Presentation

Policy-based Foveated Imaging and Perception
SessionComputational Photography
DescriptionUltra-high-resolution sensors capture fine details critical for visual perception, but acquiring all pixels at full resolution is often infeasible under realistic constraints. We introduce a real-time, task-aware foveated imaging framework that learns a sensor attention policy to dynamically allocate pixel bandwidth, extensively validated in simulation and on a 200-megapixel prototype.
Authors
Howard Xiao
Stanford University
Jan Ackermann
Stanford University
Boyang Deng
Stanford University
Gordon Wetzstein
Stanford University
Event Type
Technical Paper
TimeTuesday, 21 July 20262:50pm - 3:00pm PDT
LocationRoom 408 B
Session TimeTuesday, 21 July 20262:00pm - 3:30pm PDT
LocationRoom 408 B
Registration Categories
Full Conference Supporter
Full Conference
Next PresentationNext PresentationComputational Photography - Interactive Discussion
Similar Presentations
Broadband Hyperspectral 3D Imaging using Dispersed Structured Light
It’s Not Just a Phase: Creating Phase-Aligned Peripheral Metamers
LongE2V: Long-Horizon Event-based Video Reconstruction, Prediction, and Frame Interpolation with Video Diffusion Models