Yuseok Bae

Electronics and Telecommunications Research Institute

Papers

Total Citations

H-Index

About

Yuseok Bae is a researcher advancing the reliability of object detection in safety-critical domains such as autonomous driving and surgical robotics. His primary research focus lies in uncertainty estimation for computer vision, particularly within anchor-free object detection frameworks. Bae’s most influential work, "Localization Uncertainty Estimation for Anchor-Free Object Detection," has accumulated over 30 citations across its 2020 and 2023 iterations, reflecting growing interest in making AI perception systems robust to sensor noise and incomplete data. He identified critical limitations in existing detectors that fail to account for localization uncertainty in unstable environments, proposing novel methods to quantify and integrate this uncertainty into detection pipelines. This contribution directly addresses the reliability gap between laboratory-trained models and real-world deployment, where sensor degradation or environmental disturbances can lead to catastrophic failures. By enabling object detectors to express confidence in their spatial predictions, Bae’s work enhances the safety and trustworthiness of autonomous systems, making it a valuable reference for researchers tackling uncertainty-aware perception. His ongoing efforts continue to bridge the gap between theoretical uncertainty quantification and practical, high-stakes applications.