Capturing the Captcha
Amit Dhar, Kamalanathan Kandasamy, Sethuraman Srinivas · 2014
Captchas are widely used nowadays by many websites to check whether the user is human or a bot trying to be harmful. Lot of research has been done in the area of captchas. Whether the research is about strengthening the captcha, breaking captchas or optical character recognition, a large and accurate training dataset is needed for better results. Most of the research with captchas has been using either a limited training data set or self generated data set. In this paper, we present a novel method of building a self updating domain specific training dataset for Captchas, that can be used in various areas of research and development. Our technique is not limited to a particular kind of captcha. We prevent the misuse of this training data set using API keys that are issued only to trusted institutions/organizations. We further prevent the misuse by validating the domains during access to all our APIs.