Gabriel Brostow
Papers
2
Total Citations
8
H-Index
2
About
Gabriel Brostow is a leading figure in computer vision, with his research primarily focused on 3D scene reconstruction, augmented reality (AR), and semantic understanding of dynamic environments. He is best known for pioneering work in non-rigid scene reconstruction, enabling the 3D modeling of deformable objects and human motion from monocular video—a foundational contribution that has garnered thousands of citations. Brostow’s lab has consistently pushed the boundaries of efficient 3D representation, as seen in recent works like "Heightfields for Efficient Scene Reconstruction for AR," which tackles real-time RGB-based reconstruction for augmented reality. His team’s 2024 paper, "AirPlanes," introduces a novel method for accurate plane estimation using 3D-consistent embeddings, advancing geometric understanding for robotics and AR. A Professor at University College London, Brostow has also made significant strides in semantic segmentation and video understanding, with his research regularly appearing at top venues like CVPR and ECCV. His work has been instrumental in bridging the gap between 2D vision and 3D geometry, influencing both academic research and practical AR applications.
Research Focus
Key Achievements
Top Papers
- 1Heightfields for Efficient Scene Reconstruction for AR5 citations · 2023
- 2AirPlanes: Accurate Plane Estimation via 3D-Consistent Embeddings3 citations · 2024