McFRETA: Multi-camera real-time Feature Recognition, Extraction&Tagging Automation

ABSTRACT: Intelligent Automation, Inc. (IAI) proposes an open and scalable framework for Multi-camera real-time Feature Recognition, Extraction & Tagging Automation (McFRETA). Our approach has the following key innovations: (1) A real-time moving targets detection and tracking based on planar motion parallax analysis and scene contextual information. (2) Design of an open metadata schema, based on Web Ontology Language (OWL), (3) Develop an open processing architecture that support correlation and fusion of multiple video streams and support rule-based inference and alert notification, and (4) a geo-browser visualization tool for displaying extracted metadata with corresponding video, as well as support searching and querying of visual features and metadata. BENEFIT: The proposed framework will enable real-time automated feature extraction and parsing of multiple video streams, and the generation of metadata representing visual features and detected entities. The framework will assist analysts to examine large amount of video imagery, and provide capabilities for querying and finding objects in the scene.