Machine Learning and Pattern Recognition for Earth Observation: Hands-on Python Examples

Curso de Formação

Ano Letivo 2025/2026

The use of machine learning techniques has become increasingly relevant in the analysis of aerospace data, given the complexity and growing volume of information from remote sensors and satellites. Understanding the fundamentals of both supervised and unsupervised learning is essential for developing effective solutions for classification, segmentation, and the extraction of hidden patterns in large geospatial datasets. Recognizing patterns in satellite images?such as changes in land use, detection of bodies of water, or urban structures?requires mastery of robust computational techniques that can transform raw images into useful knowledge. The use of Python, along with libraries such as scikit-learn and OpenCV, provides a powerful and accessible environment for the development of practical applications, ranging from classical algorithms to advanced deep learning models. These models, particularly convolutional neural networks (CNNs), enable the automatic identification of objects in high-resolution imagery?such as aircraft, vehicles, or weather formations?optimizing monitoring processes and response to extreme environmental events. The thematic unit "Machine Learning and Pattern Recognition in Aerospace Data" aims to equip students with the critical and creative use of these tools, allowing them to implement solutions that meet the demands of contemporary aerospace engineering. By the end of the module, students are expected to understand the core principles of machine learning algorithms, know how to apply them to various types of aerospace data, and be capable of building intelligent systems to support data-driven decision-making.

Training proposed within the scope of the New Space Portugal Mobilizing Agenda (Ref. C644936537-00000046, Notice ACC02/CO5-i01/2022), financed by the Mobilizing Agendas for Business Innovation, through the Recovery and Resilience Program (RRP).