Multispectral Desert Fauna Surveillance and Recognition System

We propose the creation and deployment of an automated machine vision based sensor, tracking, and identification system for monitoring wildlife in support of managing natural resources. However, far more is needed beyond basic tracking. Our proposed solution is twofold – software and hardware. The software solution is to develop an architecture consisting of • Multiple networked cameras (with normal optical, IR, UV, and other desired sensor types), • Software to enable the real time storage and extraction of the video for analysis and later review, • A Front-end user interface module that allows the surveillance operator to define contextual areas of interest, • A Tracking module that identifies all moving human/animal shaped objects for future evaluation, • A Vigilance module that triggers alerts based on the type of animal and/or contextual areas of interest that were observed in the tracked objects in the video, and • A display (visual and audio) that shows the tracking of objects and the alerts. The second part of our solution is the hardware that will be deployed for the capturing of the video/IR/etc. data BENEFIT: This project closely fits with Cybernet’s existing vision-based gesture tracking line, and the technology could potentially be incorporated into a number of our products. This will potentially support a fast track approach to moving the proposed development work into full scale DoD deployment more quickly. We have identified a number of potential markets for this technology. Within the U.S. Government, there is the Army sponsor, the Department of Homeland security (monitor border crossings for suspicious behavior), and NASA (monitoring the health and well-being of astronauts on extended missions without invasive sensors). In the commercial market, this system could be employed along with security cameras in stores to detect potential thieves. We will investigate these and other commercial potentials during the Phase-I effort. Also for commercialization, this effort will support the U.S. RDECOM Communications & Electronic Research, Development, and Engineering Center (CERDEC) Night Vision & Electronic Sensors Directorate (NVESD) efforts. Specifically, this work will directly fit in with their technology to understanding and discrimination of intent and possible hostile status by signature, behavior, physiology, contact, and cultural factors, with no constraints on scale, waveband, environment, number of people in a group, or any limitations on possible scenarios.