-A +A
In surveillance systems, automatic analysis of audio-visual data is a crucial requirement. The main aim of AGS Project is to apply our concept recognition knowledge-base and software framework to surveillance systems.

Concept recognition can be described as the process of teaching audio-visual concepts to computers and automatic detection of the concepts.

At first stage, concept models are learned by means of training samples. At detection stage, samples are tested against the learned models. Models can be adapted in order to cope with new instances. Furthermore, the sensitivity of the system can be changed, so that false-alarm vs. miss trade-off can be adjusted according to the application needs.

In AGS Project, knowledge-base and software framework formed during KAVTAN Project is applied to security related concepts and a prototype system is developed.


The protype system has three main functionalities:
•    Real-time video monitoring
•    Concept recognition
•    Archiving

Concept List

•    Fire
•    Explosion Sound
•    Crowd
•    Human Existence
•    Bullet Sound
•    Crying
•    Flag
•    Fight
•    Traffic Accident
•    Abundant Object
•    Stolen Object
•    Border Intrusion
•    Flaming Object