Motivation

1
Motivation
Technion-Israel Institute of Technology
Hybrid Classification of Moving Targets
Michael Rudzsky Joint work with Ehud Rivlin, Uri
Bogomolov, Stas Lapchev and Gideon Dror
2
Tracking
3
Tracking
4
Classification -people
walk
run
run45
5
Shape is not enough for reliable
classification !
Solution- hybrid of appearance-based and
motion-based methods !
6
Main Goals

• Detection of moving objects
• Tracking of detected targets
• Classification to one of predefined classes
• human,
• human group,
• animal,
• vehicle

E.Rivlin, M.Rudzsky, R.Goldenberg, U.Bogolmolov
and S.Lapchev.,ICPR'02 Y.Bogomolov, G.Dror,
S.Lapchev, E.Rivlin, M.Rudzsky. BMVC03
7
Classification
Flow chart
8
Main Goals

• Detection of moving objects
• Tracking of detected targets
• Classification to one of predefined classes
• human,
• human group,
• animal,
• vehicle

E.Rivlin, M.Rudzsky, R.Goldenberg, U.Bogolmolov
and S.Lepchev.,ICPR'02 Y.Bogomolov, G.Dror,
S.Lapchev, E.Rivlin, M.Rudzsky. BMVC03
9
The classes handled by the system
10
Motion Based Feature Extraction
Fitted ellipse
Star skeleton
11
Combined classifier for all classes
12
Targets Tracking Classification
13
Segmentation Tracking in Difficult
Conditions