Bahram Zonooz

Advanced Engineering (Czechia), Eindhoven University of Technology

Papers

Total Citations

H-Index

About

Bahram Zonooz is a leading researcher in computer vision, with a primary focus on 3D scene understanding, adversarial robustness, and continual learning. His work centers on monocular depth estimation and structure-from-motion, addressing critical challenges in deploying deep learning models for real-world applications like robotics and autonomous driving. Zonooz’s major contributions include pioneering adversarial attack methodologies for monocular pose estimation, revealing vulnerabilities in deep neural networks that threaten reliable deployment. He has also advanced self-supervised and unsupervised depth estimation, notably through innovative image masking techniques that enhance robustness and continual learning frameworks that allow models to adapt from diverse, crowd-sourced videos without forgetting prior knowledge. His research, including papers on adversarial attacks (8 citations) and continual depth learning (7 citations), has garnered attention for its practical impact on safety-critical systems. Zonooz’s work on transformers in unsupervised structure-from-motion further underscores his commitment to pushing the boundaries of 3D perception. Through these efforts, he has established himself as a key figure in making computer vision systems more resilient, adaptable, and trustworthy for autonomous navigation and beyond.