Project Overview
This project focuses on building a system that automatically segments and classifies different land types such as forests, water bodies, deserts, urban areas, and industrial zones from satellite images.
The gained insights will contribute directly to strategic decision-making, enabling data-driven market intelligence

Key Objectives

  *   Develop a system to segment and classify natural terrain, urban, and industrial zones using satellite imagery
  *   Design and train semantic segmentation models based on deep learning architectures, including recent models such as Segment Anything (SAM)
  *   Investigate open-source satellite imagery (e.g., Sentinel-2, Landsat) and additional geospatial layers (e.g. OpenStreetMap, Google Maps)
  *   Perform error analysis to guide iterative improvements in model accuracy and robustness
  *   Address practical challenges such as low-resolution
  *   Integrate evaluation pipelines to continuously assess performance across regions and timeframes
 

Requirements

  *   Theoretical and practical understanding of the basics of machine learning and deep learning for computer vision
  *   Familiarity with Python and PyTorch

Apply to this project here