ECCV 2022 Workshop
Frontiers of Monocular 3D Perception:
Explicit vs. Implicit
Toyota Research Institute (TRI) & Toyota Technological Institute at Chicago (TTIC)
24th October 2022, Tel Aviv, Hybrid
Join live stream here (ECCV registration required) or join us in person in David Intercontinental Hotel - Grand Ballroom C.
Recent advances in neural implicit representations introduced a new powerful way to think about scene understanding. As opposed to discrete explicit representations, learned implicit representations can effectively encode both geometry and appearance by representing them as continuous mappings over vector spaces. In this workshop we aim to discuss the benefits and shortcomings of both explicit and implicit approaches for the problem of monocular 3D understanding, as well as ways they could be merged, combining the benefits of each.
This workshop will consider challenges and opportunities at the frontier of 3D reasoning from monocular visual input, the main modality common to most mobile phones, robots, and cars. Our main focus will be on 3D perception of scenes and objects directly from images and videos. Exciting new work is being done in this area, including significant improvements on existing benchmarks thanks to novel self-supervised approaches (e.g. NeRF) that effectively combine the representational capacity of deep learning with geometry. We aim to discuss recent exciting progress on 3D tasks using a single RGB camera, including:
monocular depth estimation
visual odometry
3D detection
shape modeling
scene reconstruction
Program Details (timezone: IDT)
2:00pm - 2:05pm — Introductory remarks
2:05pm - 2:35pm — Georgia Gkioxari
2:35pm - 3:05pm — Vincent Sitzmann
3:05pm - 3:15pm — Coffee Break
3:15pm - 3:45pm — Katerina Fragkiadaki
3:45pm - 4:15pm — Panel: Noha Radwan, Fergal Cotter, Pete Florence
(moderator: Adrien Gaidon, topic "Are explicit 3D representations still needed in robotics and driving")
4:15pm - 4:45pm — Andrew Davison
4:45pm - 4:55pm — Coffee Break
4:55pm - 5:25pm — Jeannette Bohg
5:25pm - 5:55pm — Panel: Angela Dai, Jiajun Wu, Rana Hanocka
(moderator: Greg Shakhnarovich, topic "Meshes x Pointclouds")
5:55pm - 6:00pm — Concluding remarks
Speakers
Panelists
Organizers
The workshop is mainly organized by researchers from Toyota Technological Institute at Chicago (TTIC) & Toyota Research Institute (TRI).
Contact
If you have any questions about the workshop, please contact us at vitor.guizilini@tri.global / rares.ambrus@tri.global / ivas@ttic.edu.