Multi-Modal Land Cover Segmentation from Remote Sensing Images

Requirements

• M.Sc. in Machine Learning, Data Science, Computer Science, Mathematics, Telecommunications, or similar
• Knowledge of Python
• Software development skills
• Basic concepts of image processing
• Basic concepts of data science, concerning data analysis, processing and machine learning

Description

• Satellite remote sensing images provide valuable information about land cover and land use, which can be used for various applications, including environmental monitoring, resource management, and urban planning. However, interpreting these images is a challenging task due to the complexity and variability of the Earth’s surface. This thesis proposes to investigate a multi-modal approach for land cover segmentation from Sentinel images at European Scale. The primary objective is to develop accurate and efficient models that can effectively leverage multi-modal information for land cover segmentation. Starting from existing datasets and resources, the aim of the thesis is to retrieve and process Sentinel feeds and other modalities to evelop a pipeline for computing the land cover on a European scale, starting from baseline approaches, and eventually developing ad-hoc methods. The thesis includes the following activities: (i) Data identification and acquisition (ii) state of the art analysis (iii) data analysis and preprocessing (iv) experiments evaluation

Contacts

• Edoardo Arnaudo
• Marco Galatola