Graphics Talk

— 2:00pm

AKSHAT DAVE , Ph.D. Candidate, Computational Imaging Lab, Rice University

Polarization-aided Inverse Rendering

Reconstructing the geometry and appearance of an object from multiple images is inherently ill-posed.  Recent progress in representing scene properties as coordinate-based neural networks have facilitated neural inverse rendering resulting in impressive geometry reconstruction and novel-view synthesis. Our key insight is that polarization of light is a useful cue for neural inverse rendering as polarization strongly depends on surface normals and is distinct for diffuse and specular reflectance. With the advent of commodity, on-chip polarization sensors, capturing polarization has become practical. First, we introduce a single-shot approach that exploits these polarization cameras for accurate separation of diffuse and specular radiance. Then we propose PANDORA, a polarimetric inverse rendering approach based on implicit neural representations. From multi-view polarization images of an object, PANDORA jointly extracts clean object geometry free from texture artefacts, separates out the diffuse and specular radiance fields, and estimates illumination incident on the object under practical unstructured scenarios and strong specularities. — Akshat Dave is a Ph.D. candidate in the Computational Imaging Lab at Rice University, advised by Prof. Ashok Veeraraghavan. His research interests are at the intersection of vision, graphics, and optics. He is interested in co-designing cameras and algorithms that exploit unique properties of light to solve challenging imaging and vision problems. His research includes non-line-of-sight imaging using ultrafast time-of-flight cameras and inverse rendering using polarization cameras. He received his Masters and Bachelors from Indian Institute of Technology Madras advised by Prof. Kaushik Mitra. Faculty Host: Ioannis Glioulekas In Person and Zoom Participation. See announcement.

For More Information:

Add event to Google
Add event to iCal