Home // COGNITIVE 2018, The Tenth International Conference on Advanced Cognitive Technologies and Applications // View article

Human and Machine Capabilities for Place Recognition: a Comparison Study

Authors:
Carlos Alberto Martínez-Miwa
Mario Castelán
Luz Abril Torres-Méndez
Alejandro Maldonado-Ramírez

Keywords: Place recognition; Human-machine capabilities; Computer Vision; Feature detection

Abstract:
This paper is aimed at exploring the way humans perceive previously visited places under challenging circumstances while driving a car. The term challenging refers to scenarios that do not contain specially rich visual information. We developed a series of experiments to investigate the performance of humans and computer vision algorithms, in order to recall locations in video sequences that were gathered with a camera mounted on a car. Our experiments reveal that the state of the art in machine vision fails against humans for identifying places where subtle changes occur, for instance, when illumination varies depending on the daytime. However, machines do present greater capabilities than humans when the reference image appears within a video sequence that does not exhibit spatial and daytime variations.

Pages: 72 to 77

Copyright: Copyright (c) IARIA, 2018

Publication date: February 18, 2018

Published in: conference

ISSN: 2308-4197

ISBN: 978-1-61208-609-5

Location: Barcelona, Spain

Dates: from February 18, 2018 to February 22, 2018