Patrick Poirson

University of North Carolina at Chapel Hill

Papers

Total Citations

218

H-Index

About

Patrick Poirson is a leading researcher in computer vision and robotics, with a primary focus on bridging the gap between object detection and real-world robotic perception. His most influential work centers on developing datasets and efficient algorithms for active vision and 3D scene understanding. Poirson’s landmark contribution, the creation of a comprehensive dataset for benchmarking active vision (193 citations), provides over 20,000 RGB-D images and 50,000 densely annotated bounding boxes across nine indoor scenes, enabling the simulation of robotic vision tasks in everyday environments. This resource has become a critical benchmark for training fast object category detectors. Additionally, his pioneering work on "Fast Single Shot Detection and Pose Estimation" (25 citations) addresses the crucial need for simultaneous object detection and 3D pose estimation in navigation and robotics. By building on state-of-the-art convolutional networks, Poirson developed efficient sliding-window methods that allow robots to not only identify objects but also understand their spatial orientation in real time. His research directly impacts the development of more capable, perceptive autonomous systems, making him a key figure in advancing practical computer vision for robotics.