2013 Poster Sessions : On Being the Right Scale: Sizing Large Collections of 3D Models

Student Name : Manolis Savva
Advisor : Patrick Hanrahan
Research Areas: Graphics/HCI
Abstract:
Recently, 3D models are becoming increasingly available in large, publicly available online repositories. Unfortunately, many 3D models are defined using scales with unknown correspondence to physical units. This makes it hard to reuse and combine 3D models to create larger scenes. We address this problem of recovering physical sizes for collections of 3D models. Our algorithmic approach provides absolute size estimates for 3D models by combining category-based size priors and size observations from existing 3D scenes. Our approach handles unobserved 3D models without any user intervention. It also scales to large public 3D model databases and is appropriate for handling the open-world problem of rapidly expanding collections of 3D models. We use two datasets extracted from online 3D model repositories and evaluate our results against both human judgments of size and ground truth data for physical sizes of 3D models.

Bio:

Manolis Savva
is a third year Computer Science Ph.D. student at Stanford University advised by Dr. Pat Hanrahan. His research interests lie at the intersection of computer graphics, data visualization and human computer interaction. Prior to Stanford, he received a BA in Physics and Computer Science from Cornell University.