I am an assistant professor in the Department of Computer Science at Brown University. My research is in 3D computer vision and machine learning. Specifically, I am interested in 3D spatiotemporal visual understanding of human physical interactions. I build methods for human-centric, object-centric, and interaction-centric understanding of our world from videos and images.
Prospective Students: We are always looking for motivated students to join us. Please see this page for information on PhD admissions. If you are already an undergrad/grad at Brown, please email me directly.
Outside of Brown, I spend part of time as a Visiting Academic at Amazon Robotics AI. I was previously a postdoc at Stanford University with Leonidas Guibas. I obtained my PhD at the Max Planck Institute for Informatics advised by Christian Theobalt and Antti Oulasvirta. I have also spent time at Microsoft Research Redmond, the University of Michigan, Ann Arbor, the Honda Research Institute, and College of Engineering Guindy.
srinath@brown.edu
Updates
[ Nov-2021 ]: Our new report and companion website provides a consolidated overview of coordinate-based neural networks (neural fields) in visual computing and beyond by reviewing over 250 papers.- [ Oct-2021 ]: I helped co-organize the Second 3DReps workshop at ICCV. The workshop recording is now available at this link.
- [ Oct-2021 ]: HuMoR, a human motion model for robust pose estimation will be presented at ICCV 2021.
- [ Apr-2021 ]: Srinath received a Google Research Scholar award to further the group's research on object-centric perception/synthesis for mixed reality.
- [ Mar-2021 ]: Our upcoming ICRA paper shows how to learn to densely reconstruct and canonicalize shapes with only weak supervision. Please see the project page for details, code, and data.
- [ Nov-2020 ]: Our upcoming NeurIPS spotlight paper shows how to learn 3D canonical spatiotemporal representations of dynamically moving point clouds. Please see the project page for details.
- [ Aug-2020 ]: We organized the 3DReps Workshop at ECCV 2020. You can watch the recorded sessions on YouTube (Link: Session 1, Session 2).
- [ Aug-2020 ]: Upcoming paper at ECCV on sparse multiview 3D surface recontruction aka Pix2Surf.
- [ Jan-2020 ]: I am serving on the program committee (Area Chair) for IEEE VR 2020.
- [ Jan-2020 ]: Paper on predicting object dynamics of previously unseen objects accepted to WACV.
- [ Dec-2019 ]: Davis Rempe is presenting our paper on multi-view aggregation for 3D reconstruction at NeurIPS. Code and datasets are available.
- [ Jun-2019 ]: Watch the interview we did for our CVPR 2019 paper:
- [ May-2019 ]: Our paper on learning to generate human-object interactions was awarded an honorable mention at Eurographics 2019. Read my blog post on this work on the SAIL Blog.
- [ Mar-2019 ]: Can we estimate the 6D pose and size of novel object instances never encountered before? Our upcoming CVPR oral paper shows one way (video).
- [ Mar-2019 ]: Read my Twitter thread about interesting hand facts that you may not have known before.