Yonghao Xu: Machine learning for remote sensing

Recent years have witnessed significant advancements in machine learning for remote sensing and Earth observation. However, there are still many challenges in constructing intelligent and secure machine learning models within the geoscience and remote sensing realm. 

This presentation will introduce recent developments in machine learning for remote sensing through four perspectives: intelligent remote sensing data interpretation, vision and language for remote sensing, trustworthy remote sensing models, and AI for environmental monitoring. 

To begin, I will introduce a consistency-regularized region-growing network, which can achieve robust land cover classification performance with limited point-level annotations. Following this, I will discuss the threat of adversarial attacks in the remote sensing domain and present the proposed Mixup-Attack in detail. 

Subsequently, I will briefly introduce our recent work on text-to-image generation for remote sensing data. Finally, I will share some preliminary works on natural hazard monitoring (e.g., landslide and wildfire detection) using machine learning techniques and satellite remote sensing data.

Yonghao Xu is an Assistant Professor at the Computer Vision Laboratory (CVL), Linköping University, Sweden. He received his Ph.D. degrees in photogrammetry and remote sensing from Wuhan University, China, in 2021. From 2021 to 2023, he was a Postdoctoral Researcher with the Institute of Advanced Research in Artificial Intelligence (IARAI), Austria. 

He was a recipient of the First Place award in the IEEE Geoscience and Remote Sensing Society (GRSS) Data Fusion Contest in 2018. Since 2022, he has been working as the co-lead of the Benchmarking Working Group in the IEEE GRSS Image Analysis and Data Fusion Technical Committee. His research interests include remote sensing, computer vision, and machine learning.