Advanced AI-powered coral reef health monitoring through acoustic analysis
AquaListen is a cutting-edge marine conservation tool that uses machine learning to analyze underwater audio recordings and assess coral reef ecosystem health. Our system provides real-time insights into reef biodiversity and stress levels, enabling proactive conservation efforts.
- 🎵 Audio Analysis: Process underwater recordings to detect marine life patterns
- 🤖 AI Classification: Custom-trained neural network for reef health assessment
- 📊 Real-time Dashboard: Interactive web interface with live monitoring
- 🗺️ Site Management: Track multiple reef locations and monitoring sites
- 📈 Batch Processing: Analyze large datasets efficiently
- 🚨 Alert System: Automated notifications for reef stress detection
- Training Dataset: 57,000 audio samples from ReefSet v1.0
- Model Architecture: Deep neural network with feature extraction
- Performance: 82.3% accuracy on reef health classification
- Categories: Healthy, Stressed, Ambient water conditions
- Backend: FastAPI + TensorFlow + Librosa
- Frontend: React + TypeScript + Tailwind CSS
- Audio Processing: 16kHz sampling, 10-second segments
- Deployment: Docker-ready with TensorFlow SavedModel
- Python 3.8+
- Node.js 16+
- 4GB+ RAM for model inference
-
Clone the repository
git clone https://github.com/Bavan23/AquaListen.git cd AquaListen -
Install Python dependencies
pip install -r requirements.txt
-
Install Node.js dependencies
cd app npm install -
Start the system
# Start API server python aqualisten_api.py # Start frontend (new terminal) cd app npm run dev
-
Access the application
- Web Interface: http://localhost:3000
- API Documentation: http://localhost:8000/docs
| Metric | Score |
|---|---|
| Accuracy | 82.3% |
| Precision | 81.9% |
| Recall | 82.5% |
| F1-Score | 82.2% |
Training Data: 57,000 marine audio samples Categories: Healthy (32k), Stressed (18k), Ambient (7k)
- WAV (recommended)
- MP3
- FLAC
- M4A
Optimal Settings: 16kHz sample rate, 10-second duration
- Navigate to Upload page
- Select audio file (.wav, .mp3, .flac, .m4a)
- Click "Analyze" to get reef health assessment
- View confidence scores and acoustic features
- Go to Batch Processing page
- Upload multiple files (max 20 per batch)
- Monitor processing progress
- Download results as CSV
- Add monitoring sites in Sites page
- Upload regular recordings for each site
- Track health trends over time
- Set up automated alerts
- Spectral centroid and bandwidth
- Zero-crossing rate analysis
- 13 MFCC coefficients
- Energy distribution patterns
- Healthy: High biodiversity, complex acoustic patterns
- Stressed: Reduced acoustic complexity, mechanical noise
- Ambient: Background ocean sounds, minimal biological activity
AquaListen supports coral reef conservation by:
- Early Detection: Identify reef stress before visual degradation
- Non-invasive Monitoring: No physical disturbance to marine life
- Scalable Assessment: Monitor large areas cost-effectively
- Data-driven Decisions: Provide quantitative health metrics
We welcome contributions to improve AquaListen:
- Fork the repository
- Create feature branch (
git checkout -b feature/amazing-feature) - Commit changes (
git commit -m 'Add amazing feature') - Push to branch (
git push origin feature/amazing-feature) - Open Pull Request
This project is licensed under the MIT License - see the LICENSE file for details.
- ReefSet v1.0 dataset for training data
- Marine biology research community
- Open source audio processing libraries
- Coral reef conservation organizations
- Project Lead: [Bavan]
- Email: [bavan2312@gmail.com]
- Website: https://aqualisten.org
🌊 Protecting coral reefs through innovative acoustic monitoring technology