9/16/2011
Written by
A new text in computer vision co-authored by University of Illinois computer science professor Derek Hoiem introduces the subject of 3D object recognition and scene interpretation in depth. The text, Representations and Techniques for 3D Object Recognition & Scene Interpretation has a primary focus on recent efforts to fuse models of geometry and perspective with statistical machine learning.
The text aims to make the latest developments in 3D scene understanding and object recognition accessible to newcomers to the field. With existing research scattered across journal and conference papers, the subject was left to the purview of experts. Hoiem’s text organizes this research and provides an historical and technical background so that newcomers to the field can learn about this emerging area.
“In recent years, the rigid, algebraic view of 3D geometry has given way to a more statistical, probabilistic view. In consequence, we’ve seen amazing new abilities to reconstruct 3D scenes and recognize 3D objects from a photograph,” said Hoiem. “These technologies could have far-ranging impact, from robotics, to vehicle safety, to content creation and photo enhancement. “
In 2011, Hoiem received an NSF CAREER Award for his work in computer vision. His project, “Large-Scale Recognition Using Shared Structures, Flexible Learning, and Efficient Search” aims to enable computers to interpret objects in images. By developing algorithms to recognize parts, materials, pose, and other properties of objects, Hoiem aims to give computers the ability to make predictions about new objects that they encounter.