Behavioral Recognition Systems: Recognizing suspicious scenes.

Pattern recognition for video cameras: Systems for behaviour recognition - Comment on 2013 March 16

Home | Comments | Creation | Redemption Period | Miscellaneous
Home > Comments > 2013 > Comment on 2013 March 16
2008 | 2009 | 2010 | 2011 | 2012  |  2013 |
End: Go to the end of this webpage.

Intelligent surveillance cameras. The video scouts learn to recognize conspicuous behaviour. Read more:

Here something about the progress of surveillance cameras:


Software for video cameras

How surveillance is to become intelligent

Worldwide millions of cameras watch the behaviour of citizens. Because men can no longer cope with all the picture material, software is now to recognize conspicuous scenes. That is the dream of security authorities – and a nightmare for data protectors.

The US firm BRS Labs is an instant success which hardly anybody knows and it would be best if nobody would know it. Within one year the turnover has increased ten times to 200 million dollars. New offices were opened in Houston, São Paulo and London. In San Francisco the technique of the firm is already employed and exported as far as to Dubai. The reason for the success of BRS – an abbreviation for Behavioral Recognition Systems – is surveillance software called AiSight.

At present it revolutionizes worldwide the surveillance at airports and railway stations, because AiSight makes cameras clever: Equipped with the program the video lookouts learn to recognize conspicuous behaviour. No matter whether someone puts down a suitcase where he should not, gets out of a car where he is not allowed to or climbs over fences – the computer raises the alarm. Man is no longer necessary to assess activities of other people recorded per camera – a dream for security authorities.

The calculation behind it is simple. Worldwide there are 50 million surveillance cameras according estimates by to BRS Lab. Just in Britain they record every citizen 300 times a day. But the pictures end in nothing most of the time. There are simply not enough personnel to attentively look at and assess the video material, whether live or not. On the other hand a machine is never distracted, always attentive, never needs a coffee break. The idea of an intelligent surveillance camera hits a gigantic gap in the market.

European security experts would like to take part in this business. The German research ministry (BMBF) alone supports nine projects for pattern recognition with 21 million euro. For three years German researchers try to teach computers, how one reads out of faces, recognizes weapons or registers suspicious movements.

Here an algorithm is to track down dangerous situations in underground railway stations, for example a lying down person, a crowd of people, aggression or generally conspicuous behaviour.

AiSight is built up self-learning, similar to a neuronal net it can gather experiences and learn what in which situation is regarded as normal and what not – for example which door passengers are not allowed to enter.

"Where do the criteria for pattern recognition come from? When they come from experts then there is the danger that a certain world-view is programmed here into the system." On the other hand if the criteria are determined statistically, the statistic normal behaviour, therefore the most frequent behaviour, could become the wished for behaviour – the behaviour the machine wishes for.


Back to: 2013 March 16
Top: Go to the top of this webpage.
2008 | 2009 | 2010 | 2011 | 2012  |  2013 |
Home | Comments | Creation | Redemption Period | Miscellaneous
Site Map: For an overview of this website and for access to the individual webpages.
The web address of this webpage is: