Stuart Golodetz

Papers

Total Citations

216

H-Index

About

Stuart Golodetz is a leading researcher in computer vision, specializing in dense 3D reconstruction, semantic scene understanding, and real-time instance segmentation. His most impactful work, "Incremental Dense Semantic Stereo Fusion for Large-Scale Semantic Scene Reconstruction" (2015, 209 citations), pioneered a method that enables robots to simultaneously perceive 3D structure and recognize objects in large, complex environments—a critical step toward autonomous navigation and manipulation. This work addresses the challenge of moving beyond controlled settings to dynamic, real-world scenes. Golodetz further advanced the field with "Straight to Shapes++: Real-time Instance Segmentation Made More Accurate" (2019), which improved the speed and precision of instance segmentation for time-sensitive applications like autonomous driving and drone navigation. His research on 3D camera re-localization in changing indoor scenes (2020) tackles the practical problem of maintaining accurate pose estimation when environments evolve. With over 200 citations on his seminal work, Golodetz has made foundational contributions that bridge the gap between academic computer vision and real-world robotic deployment, inspiring new approaches to large-scale scene understanding.