Konrad Karanowski

Wrocław University of Science and Technology

Papers

Total Citations

H-Index

About

Konrad Karanowski is a researcher focused on advancing continual learning in 3D computer vision, particularly through the lens of point cloud processing. His major contribution lies in developing memory-efficient methods that allow deep learning models to learn new 3D data without forgetting previously acquired knowledge—a critical challenge in autonomous systems and robotics. His most-cited work, "Continual learning on 3D point clouds with random compressed rehearsal" (2023), introduces a novel rehearsal strategy that uses random compressed representations to store past data, drastically reducing memory overhead while maintaining performance. This paper has garnered 11 citations, reflecting its early impact in a rapidly evolving field. Karanowski’s research bridges the gap between lifelong learning and 3D perception, offering practical solutions for real-world applications where models must adapt to new environments over time. His work is particularly notable for its focus on scalability and efficiency, making it relevant for resource-constrained platforms like drones or autonomous vehicles. As a rising voice in continual learning, Karanowski is shaping how machines can continuously learn from dynamic 3D worlds.