Stefanos Kollias

National Technical University of Athens, University of Lincoln

Papers

Total Citations

H-Index

About

Stefanos Kollias is a leading researcher in computer vision and human activity recognition, with a particular focus on developing robust, real-world applicable systems. His work bridges the gap between traditional machine learning and modern deep learning approaches. A key contribution is his exploration of the Trace Transform for human action recognition, a method that captures invariant features from video sequences, as detailed in his highly cited 2013 paper (55 citations). This work laid a foundation for more resilient recognition in cluttered or dynamic environments. More recently, Kollias has advanced the field with his work on "Deep Bayesian Self-Training" (2020, 36 citations), a semi-supervised learning framework that leverages uncertainty to improve model performance with limited labeled data—a critical challenge for practical deployment. He has also contributed to online human activity recognition for Human-Robot Interaction (HRI), developing compact sequence encoding schemes that enable efficient, real-time analysis. His research is characterized by a commitment to methodological rigor and practical applicability, making significant strides toward machines that can understand and anticipate human actions in natural settings.