Research Papers
Explore our current research projects. More exciting projects coming soon!
A Theory of Joint Light and Heat Transport for Lambertian Scenes
How can we measure absorbed light?
CVPR 2024
When light gets absorbed, it becomes heat. The intensity of this heat source can be measured upto scale from thermal transients without any knowledge about the shape, illumination or material properties! This allows us to solve albedo-shading separation for arbitrary shape and illumination from a single view analytically!
Shape from Heat Conduction
Can thermal vibrations reveal the shape of an object?
ECCV 2024
We propose a novel approach to recover shape from thermal transients induced on objects heating up (or cooling down). The shape-dependent conduction of heat is robust to the complexities of light transport in visible spectrum and our method works for dark, textureless and even transparent objects.
Resolving Shape Ambiguities using Heat Conduction and Shading
Can Shape Laplacians and Shading uniquely determine the shape of an object?
ICCP 2025
The well-studied Shape-from-Shading problem suffers from a per-pixel cone ambiguity even for known light source direction. On the other hand, Shape Laplacians encode the intrinsic geometry of a surface and can be recovered for arbitrary unknown lighting, but suffers from local binary (convex/concave) ambiguities. This paper shows that the Shading constraint can break the binary ambiguity of Shape Laplacians. It further provides an approach to recover the light source direction by estimating a pseudo-shape using only the Shape Laplacians.
Projecting Trackable Thermal Patterns
How can we track points on dark and textureless surfaces?
CVPR 2024
We demonstrate that dynamic laser (visible light) patterns projected onto a surface cause local heating and produce thermal patterns that stick to the object when viewed using a thermal camera.
These patterns can be used to track points on dark, textureless and even transparent surfaces.
Research Poster
Explore our comprehensive research overview in poster format
Acknowledgements
These works were partly supported by NSF grants IIS-2107236, CCF-1730147, and NSF-NIFA AI Institute for Resilient Agriculture.