Keywords
motion, 3D shape, non-rigid shape, structure-from-motion
Abstract
Our world is full of objects that deform over time, for example animals, trees and clouds. Yet, the human visual system seems to readily disentangle object motions from non-rigid deformations, in order to categorize objects, recognize the nature of actions such as running or jumping, and even to infer intentions. A large body of experimental work has been devoted to extracting rigid structure from motion, but there is little experimental work on the perception of non-rigid 3-D shapes from motion (e.g. Jain, 2011). Similarly, until recently, almost all formal work had concentrated on the rigid case. In the last fifteen years, however, Computer Vision researchers have made significant advances in non-rigid-structure-from-motion. In this talk we will present the history of these advances, while examining the validity of the assumptions, and the performance of the models in the light of what we know about human vision. We will discuss how these models can be modified for human vision, particularly by testing assumptions in psychophysical and physiology experiments. We hope that this will spark interest in this new and exciting topic of research, and create cross talk between Computer Scientists and Vision Researchers.
Start Date
13-5-2015 2:50 PM
End Date
13-5-2015 3:15 PM
Session Number
02
Session Title
Shape and Form
Formal Aspects of Non-Rigid-Shape-from-Motion Perception
Our world is full of objects that deform over time, for example animals, trees and clouds. Yet, the human visual system seems to readily disentangle object motions from non-rigid deformations, in order to categorize objects, recognize the nature of actions such as running or jumping, and even to infer intentions. A large body of experimental work has been devoted to extracting rigid structure from motion, but there is little experimental work on the perception of non-rigid 3-D shapes from motion (e.g. Jain, 2011). Similarly, until recently, almost all formal work had concentrated on the rigid case. In the last fifteen years, however, Computer Vision researchers have made significant advances in non-rigid-structure-from-motion. In this talk we will present the history of these advances, while examining the validity of the assumptions, and the performance of the models in the light of what we know about human vision. We will discuss how these models can be modified for human vision, particularly by testing assumptions in psychophysical and physiology experiments. We hope that this will spark interest in this new and exciting topic of research, and create cross talk between Computer Scientists and Vision Researchers.