Challenges in Video Analytics

Abstract

Video analytics technology has matured and found application in a variety of fields over the past decade. This chapter discusses the current state-ofthe-art, and describes challenges for future video analytics implementations. Current applications and markets for video analytics are described in the context of a processing pipeline. Application-specific challenges are described with potential solutions to those challenges. This chapter also lists some implementation considerations for embedded video analytics and concludes with future and emerging applications of video analytics. IntroductionVideo analytics is an industry term for the automated extraction of information from video for a variety of purposes. It is a combination of imaging, computer vision, pattern analysis, and machine intelligence applied to real-world problems. Its utility spans several industry segments including video surveillance, retail, and transportation. Video analytics is distinct from machine vision or machine inspection and is similar to automotive vision. Some applications of analytics include the detection of suspicious objects and activities for offering better security, in license plate recognition and traffic analysis for intelligent transportation systems, and in customer counting and queue management for retail applications.The past decade has seen the maturation of algorithms and the adoption of analytics solutions in these markets. Analytics has progressed from research labs, with algorithms running on powerful workstations and PCs to current real-time embedded implementations on consumer-grade embedded processors. At the same time, the range of applications for analytics has also grown, with current trends indicating

An assessment of the video analytics technology gap for transportation facilities

“…Many commercially available tools offer a standard set of video interpretation capabilities and are meant to work out of the box, with little or no tuning required by the end user [2]. Some critical infrastructure facilities have already deployed such tools within limited capacities, especially for perimeter monitoring and archived content retrieval [3].…”

Object recognition and tracking for surveillance applications using deep learning techniques

“…Some examples of such footage are given in Figure 1. 3. As discussed in [6], most video analytics applications are following a pipeline of well established processing steps, depicted in Figure 1. 4.…”

“…Enable video analytics for mobile camera devices producing unstable content. 6. Demonstrate real-world applications of the contributions.…”