CVPR Daily - Thursday

by Emma Alexander For its 11th annual meeting, the Computational Cameras and Displays workshop reunited in person to great success. In addition to several remote presenters and attendees in the morning, many of the in-person talks were standing-room-only, spilling past the 100+ seats in the room into the hallway. Speakers presented on a wide range of topics, from novel sensors and novel optics arranged in unique camera configurations to the data structures and algorithms that power their sensing capabilities. Vivek Goyal of Boston University spoke on the pitfalls and potentials of quantization in computational imaging, with a focus on techniques for improving the depth resolution of single-photon lidar. Carola-Bibiane Schönlieb of Cambridge brought deep mathematical rigor to data-driven image priors for inverse rendering. Peyman Millanfar of Google gave a clear and insightful explanation of denoising and its use as a building block for imaging, inverse problems, and machine learning. Kristina Monakhova , just graduated from UC Berkeley and soon to join MIT as a postdoc, described some of her thesis work, with a focus on physics-informed machine learning for lensless computational cameras. These remote talks will be available shortly on the CVPR virtual platform. 24 DAILY CVPR Workshop Thursday