Kevin J Liang

Papers

Total Citations

H-Index

About

Kevin J Liang is a rising researcher in computer vision whose work pushes the boundaries of object detection toward open-world understanding. His primary research focuses on enabling detection systems to recognize both known and novel object categories—a critical capability for real-world applications like autonomous driving, robotic manipulation, and navigation. Liang's most cited work, "Extending One-Stage Detection with Open-World Proposals" (2022, 9 citations), tackles the challenging problem of Open World Detection (OWD), where models must generalize to unseen classes without retraining. By extending one-stage detectors to generate proposals for novel objects, Liang addresses a fundamental limitation in traditional closed-set detection. His contributions are particularly significant for safety-critical systems that must handle unexpected scenarios. While still early in his career, Liang's work demonstrates a clear trajectory toward making perception systems more robust and adaptable. His research sits at the intersection of object detection, open-set recognition, and practical deployment, offering promising directions for students and researchers interested in building vision systems that can truly operate in the wild.