🌍 Satellite Image Segmentation Space

This project allows you to upload or capture satellite images, run segmentation predictions using YOLOv11, UNet, and Mask R-CNN models, and visualize each segmented class with corresponding area calculations.

🧪 Prototype Setup with GUI

---

🚀 Features

• 📤 Upload custom images or 🗺️ capture from live map view
• 🧠 Predict segmentation masks using deep learning models
• 🖼️ Split output into 6 individual class-wise images:
  • Buildings 🏢
  • Hills ⛰️
  • Land 🌾
  • Road 🛣️
  • Vegetation 🌳
  • Water 🌊
• 📏 Calculate and display the pixel area covered by each class

---

🛠️ Installation

1. Clone the repository

git clone https://github.com/praveensunkara19/SatelliteSeg-Yolo-Unet-MaskRcnn.git
cd SatelliteSeg-Yolo-Unet-MaskRcnn

2. Set up a virtual environment (optional but recommended)

python -m venv myenv
myenv\Scripts\activate     # On Windows
#or 
source myenv/bin/activate  # On macOS/Linux

3. Install dependencies

pip install -r requirements.txt


4.▶️ How to Run
py app.py


📂 Folder Structure

SatelliteSeg-Yolo-Unet-MaskRcnn/
├── app.py                     # Main Gradio app
├── utils/                     # Helper scripts and predictors
├── models/                    # Model weights (.pt, .h5, etc.)
├── assets/                    # Visual assets (optional)
├── config.py                  # Class labels and config
├── requirements.txt           # Dependencies
├── .gitignore
├── LICENSE
└── README.md

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

🙌 Acknowledgments Ultralytics YOLO Detectron2 by Facebook Research Gradio for the UI Esri Satellite Maps for live map tiles

Author @praveensunkara19

---