Traditional security cameras do a good job of reacting whenever someone steps into their field of view. But what exactly is that person captured on camera doing? And where are they going once they leave the view of the camera? To help companies gain a deeper sense of the whereabouts of employees and provide a better, more intelligent form of physical security, CrowdOptic has developed a software platform that that utilizes artificial intelligence in the video analytics process to propel its own patented triangulation algorithms. The system, developed in collaboration with Hewlett Packard Enterprise and Intel Corporation, will be used initially to provide indoor location, gaze analytics, and subject tracking.
Triangulation Algorithms Provide Better Surveillance
Its hybrid architecture enables CrowdOptic Intersect to run securely at the edge to support retail use cases and as a component of a larger centralized system to support public safety use cases, with flexibility to support virtually any environment in between. CrowdOptic Intersect works with pre-trained AI models from some of the world’s leading research institutions and has the ability to train custom models, depending on the specific application. Analysis of the video incorporates CrowdOptic’s patented triangulation algorithms without a dependency on device sensor data, opening the doors to a wide range new use cases.
An example of how a basic AI model works with CrowdOptic algorithms can be viewed at: https://crowdoptic.com/files/ai-demo/.
A typical configuration of CrowdOptic Intersect runs on an HPE EL300 and utilizes the Intel OpenVINO SDK to spread the compute load across the EL300’s in-built GPU and an HPE neural compute accelerator card with four Intel Movidius Myriad X Vision Processing Units on board.
Enterprises, retail organizations or public agencies interested in deploying this system within existing camera infrastructures should contact CrowdOptic at email@example.com. Demo kits may be purchased online at: https://intersect.crowdoptic.com/.