sheng-yu_wang at berkeley.edu
I will be serving a mandatory military service for 4 months during Spring 2020. It will likely start in April.
I am a student researcher at BAIR. Recently, I received my BA degree from UC Berkeley where I spent one wonderful year at BAIR, working with Oliver Wang, Richard Zhang, Andrew Owens and Alexei (Alyosha) Efros. Prior to joining BAIR, I worked with Fisher Yu at Berkeley DeepDrive and Michael Crommie on condensed matter physics research.
Currently I work on computer vision, graphics, and machine learning. I started from understanding the structures and properties of materials to those of perception. Perceptual data contains diverse and intricate concepts, ranging from low-level features to high-level semantics, all of which fascinates me a lot.
My main research interests are, but not limited to:
- Building controllable models that synthesize beautiful pictures/music: I am interested in constructing models that are able to provide better controls in generation (e.g., shape, pose, style, composition). I would love to have them as tools for people to express themselves artistically more freely.
- Capturing realism: What is the set of perceptual perturbations in the real image/audio manifold? In other words, can we find a family of transformations that serves as "small" changes for humans? Can we use it to understand the boundaries of the real manifold (i.e. apply perturbations until perceive as unreal)?
Many thanks to Richard Zhang for sharing his website template!
Detecting Photoshopped Faces by Scripting Photoshop
Sheng-Yu Wang, Oliver Wang, Andrew Owens, Richard Zhang, Alexei A. Efros.
In ICCV, 2019.
Direct reversible writing of high carrier density, high mobility circuitry into two-dimensional van der Waals heterostructures
Wu Shi, Salman Khan, Lili Jiang, Sheng-Yu Wang, Hsin-Zon Tsai, Dillon Wong, Takashi Taniguchi, Kenji Watanabe, Fen Wang, Michael Crommie, Alex Zettl.