Evaluation and fair comparison of human tracking methods with PTZ cameras

Abstract

Evaluation and comparison of methods, repeatability of experiments, and availability of data are the dynamics driving science forward. In computer vision, a database with ground-truth information enables fair comparison and facilitates rapid improvement of methods in a particular topic. Being a high-level discipline, Human-Computer Interaction (HCI) systems rises on numerous computer vision building blocks, including eye-gaze localization, human localization, action recognition, behavior analysis etc. using mostly active systems employing lasers, projectors, infrared scanners, pan-tilt-zoom cameras and other various active sensors. In this research, we focus on fair comparison of human tracking methods with active (PTZ) cameras. Although there are databases on human tracking, no specific database is available for active (pan-tilt-zoom) camera human tracking. This is because active camera experiments are not repeatable, as camera views depend on previous decisions made by the system. Here, we address the above problem of systematical evaluation of active camera tracking methods and present a survey of their performances.