Event Location: NSH 1305
Bio: Saad Ali is currently a PhD candidate at the University of Central
Florida, advised by Prof. Mubarak Shah. His research interests include
surveillance in crowded and aerial scenes, action recognition, object
recognition and dynamical systems. He is a student member of IEEE.

Abstract: Automatic localization, tracking, and event detection in videos of
crowded environments is an important visual surveillance problem.
Despite the sophistication of current surveillance systems, they have
not yet attained the desirable level of applicability and robustness
required for handling crowded scenes like parades, concerts, football
matches, train stations, airports, city centers, malls etc.

In this talk, I will first present a framework for segmenting scenes
into dynamically distinct crowd regions using Lagrangian particle
dynamics. For this purpose, the spatial extent of the video is treated
as a phase space of a non-autonomous dynamical system where transport
from one region of the phase space to the other is controlled by the
optical flow. A grid of particles is advected through the phase space
using the optical flow using a numerical integration scheme, and the
amount by which neighboring particles diverge is quantified by using a
Cauchy-Green deformation tensor. The maximum eigenvalue of this tensor
is used to construct a Finite Time Lyapunov Exponent (FTLE) field,
which reveals the time-dependent invariant manifolds of the
non-autonomous dynamical system which are called Lagrangian Coherent
Structures (LCS). The LCS in turn divides the crowd flow into regions
of different dynamics, and therefore are used to the segment the scene
into distinct crowd regions. This segmentation is then used to detect
any change in the behavior of the crowd over time. Next, I will
present an algorithm for tracking individual targets in high density
(hundreds of people) crowded scenes. The novelty of the algorithm lies
in a scene structure based force model, which is used in conjunction
with the available appearance information for tracking individuals in
a complex crowded scene. The key ingredients of the scene structure
force model are three fields namely, `Static Floor Field’ (SFF),
`Dynamic Floor Field’ (DFF), and `Boundary Floor Field’ (BFF). These
fields determine the probability of a person moving from one location
to another in a way that the object movement is more likely in the
direction of higher fields.