Smart Surveillance with Context and Location Sensitivity and Quality Control

Abstract

Smart video surveillance systems are essentials in modern environments to ensure safety and security for lives and property. In order to capture critical clues of reconnaissance, it is important to automatically and dynamically control the surveillance sensitivity and quality for detecting abnormal/suspicious events to perform high quality monitoring. In this work, we design a smart video surveillance system, which fully utilizes pan tilt-zoom (PTZ) cameras and well integrates with environment sensors (such as fire/motion/door sensors) while cooperating with wearable devices that can adaptively monitor events and ensure the surveillance quality in terms of pixel-per-foot (PPF), viewing angles, timeliness, and accuracy. In addition, this system also supports to define the special monitoring tasks and surveillance requirements, such as the surveillance timeliness and duty cycle for particular moving objects to be tracked per time unit. Then, a smart camera dispatch algorithm will calculate and determine the best cameras set with the corresponding configurations to accomplish the surveillance tasks effectively and precisely. We will show how our system realizes smart detecting and monitoring in this demonstration by leveraging PTZ cameras, environmental sensors and wearable devices.