tailieunhanh - A Survey of Human-Sensing: Methods for Detecting Presence, Count, Location, Track, and Identity

An increasingly common requirement of computer systems is to extract information regarding the people present in an environment. In this article "A Survey of Human-Sensing: Methods for Detecting Presence, Count, Location, Track, and Identity "we provide a comprehensive, multi-disciplinary survey of the existing literature, focusing mainly on the extraction of five commonly needed spatiotemporal properties: Namely presence, count, location, track and identity,. | A Survey of Human-Sensing Methods for Detecting Presence Count Location Track and Identity THIAGO TEIXEIRA Yale University GERSHON dUbLON Massachusetts Institute of Technology1 and ANDREAS SAVVIDES Yale University An increasingly common requirement of computer systems is to extract information regarding the people present in an environment. In this article we provide a comprehensive multi-disciplinary survey of the existing literature focusing mainly on the extraction of five commonly needed spatiotemporal properties namely presence count location track and identity. We discuss a new taxonomy of observable human properties and physical traits along with the sensing modalities that can be used to extract them. We compare active vs. passive sensors and single-modality vs. sensor fusion approaches in instrumented vs. uninstrumented settings surveying sensors as diverse as cameras motion sensors pressure pads radars electric field sensors and wearable inertial sensors among others. The goal of this work is to summarize the existing solutions from various disciplines to guide the creation of new systems and point toward future research directions. Categories and Subject Descriptors Artificial Intelligence Robotics Sensors Image Processing and Computer Vision Scene Analysis Sensor Fusion Tracking General Terms Algorithms Design Additional Key Words and Phrases Human counting human detection identification localization people counting person detection sensor fusion tracking 1. INTRODUCTION As the sensor network and ubiquitous computing communities increasingly focus on creating environments that are seamlessly aware of and responsive to the humans that inhabit them the need to sense people will become ever more pressing. Humansensing encompasses issues from the lowest level instantaneous sensing challenges 1Work done while a member of ENALAB at Yale University. Author s address Yale University 10 Hillhouse Ave New Haven CT 06511. Permission to make digital .