Automated collection of high quality 3D avatar images

James Kim, University of Louisville
Darryl D'Souza, University of Louisville
Roman V. Yampolskiy, University of Louisville

Abstract

CAPTCHAs are security tests designed to allow users to easily identify themselves as humans; however, as research shows (Bursztein et al. 2010) these test aren't necessarily easy for humans to pass. As a result a new test, which requests users to identify images of real humans among those of 3D virtual avatars, is proposed that would create a potentially unsolvable obstacle for computers and a quick, easy verification for humans. In order to provide test cases for this new test, an automated bot is used to collect images of 3D avatars from Evolver.com.