Satellite Image Segmentation Application

Welcome to the Satellite Image Segmentation Application, an intuitive tool designed to harness the power of deep learning for precise land cover classification from satellite imagery. This application leverages state-of-the-art machine learning techniques to segment images into various classes, aiding in environmental monitoring, urban planning, and agricultural analysis.

Table of Contents

• Features
• Technologies Used
• Installation
• Usage
• How It Works

Features

• User-Friendly Interface: Built with Gradio, the application provides an easy-to-use interface for uploading images and displaying results.
• Real-Time Segmentation: Process satellite images quickly and efficiently to generate segmented outputs in real time.
• Color-Coded Segmentation: Each class (water, land, road, building, vegetation, unlabeled) is represented by a distinct color for easy visualization and interpretation.
• Interactive Legend: A clear legend provides an overview of the segmentation classes and their corresponding colors.

Technologies Used

• Python: The core programming language for model development and data manipulation.
• TensorFlow/Keras: For building and training the deep learning model.
• Gradio: To create an interactive web interface for the application.
• NumPy: For numerical operations and handling image data.
• Pillow: For image processing tasks.

Installation

To run the application locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/yourusername/satellite-image-segmentation.git
   cd satellite-image-segmentation

2. pip install -r requirements.txt
3. Load your pre-trained model into the code where specified
4. python app.py Open your web browser and navigate to http://localhost:7860 to access the application.

Usage

Upload a satellite image by clicking the "Please select source Image" button. Click "Load above Image" to process the image. View the segmented output displayed alongside the interactive legend.

How It Works

The application uses a convolutional neural network (CNN) trained on a dataset of labeled satellite images. It processes the input image to predict pixel-wise labels based on learned patterns from the training data. The RGB representation of each class is applied to visualize the segmented areas effectively.

The segmentation classes include:

Water: Represented in yellow Land: Represented in purple Road: Represented in blue Building: Represented in dark purple Vegetation: Represented in green Unlabeled: Represented in gray