13 Smart Surveillance Video Stream Processing at the Edge for Real‐Time Human Objects Tracking

Seyed Yahya Nikouei Ronghua Xu and Yu Chen

13.1 Introduction

The past decade has witnessed worldwide urbanization because of the benefits and diverse lifestyles in bigger cities. While it brings higher living quality, it introduces new challenges to city administrators, urban planners, and policy makers. Safety and security are among the top concerns when more and more people live in an area with such a high density. Situational awareness (SAW) has been recognized as one of the key capabilities in order to timely deal with urgent issues. To serve this purpose, more and more surveillance cameras and sensors are installed in urban area to monitor the daily activities of the residents. For example, North America alone had more than 62 million cameras by 2016 [1]. The enormous surveillance data generated by these cameras requires extraordinary supervisory action to extract useful information, which implies 24/7 attention to the captured video streams. It is not realistic to rely on human operators facing the ubiquitously deployed cameras. Recent machine‐learning algorithms are promising to make smarter decisions based on surveillance video in real time. However, intelligent decision‐making approaches is not mature yet today.

When each frame is taken, it must be transferred from the field to the data center where further processing. Nowadays, the video data dominate the real‐time ...

Get Fog and Edge Computing now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.