What cutting-edge project are you working on right now?
Visual 3D perception is fundamental towards enabling machine perception to "see" environments as we do, for instance for robots interacting with their environments, autonomous driving, or creating virtual content reflecting real objects or scenes.
We are working on driving forward 3D perception to the next level by coupling high-quality geometry, color, and semantics to enable reconstruction and recognition of complex, fine-grained phenomena in real world environments that we see around us.Our ScanNet++ project aims to enable learning of general priors for 3D scene representations to allow for fine-grained semantic understanding as well as photorealistic synthesis of RGB views and millimeter-resolution geometry.
What was the key experience that made you want to do research in data science and what fascinates you about working with MDSI?
I really like working with visual data (particularly 3D visual data), and many of the interesting research questions regarding machine perception or synthesis of 3D (e.g., recognizing objects and their functions) are data-driven problems. The MDSI bridges across various fields, and I find that exchange of ideas across disciplines is very beneficial towards driving cutting-edge science.
What paradigm shift do you expect within the next ten years, triggered in particular by the Institute's interdisciplinary approach?
Rapid developments across a variety of fields in data-driven sciences have made the whole area very exciting. I'm pretty interested -- both algorithmically and interpersonally -- in more efficient distillation of knowledge and insights across different kinds of data and fields (especially since in the 3D domain, data is typically more limited).