Neural Geometry and Rendering: Advances and the Common Objects in 3D Challenge @ECCV2022

The workshop is over. The recordings can be found below.

The success of neural geometry and rendering has created enormous interest in 3D reconstruction and its potential in graphics and image understanding. The goal of this full-day workshop is to bring together the neural geometry and graphics communities to discuss recent advances and future challenges, such as reconstructing dynamic events from limited views. We will have invited talks from an incredible line-up of distinguished speakers in this field.

Alongside the workshop, we are hosting a new CO3D Challenge, based on a the Common Objects in 3D dataset (CO3Dv2). The task is to synthesize novel views of common 3D objects in two settings, when many views (100+) of the objects are available and when only a few views (1-9) are available. The Challenge is now accepting submissions until 11:59AM PDT, Oct 20, 2022 16:59PM PDT, Oct 22, 2022.

Program

Speaker

Title

Recordings

David Novotny "Opening and the CO3D Challenge" vimeo, bilibili
Ben Mildenhall "Modeling Light for View Synthesis" vimeo, bilibili
Angjoo Kanazawa "Dynamic View Synthesis: A Reality Check" vimeo, bilibili
Coffee Break
Niloy Mitra "Learning Dense Surface-Surface Maps" vimeo, bilibili
Thomas Müller "Instant Neural Graphics Primitives" vimeo, bilibili
Taewan Ethan Kim CO3D Many-view Winner Presentation vimeo, bilibili
Zhizhuo Zhou CO3D Few-view Winner Presentation vimeo, bilibili
Lunch Break
Lourdes Agapito "Learning Object-Aware 3D Representations" vimeo, bilibili
Christian Rupprecht "(De)-Rendering 3D Objects in the Wild" vimeo, bilibili
Deva Ramanan "Reconstructing Deformable Objects from Monocular Videos" vimeo, bilibili
Coffee Break
Andreas Geiger "Constraining 3D Fields for Reconstruction and View Synthesis"
Yaser Sheikh "Neural Rendering for Photorealistic Telepresence"
Panel Discussion vimeo, bilibili

Speakers

Organizers