score: Percentage confidence (0-100%)
• Inference Time: Milliseconds for prediction
• Probability Distribution: Full softmax output for all classes

📈 Performance Optimization

• Model Caching: Models are cached using @st.cache_resource
• Lazy Loading: Models loaded only when needed
• Efficient Preprocessing: Optimized image pipeline
• Memory Management: Automatic cleanup of model objects

🚢 Deployment

Hugging Face Spaces

The application is deployed on Hugging Face Spaces using:

• Runtime: Python 3.9
• Framework: Streamlit
• Resources: CPU-optimized for inference

Docker Deployment

FROM python:3.9-slim

WORKDIR /app
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

COPY . .
EXPOSE 7860

CMD ["streamlit", "run", "app.py", "--server.port=7860", "--server.address=0.0.0.0"]

🤝 Contributing

We welcome contributions! Please follow these steps:

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

Development Setup

# Clone your fork
git clone https://github.com/YOUR_USERNAME/satellite-classification-dashboard.git

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install development dependencies
pip install -r requirements.txt

# Run in development mode
streamlit run app.py --logger.level=debug

📄 License

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

🙏 Acknowledgments

• Dataset: Spark 2021 dataset used for training
• Frameworks: TensorFlow, Streamlit, Plotly
• Models: Pre-trained architectures from TensorFlow/Keras
• Hosting: Hugging Face Spaces for deployment

📞 Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Email: bhavithrass@gmail.com

🔮 Future Enhancements

• Real-time Video Classification
• Batch Processing for multiple images
• Model Fine-tuning Interface
• Custom Model Upload
• Export Predictions (CSV/JSON)
• API Endpoint for programmatic access
• Mobile App version
• Advanced Visualization with satellite orbit data

📊 Usage Statistics

Track your application usage:

• Classifications Performed: Real-time counter
• Popular Models: Usage analytics
• Performance Metrics: Response time tracking

---

Built with ❤️ using Streamlit and TensorFlow

For more information, visit our Hugging Face Space