🌲 Campus Forest Classification using Satellite Imagery

A machine learning-based land cover classification project that identifies Forest and Non-Forest regions on a university campus using Sentinel-2 satellite imagery and Google Earth Engine.

---

📍 Study Area

University campus region defined using a custom polygon boundary (centered at approximately 80.02°E, 23.17°N).

---

🛰️ Data Source

• Satellite: Sentinel-2 Surface Reflectance (COPERNICUS/S2_SR_HARMONIZED)
• Date Range: November 2025 - January 2026
• Cloud Filter: < 30% cloudy pixel percentage

---

🧹 Preprocessing

• Cloud masking using QA60 band
• Median composite generation
• NDVI calculation: (B8 - B4) / (B8 + B4)

---

🤖 Classification Models

Three classifiers trained on manually digitized points:

• Random Forest - ee.Classifier.smileRandomForest(150)
• SVM - ee.Classifier.libsvm()
• XGBoost - ee.Classifier.smileGradientTreeBoost(10)

Labels:

• Forest = 1
• Non-Forest = 0

---

📊 Model Comparison

All models trained with 70/30 train-test split.

Model	Algorithm	Accuracy	Kappa	Confusion Matrix
Random Forest	smileRandomForest(150)	0.9230	0.8454	[[20, 3], [0, 16]]
SVM	libsvm	0.9574	0.9126	[[19, 1], [1, 26]]
XGBoost	smileGradientTreeBoost(10)	0.9761	0.9523	[[20, 1], [0, 21]]

Confusion Matrix Format: [[TN, FP], [FN, TP]]

---

� Project Structure

├── notebooks/
│   ├── random_forest/
│   │   └── forest_classification.ipynb
│   ├── svm/
│   │   └── forest_classification.ipynb
│   └── xg_boost/
│       └── forest_classification.ipynb
├── scripts/
│   └── classify_forest.py
├── .env
├── .gitignore
├── requirements.txt
└── README.md

---

⚙️ Installation

Prerequisites

• Python 3.10+
• Google Earth Engine account

Setup

1. Clone the repository

   git clone <repository-url>
   cd <project-directory>

2. Create virtual environment & install dependencies

   python -m venv .venv
   source .venv/bin/activate
   pip install -r requirements.txt

3. Authenticate Earth Engine

   earthengine authenticate

4. Configure environment

   Create a .env file:

   EE_PROJECT_ID=your-earth-engine-project-id

---

🚀 Usage

Run Notebooks

jupyter notebook

Open any notebook in notebooks/ directory and run all cells.

Run Script

python scripts/classify_forest.py

---

🗺️ Output

Binary classification map:

• 🌲 Dark Green = Forest
• ⬜ Light Gray = Non-Forest

---

Made with ❤️ using Google Earth Engine