Angela Dai receives Eurographics Young Researcher Award

Based on an email interview with Kate Salesin
Posted on October 7, 2022

We sat down with Professor Angela Dai, recipient of the 2022 Eurographics Young Researcher Award, to chat about the professor life, her long-term vision, and more. Congratulations, Angela!

Meet Angela

Angela Dai is an Assistant Professor at the Technical University of Munich where she leads the 3D AI group. Prof. Dai's research focuses on understanding how the 3D world around us can be modeled and semantically understood. Previously, she received her PhD in computer science from Stanford in 2018 and her BSE in computer science from Princeton in 2013. Her research has been recognized through a Eurographics Young Researcher Award, Google Research Scholar Award, ZDB Junior Research Group Award, an ACM SIGGRAPH Outstanding Doctoral Dissertation Honorable Mention, as well as a Stanford Graduate Fellowship.

Q & A with WiGRAPH

What do you love about your job?

Angela: I really like working on tangible topics with cool visuals, which is mostly how I ended up working on reconstruction and semantic understanding of 3D scenes. In particular, understanding indoor environments – because I get to interact with them all the time :) It's fun to map things that we see to machine representations. Being a professor allows me to sketch out and define problems and solutions that I believe in, with the ability to tackle large-scale problems with a team of budding researchers – I think that freedom is difficult to find in other environments.

Has any aspect of your life as a professor surprised you? What advice would you share with someone who is thinking of pursuing this career?

Angela: There are a lot of hats to wear as a professor, many of which one might not learn about explicitly in grad school – not just going from student to advisor in terms of research, but fundraising and budgeting, teaching and developing new courses, service, community activities, marketing, so I'm still learning. Figuring it out as you go is kind of par for the course (hopefully!).

I think being extroverted helps, but I tend to be more the diffident type, so for anyone like that who might sometimes wonder if they're capable, we're generally all here because we like to learn, and mostly everything is learnable :)

Your body of work has consistently focused on 3d scene understanding – what initially drew you to this topic?

Angela: When I started my PhD, commodity depth sensors were getting attention and deep learning had gained significant traction in image classification, so that combination made a lot of sense. Plus, I really liked the tangible appeal of reconstruction of 3D scenes, which is concrete enough to measure and visualize, but not without significant challenges... I reconstructed a bunch of rooms for my parents while visiting once, and they were more confused that they didn't look like movie-quality assets. So that's of course one part of the goal.

What levels or facets of scene understanding do you hope to achieve in a year, 5 years, and your whole career?

Angela: That's a big question :) I'm pretty interested in developing parametric models for 3D shape and semantics, for instance learned from clean, complete synthetic data that can then be fit to real-world observations like single-view images or monocular video. Longer-term, I think about re-imaging 3D operators and representations used for learning, as (sparse) convolutions or unstructured point processing are not the most efficient nor the most intuitive for understanding 3D/4D environments. Overall, I'm fascinated by re-creating the structures and interactions of real environments around us in digital representations that can be manipulated and interacted with, democratized to commodity-level content creation.

Your research lies at the intersection of computer graphics, computer vision, and machine learning – three prolific fields. Do you ever worry about “keeping up” with their rapid pace? If so, how do you manage that anxiety?

Angela: The pace is certainly remarkable these days, which makes it super cool to see; between the advances in shape reconstruction, neural rendering, etc., there is no lack of new things to try. So I find it pretty cool to always be learning and appreciating new ideas, though I can't say that it doesn't induce any pressure at the same time. But I do think the positive aspects outweigh it; there's always a great set of people to chat with, and still big open challenges in problems like content creation, reconstruction, dynamics and interactions.

What activities or discoveries have brought you joy outside of work recently?

Angela: I quite enjoy jogging and going to the gym again – there's a lot of nature near where I live, which is really nice for running. Also turn-based strategy games!