IDIS has launched its AI in the Box (DV-2116), boosting the power of surveillance systems with the most accurate deep learning analytics yet developed.
One Korean surveillance manufacturer says in independent tests IDIS
Deep Learning Analytics (IDLA) has achieved industry-beating accuracy rates of 97%, a record performance which is further boosted by high speed processing.
The DV-2116 makes deep learning analytics more affordable for small to mid-sized applications, enhancing security and control room efficiency. The plug-and-play IDLA-ready appliance comes embedded with an NVDIA GTX1060 GPU chipset allowing the analysis of up to 16 channels simultaneously. Users benefit from robust and calibration-free object detection and classification (objects such as people, cars, and bicycles); intrusion and loitering detection; powerful. intelligent search functions; and tracking by colour, object and number;
The introduction of AI in the Box makes deep learning analytics now easier to adopt through trouble-free plug-and-play installation via IDIS Solutions Suite video management software (VMS). This allows installation without costly disruption. And the 97% accuracy minimises false alarms, significantly improving detection and monitoring performance.
James Min, Managing Director, IDIS Europe, says this latest innovation has the potential to make surveillance much less labour intensive – and more effective – for a wide range of users. "Our high accuracy analytics can process vast amounts of data, without break, in a way that human operators can’t. This means that high-resolution video streams can be automatically monitored to spot suspicious behaviour or distinguish potential threats from every day activity.”
IDIS’s Deep Learning Engine, which powers the new DV-2116 AI in the Box solution, can recognise potentially significant movements and characteristics of people and vehicles, while ignoring activity that isn’t relevant. The technology can quickly check through hours of video to find specific individuals. It also becomes more accurate over time due to its self-learning characteristics.
“This is very exciting as it means that time critical activities - such as investigating incidents - will become increasingly efficient as our analytics are embedded in operations,” adds James Min.