Research Profile
Current Role and Institutional Context
Leandro Stival is a Postdoctoral Researcher in the Artificial Intelligence Group at Wageningen University & Research (WUR), Netherlands. His research is situated at the intersection of advanced machine learning and environmental science, with a primary focus on developing scalable, self-supervised representation learning frameworks for multimodal spatiotemporal data.
Research Methodologies and Scientific Inquiries
My work investigates how high-dimensional, unlabeled data from Satellite Imagery (Remote Sensing), Video, and Time Series can be utilized to build generalizable foundation models. I employ techniques from contrastive learning, masked autoencoders, space state models and generative modeling to autonomously discover structural patterns in Earth observation data.
Key scientific questions I address include:
- How can self-supervised learning capture long-range temporal dependencies in irregular environmental time series?
- To what extent can cross-modal alignment between multispectral and spatial data improve representation robustness?
- How can we design domain-agnostic architectures that generalize across diverse geographical regions and sensor types?
Academic Biography
Leandro Stival earned his Ph.D. in Computer Science from the University of Campinas (UNICAMP), Brazil, where his doctoral research focused on self-supervised deep learning and computer vision applied to video colorization and spatiotemporal analysis. During his doctoral studies, he conducted a research period at Wageningen University & Research, focusing on remote sensing applications. His academic background is characterized by a strong foundation in computer science, specifically in digital signal processing, graph theory, and neural network architectures.
Throughout his career, he has published in high-impact venues such as the ISPRS Journal of Photogrammetry and Remote Sensing and PLOS ONE. His research spans multiple application domains, including change detection on earth observation, natural video processing, and sports analytics, demonstrating a commitment to interdisciplinary AI. Currently, he contributes to the development of AI-driven solutions for Earth observation, aiming to bridge the gap between theoretical machine learning and practical environmental monitoring challenges.