|
|
KOBUS BARNARD
UNCERTAINTY QUANTIFICATION GROUP
|
My students and I work on statistical approaches to understanding images both
in the domain of pure computer vision and applied to scientific problems.
One example is the fitting structural
models of biological organisms to data. For example, in project we have
developed a stochastic
L-system model for
microscopic fungus from the genus Alternaria,
and the corresponding
inference process to fit the model to stacks of brightfield images (paper).
Being able to fit such models enables quantification of form which is critical
for high throughput experiments targeted at understanding how form relates to
organism function and its molecular biology.
Fungus across the genus vary widely
structure. By parameterizing possible species level characteristics, we can
learn structure models for different species. This then can be used, for
example, to classify species based on morphology.
I am with the
Computer
Science department at the
University of
Arizona.
Additional affiliations include Electrical and Computer Engineering
(courtesy appointment),
the Cognitive
Sciences GIDP,
the Statistics GIDP,
and BIO5.