🦠 PathoWatch — Pathogen Risk Intelligence Platform

Real-time vegetation pathogen detection and human disease risk analysis using satellite imagery, spectral intelligence, and live environmental data.

---

🚨 Problem Statement

Every year, millions of people in India and across South Asia are affected by preventable diseases — malaria, dengue, cholera, typhoid, and respiratory illnesses caused by air pollution. At the same time, crop diseases silently devastate agricultural yields before farmers even notice symptoms on the ground.

The core problem is a lack of early, location-specific risk intelligence:

• Farmers cannot detect vegetation stress before visible symptoms
• Public health systems lack hyperlocal, real-time insights
• Existing tools are expensive, slow, or too coarse
• No unified platform combines environmental + health intelligence

---

💡 Solution

PathoWatch is a unified platform that combines:

1. Satellite-based vegetation pathogen detection
2. Dynamic spectral intelligence using real scientific datasets
3. Human disease risk scoring using live environmental data
4. Actionable recommendations

---

🧠 PathoWatch – Complete Workflow (Simplified + Logical Flow)

---

🔷 0. System Initialization (Backend Start)

• Flask server starts
• PathogenMonitoringSystem is initialized
• Spectral library + detection models are loaded

👉 This prepares the system to process satellite images

---

🌍 MAIN WORKFLOW (5-Phase Pipeline)

---

🟢 PHASE 1: Data Infrastructure & Spectral Library

🎯 Goal:

Create reference “fingerprints” of diseases

⚙️ What happens:

• Fetch real spectral data from USGS library

• Generate synthetic pathogen signatures using:

  • Chlorophyll loss
  • Water stress
  • Leaf damage

📦 Stored Signatures:

• wheat_rust
• rice_blast
• late_blight
• bacterial_blight

👉 Each pathogen has unique wavelength behavior

---

🟡 PHASE 2: Preprocessing (Cleaning Satellite Data)

🎯 Goal:

Make raw satellite data usable

⚙️ Steps:

1. Atmospheric Correction

• Removes haze, dust, sunlight distortion
• Uses Dark Object Subtraction

2. Noise Reduction

• Smooth spectral values
• Spatial filtering (Gaussian)

3. Dimensionality Reduction

• Uses PCA
• Reduces hundreds of bands → important features

4. Feature Extraction

Calculates indices:

• NDVI (vegetation health)
• NDWI (water stress)
• NDRE (disease stress)

👉 Converts raw image → meaningful scientific data

---

🔵 PHASE 3: Detection (Core Intelligence)

🎯 Goal:

Detect disease patterns in pixels

⚙️ Techniques Used:

🧩 A. Spectral Angle Mapper (SAM)

• Compares pixel spectrum with pathogen signature
• Measures similarity (0 → 1)

👉 If similar → possible disease

---

🧠 B. 3D CNN (Deep Learning)

Uses:

• Spatial info (image)
• Spectral info (bands)

👉 Learns patterns automatically

---

🔁 Hybrid Approach

• SAM → initial detection
• CNN → refinement

👉 Combines physics + AI

---

🟠 PHASE 4: Prediction & Monitoring

🎯 Goal:

Generate disease risk maps

⚙️ Steps:

• Process hyperspectral image
• Create similarity maps (per pathogen)

📊 Risk Levels:

• 🔴 High risk (> 0.7)
• 🟡 Medium (0.4 – 0.7)
• 🟢 Low (< 0.4)

📦 Stored Outputs:

• Risk maps (PNG)
• Statistics
• Trends

👉 Output = visual + numerical insights

---

🔴 PHASE 5: Validation & Quality Control

🎯 Goal:

Ensure reliability

⚙️ Metrics:

• Spatial coherence
• Signal-to-noise ratio (SNR)
• Statistical consistency

🧠 Output:

• Quality score

• Recommendations:

  • Improve data
  • Increase resolution
  • Reduce noise

👉 Prevents false detections

---

🔄 ADDITIONAL PIPELINE (Advanced ML Version)

⚙️ Enhancements:

• SAM → generates training labels

• Feature extraction + PCA

• Train models:

  • Random Forest
  • SVM

• Ensemble prediction

👉 More accurate + adaptive system

---

🧬 Major Upgrade — Spectral Intelligence Pipeline (v2)

❌ Removed (Old System)

• Hardcoded spectral dictionaries
• Fixed wavelength arrays
• Static thresholds
• SAM-only detection

---

✅ Added (New System)

• 🌐 USGS spectral integration
• 📡 Dynamic wavelength extraction
• 📊 Adaptive thresholding
• 🤖 Hybrid ML (RF + SVM)
• ⚙️ Configurable targets

---

🧬 Pipeline Flow

Input Raster
   ↓
Extract Wavelengths
   ↓
Fetch USGS Spectrum
   ↓
SAM Matching
   ↓
Dynamic Thresholding
   ↓
Feature Extraction
   ↓
ML Training
   ↓
Ensemble Prediction
   ↓
Risk Map Output

---

🏗️ Project Architecture

Pathowatch-repo/
│
├── server.py
├── pathowatch_pipeline.py
├── pathowatch.py
├── index.html
│
├── Dockerfile
├── requirements.txt
├── Procfile
│
├── Browser_images/
│   ├── B02.tiff
│   ├── B03.tiff
│   ├── B04.tiff
│   └── B08.tiff
│
└── .env

---

🛠️ Tech Stack

Layer	Technology
Backend	Flask, Python
ML	Scikit-learn
Satellite	Google Earth Engine
Processing	rasterio, numpy
Frontend	HTML, JS, Leaflet
APIs	OpenWeather, WAQI
Deployment	Docker, Render

---

🚀 Running Locally

Python

pip install -r requirements.txt
python server.py

Docker

docker build -t pathowatch .
docker run --env-file .env -p 8080:8080 pathowatch

---

☁️ Deployment (Render)

• Runtime: Docker
• Region: Singapore
• Add environment variables
• Deploy

---

🚀 Future Scope

• Hyperspectral drones
• Deep learning models
• Global disease tracking
• Mobile app

---

📄 License

MIT License