🌊 HydroVexel - AI-Powered Oil Spill Detection System

AI-Powered Protection for Our Oceans ✨

An AI system that detects oil spills from satellite imagery with 94.57% accuracy. Built for the GDG Noida Build-a-thon to help protect marine environments through rapid detection and response.

---

🌐 Try HydroVexel Live

🎥 Demo Video: Click Here to Watch

📋 Table of Contents

• About
• Why This Matters
• Key Features
• Technology Stack
• Model Performance
• Quick Start
• How to Use
• Results
• Future Plans
• About the Developer
• Acknowledgments

---

🎯 About

HydroVexel is an AI-powered oil spill detection system developed for the GDG Noida Build-a-thon. It uses deep learning to analyze satellite images and instantly identify oil spills in our oceans.

Why This Matters

Oil spills threaten:

• 🐋 Marine Life - Kills fish, mammals, and birds
• 🏖️ Coastlines - Pollutes beaches and shores
• 💰 Economies - Damages fishing and tourism
• 🌍 Environment - Long-term ecological damage

Traditional detection is slow and manual. HydroVexel provides instant, accurate detection for faster response.

---

✨ Key Features

• 🎯 Real-time Detection - Upload images, get instant results
• 📊 94.57% Accuracy - Powered by U-Net + Attention architecture
• 🗺️ Visual Analysis - Heatmaps, overlays, and confidence scores
• ☁️ Cloud Storage - Save and track detections over time
• 📱 Easy to Use - Simple web interface, no installation needed
• 🌊 Beautiful UI - Ocean-themed design with smooth animations

---

🧠 System Architecture

HydroVexel follows a modular pipeline for accurate oil spill detection:

Architecture Overview: System Architecture

• Preprocessing: Image resizing, normalization, and augmentation
• Model: U-Net with Attention Gates for segmentation
• Post-Processing: Thresholding and mask refinement
• Visualization: Detection overlays, confidence maps, and dashboards
• Database: Supabase cloud storage for historical tracking

🛠️ Technology Stack

AI & Machine Learning:

• TensorFlow 2.19 & Keras - Deep learning
• PyTorch 2.0 - Model development
• U-Net + Attention Gates - Architecture

Data Processing:

• NumPy & Pandas - Data handling
• OpenCV - Image processing
• Matplotlib & Seaborn - Visualizations

Web App:

• Streamlit - Web interface
• Supabase - Database & storage
• Python 3.10 - Backend

---

📊 Model Performance

Metric	Score
Accuracy	94.57%
Precision	96.22%
Recall	94.69%
F1-Score	95.45%
Dice Coefficient	0.8984

What This Means:

• ✅ Detects 95% of actual oil spills
• ✅ Very few false alarms (96% precision)
• ✅ Reliable for real-world use

---

🚀 Quick Start

Option 1: Use the Live App (Recommended)

Just visit: https://hydrovexel.streamlit.app/

Option 2: Run Locally

1. Clone the repo

git clone https://github.com/simplysandeepp/Oil-Spill-Detection
cd hydrovexel

2. Install dependencies

pip install -r requirements.txt

3. Run the app

streamlit run streamlit_app.py

4. Open in browser

http://localhost:8501

---

💡 How to Use

Simple Steps:

1. Visit the Website → hydrovexel.streamlit.app

2. Upload Image → Drag & drop or browse (JPG/PNG)

3. Adjust Settings (optional)

   • Confidence threshold
   • Overlay transparency

4. Click "DETECT" → AI analyzes the image

5. View Results

   • Detection overlay
   • Confidence heatmap
   • Binary mask
   • Coverage statistics

6. Explore History → View past detections and gallery

---

🎨 Results

Dataset Distribution

Dataset Loading

Oil Spill Distribution

Model Architecture

Training Progress

Performance Metrics

Confusion Matrix

Quality Heatmap

Model Predictions

Best vs Worst Predictions

Application Screenshots

Landing Page

Oil Spill Information

Upload Interface

Detection Results

Analysis Dashboard

Live Database

Detection History

---

🔮 Future Plans

• 🛰️ Real-time Satellite Integration - Automatic monitoring
• 📈 Time-series Analysis - Track spill evolution
• 📱 Mobile App - iOS and Android versions
• 🌍 Multi-language Support - Reach more users
• 🔔 Alert System - Instant notifications
• 📊 Advanced Reports - Automated report generation

---

👥 Team

Project Lead: Sandeep Prajapati

AI Enthusiast | Environmental Tech Developer

Built for GDG Noida Build-a-thon to combine AI with environmental protection.

Education: B.Tech in CSE (AI & ML) - Galgotias University (2023-2027)

Roles:

• Google Student Ambassador
• GSSOC'25 Mentor
• Core Member @ GDG OC GU
• Super Contributor @ Hacktoberfest'25

Connect:

Team Members

Siya Kumari

Khushi Rawat

Vansh Chhabra

---

🙏 Acknowledgments

Built For:

• GDG Noida Build-a-thon - Project motivation and platform

Special Thanks:

• GDG Noida Community - Support and guidance
• Galgotias University - Academic resources

Data & Tools:

• Zenodo Dataset - Training data
• TensorFlow & Streamlit - Development frameworks
• Supabase - Cloud infrastructure

---

📜 License

MIT License - See LICENSE file

---

Built with ❤️ for GDG Noida Build-a-thon | Protecting Our Oceans with AI