Seonhoon Lee
Papers
2
Total Citations
19
H-Index
2
About
Seonhoon Lee is a researcher specializing in computer vision, with a particular focus on scene change detection—a critical capability for visual surveillance, anomaly detection, and mobile robotics. His work addresses the fundamental challenge of detecting changes in real-world environments where image pairs are often imperfectly aligned, moving beyond idealized assumptions that limit practical deployment. Lee's major contributions include pioneering a dual-task learning framework that leverages both dense correspondence and mis-correspondence, enabling robust change detection even with coarse scene matches. This work, published in 2022, has already garnered 13 citations for its practical significance. More recently, in 2024, Lee advanced the field with a semi-supervised approach that uses distillation from feature-metric alignment, reducing the need for expensive labeled data while maintaining high detection accuracy. His research bridges the gap between theoretical computer vision models and real-world applications, making autonomous systems more reliable in dynamic environments. Lee's innovative methods are shaping the next generation of visual monitoring technologies.
Research Focus
Key Achievements
Top Papers
- 1
- 2