We have designed and implemented a technique for generating event models
automatically based on qualitative reasoning and a statistical analysis
of video input. Using an existing tracking program which generates labelled
contours for objects in every frame, the view from a fixed camera is partitioned
into semantically relevant regions based on the paths followed by moving
objects. The paths are indexed with temporal information so objects moving
along the same path at different speeds can be distinguished. Using a notion
of proximity based on the speed of the moving objects and qualitative spatial
reasoning techniques, event models describing the behaviour of pairs of
objects can be built, again using statistical methods. The system has been
tested on a traffic domain and learns various event models expressed in
the qualitative calculus which represent human observable events. The system
can then be used in a variety of applications, e.g. to recognise subsequent
selected event occurrences or unusual behaviours. A video demonstrating
some of these ideas can be found here.
Constructing qualitative event models automatically from video input
J. Fernyhough, A. G. Cohn and D. Hogg
Image and Vision Computing, 18 , pp 81-103, (2000).
[BibTeX]
Newer work, also in collaboration with colleagues working in Computer Vision, and funded by an EU Framework 5 project on Cognitive Vision, Cogvis, is described in more detail here.