| |
Abstract:
The classical model of visual processing in cortex is a
hierarchy of increasingly sophisticated representations, extending
in a natural way the model of simple to complex cells of Hubel and
Wiesel. Somewhat surprisingly, little quantitative modeling has
been done in the last 15 years to explore the biological
feasibility of this class of models to explain higher level visual
processing, such as object recognition. I will review experimental
results in viewpoint-invariant object recognition and describe here
a new hierarchical model --- developed with Max Riesenhuber ---
that accounts well for this complex visual task, is consistent with
several recent physiological experiments in inferotemporal cortex
and makes testable predictions. A key element of the model is a
MAX-like response function of some neurons where the strongest
afferent determines the unit's output. The MAX operation was
suggested by trying to find the computational equivalent in cortex
of a scanning operation which is a key module in a family of
successful computer vision algorithms that we have developed during
the last few years. In particular, I will briefly describe a
trainable object detection system that automatically learns to
detect objects of a certain class, such as faces, cars and people,
in unconstrained scenes.
|
