Components

The input video stream source for OMNI Agent can be an RTSP stream, a USB camera, or a video file. OMNI Agent splits the video stream into separate frames and feeds them into the processing pipelines of selected video analytics scenarios.

A video analytics scenario handler consists of a set of interconnected modules that extract specified parameters from the video stream. A typical set of modules includes an object detector, a tracker, a region presence and line-crossing detector, as well as modules that generate output data in the form of processes and events.

Video analytics scenarios that require facial recognition additionally use modules for face attribute detection, biometric template generation, and database search.

The object tracks and/or key points, region presence indicators, face attributes and recognition results, as well as detected actions (HAR), produced by the scenario handlers, are transmitted to external systems through one or more channels in the form of processes and events.