Data-Driven Computational Abstraction and Imagination of Visual Forms
Visual data in digital form is now extremely commonplace. This is especially true for 2D images and videos, but increasingly also the case for 3D scans or RGBD videos. This wealth of data offers us opportunities for enabling computers to structure visual information in ways that approximate the tasks of abstraction, summarization, and completion of shapes or forms that our brains so naturally perform. The speaker will discuss certain new computational approaches to this goal, based on building networks that interconnect data related to shape or form in multiple modalities (2D images, 3D point clouds, 3D models, etc). Information transport and aggregation in such networks naturally lead to abstractions of objects and other visual entities, allowing data compression while capturing variability as well as shared structure. The speaker will also describe a collaborative effort between researchers at Princeton University, Stanford University and Toyota Technological Institute at Chicago called ShapeNet, as an attempt to build a large-scale repository of 3D models richly annotated with geometric, physical, functional, and semantic information - both individually and in relation to other models. More than a repository, ShapeNet is a true network that allows information transport not only between its nodes but also to new visual data coming from sensors. This effectively enables us to add missing information to signals, giving us for example the ability to imagine what an occluded part of an object in an image may look like, based on the world-knowledge encoded in ShapeNet.
About the speaker
Prof. Leonidas Guibas received his BS and MS from the California Institute of Technology in 1971 and obtained his PhD in Stanford University in 1976. He joined Stanford as Professor of Computer Science in 1984. Prof Guibas is currently the Paul Pigott Professor of Computer Science and Professor (by courtesy) of Electrical Engineering. He heads the Geometric Computation group in the Computer Science Department and is a member of the Computer Graphics and Artificial Intelligence Laboratories.
Prof. Guibas works on algorithms for sensing, modeling, reasoning, rendering, and acting on the physical world. His current foci of interest include geometric modeling with point cloud data, deformations and contacts, organizing and searching libraries of 3D shapes and images, sensor networks for lightweight distributed estimation / reasoning, analysis of GPS traces and other mobility data, and modeling the shape and motion biological macromolecules and other biological structures.
Prof. Guibas is a Fellow of the Association for Computing Machinery (ACM) and a Fellow of the Institute of Electrical and Electronics Engineers (IEEE). He was awarded the ACM/AAAI Allen Newell Award in 2007 for his pioneering contributions in applying algorithms to a wide range of computer science disciplines.