FPSica/ML4SAR-Tutorial

This tutorial approaches Synthetic Aperture Radar (SAR) as one more data modality to be integrated into Machine Learning applications, in contrast to traditional SAR-specific tutorials. The focus is on preparing SAR data to be ML-ready, with a clear goal of making it accessible for machine learning workflows.

Tutorial on Machine Learning for Synthetic Aperture Radar (ML4SAR)

Dr. Francescopaolo Sica

This tutorial approaches Synthetic Aperture Radar (SAR) as one more data modality to be integrated into Machine Learning applications. The focus is on preparing SAR data to be ML-ready, with a clear goal of making it accessible for machine learning workflows.

Key aspects covered include:

• The statistical, geometric, and radiometric characteristics of SAR data.
• Best practices for data access, processing, and transforming SAR data for use in machine learning pipelines.

This tutorial serves as a reference for ML practitioners, offering guidance on preparing SAR data for various applications, whether in image classification, detection, semantic segmentation, or regression. By focusing on data readiness for ML, it bridges the gap between SAR remote sensing expertise and machine learning methods.

Presented at the IADF School on Computer Vision for Earth Observation on September 11, 2024, University of Sannio, Benevento, Italy. The recorded tutorial will be made available soon. Link to the IADF School