Scene understanding includes: state recognition, event
detection and recognition, and situation interpretation.
We focused on the event detection and recognition which
bridges the gap between the raw data and the high
level description of identities and activities. However,
many pre-requisites are required before addressing the
”video understanding”. The research is divided into
three phases:
In video analysis, the research is focused on tracking
multiple objects considering the real-time challenges.
Many techniques have been proposed to address these
crucial issues but still no general solution exists. For
that, we have proposed two novel approaches to han-
dle the multi-object tracking under confusion based on
Correlation-Weighted Histogram Intersection (CWHI)
and the second approach multiple features of moving
object are fused along with CWHI. Further, we are
interested in integrating the inferential knowledge in-
terpretation framework to recognize the state of mov-
ing object under inter-object occlusion and separation.
After tracking, classification of detected objects will
be carried out to obtain the individual objects in the
scene. These individual objects are then interpreted
separately to infer their activities. (S. Pathan -064,
A. Al-Hamadi -18709)