Federico Stella

Institute of Science and Technology Austria

Papers

Total Citations

H-Index

About

Federico Stella is a researcher at the intersection of computer vision, robotics, and geometric deep learning, with a primary focus on 3D shape analysis and self-supervised learning. His most notable contribution is the development of "Self-supervised Spherical CNNs," a pioneering framework that learns to canonically orient 3D surfaces without manual annotations. This work, published in 2020 and garnering 9 citations, addresses a fundamental challenge in computer vision and robotics: reliably defining a consistent orientation for 3D objects. By leveraging spherical convolutions and self-supervision, Stella's approach replaces traditional handcrafted geometric heuristics with a learned, data-driven method, enabling more robust and generalizable surface alignment. This innovation has direct implications for applications like object recognition, manipulation, and scene understanding, where canonical orientation is critical. Stella's research bridges the gap between geometric reasoning and modern deep learning, offering a principled alternative to designer-specified cues. His work stands out for its elegance and practical impact, providing a foundation for future advances in 3D perception and autonomous systems.