TITLE: A Region-based Approach to Scene Understanding
PRESENTER: Stephen Gould, http://www.stanford.edu/~sgould/, firstname.lastname@example.org
AFFILIATION:Electrical Engineering, Stanford University, http://www.stanford.edu/
DATE: Monday 3rd August 2009
PLACE: CSE Seminar Room, Level 1, K17
Object detection and multi-class image segmentation are two closely
related tasks that can be greatly improved when solved jointly by
feeding information from one task to the other. However, current
state-of-the-art models use a separate representation for each task
making joint inference clumsy and leaving the classification of many
parts of the scene ambiguous.
In this work, we propose a hierarchical region-based approach to
joint object detection and image segmentation. Our approach
simultaneously reasons about pixels, regions, and objects in a
coherent probabilistic model. Pixel appearance features allow us to
perform well on classifying amorphous background classes, while the
explicit representation of regions facilitate the computation of
more sophisticated features necessary for object detection.
Importantly, our model gives a single unified description of the
scene. We explain every pixel in the image and enforce global
consistency between all random variables in our model.
We show, experimentally, that our approach competes with the
state-of-the-art on a number vision tasks including multi-class image
segmentation and geometric reasoning, and significantly improves
on state-of-the-art methods for object detection.
BIOGRAPHY OF SPEAKER:
Stephen Gould is a PhD student at Stanford University studying machine
learning and artificial intelligence. His primary focus is the use of
probabilistic models for computer vision and, in particular, scene
understanding. His PhD advisor is Daphne Koller. He also works with
Previously, Stephen Gould co-founded Sensory Networks where he was
responsible for hardware engineering and product development as VP
Engineering. Before Sensory Networks, he worked in other senior
(and junior) engineering positions in research and development companies
(including Dilithium Networks, Polartechnics, CWC). Stephen received MSEE
degree from Stanford in 1998, and BE and BSc degrees from The University
of Sydney, Australia in 1996 and 1994. He enjoy mountain bike riding,
playing golf, some travelling, and solving puzzles.
Van Hai Ho