The future of home security will see threats detected and responded to before a crime can be committed. Making use of big data, home security systems are now able to analyse camera streams and proactively alert armed response service providers. This is a massive stride in improving the means we use to secure our neighbourhoods, believes co-founder of Syntech, Ryan Martyn.
“Research indicates that South Africa has among the most security cameras per capita in the world. That creates a great platform to offer a solution on legacy infrastructure. This new approach to security actually prevents incidents from happening.”
According to Deep Data CEO Jasper Horrell, “There has been a shift in terms of what we can do with data gathering and processing. It’s now become possible to store large quantities of data gathered from different sensors, and also to process it in intelligent ways.”
The latest in home security technology uses big data analysis to revolutionise how threats are assessed. This technology is provided in a partnership between Syntech, which provides Giada commercial computer systems as hardware, and Deep Data, which develops the Deep Alert software.
The Deep Alert system makes use of analytics to identify and analyse threats based on security camera footage, it also sends alerts to homeowners or security service providers based on the type of threats that it identifies.
A proactive approach
Traditional security systems rely on cameras to record footage which, in the event of a security breach, identifies suspects and provides evidence. With this model, a perimeter breach triggers the alarm and an armed response service provider attends the scene – an entirely reactive approach. Future systems making use of big data fundamentally challenge traditional security systems by proactively identifying threats and allowing security monitoring services to respond to alerts before an incident takes place.
The software attaches to existing camera streams and analyses motion triggers, tracks the trigger and then classifies the type of object – such as a dog, person walking down the street, or somebody climbing over the wall. This software runs on a Giada computer – small, lightweight and low energy commercial grade computers – which can monitor up to 12 concurrent camera streams. Potential threats are instantaneously sent to a cloud service, where most of the intensive analytics run. From here, an alert will be sent should a valid threat be detected.
“It’s all very well installing alarms, beams and fences, but every false alarm plants a seed of doubt in terms of how reliable that information is. Over time, repeated false alarms create vulnerability because you stop reacting to them. Big data adds a layer which helps you to trust the system much more,” says Martyn.
The future of fighting crime
South Africa has proven an ideal testing ground for new security features and the big data systems have already been successfully rolled out among neighbourhood watches and residential communities.
Key to the effectiveness of this technology is how it assists armed response services: They receive more accurate information which allows assessment of threats before dispatching a patrol vehicle. This ensures resources are diverted to critical incidents and reduces the number of false alarms, while saving money for the armed response service provider.
The potential for using similar systems is extensive, and could be adapted for businesses to monitor deliveries or operations. Networks of cameras in the same neighbourhood would allow for patterns to be monitored – the difference between a car that frequents an area during a daily commute and one circling an area to carry out a crime would be immediately apparent.
“There will be analytics on analytics. Instead of intelligently telling you what’s there, the technology will be detecting interesting patterns or threats between a range of cameras or classes of objects,” explains Horrell.
For more information about Deep Alert visit https://www.deepdata.works/
© Technews Publishing (Pty) Ltd. | All Rights Reserved.
printer friendly version