Chenhao Li 李 辰昊

I am a Researcher at Sony. Previously, I was a Ph.D. student at Osaka University, supervised by Prof. Hajime Nagahara.. My research interests include computer vision, computational imaging, and computer graphics.

Email  /  Twitter  /  Google Scholar  /  Github

Most of my research is about estimating the physical world (shape, materials, illumination, etc.) using various types of sensing devices. Representative papers are highlighted.

With polarization cues, a lightweight model trained on a small dataset outperforms vision foundation models

An inverse rendering pipeline using polarization cues that supports both conductors and dielectrics.

Joint estimation of shape and subsurface scattering parameters from a single polarized image.

An inverse rendering pipeline using a combination of explicit-implicit polarimetric lighting representations and a physically-based polarimetric renderer.

Jointly estimate SSS parameters, surface normal, depth, roughness, and illumination by observing a flash and no-flash image pair of translucent objects.

Generate the entire Chinese font library by observing a few (less than 10) English letters.