description | none |
repository URL | https://github.com/ECHibiki/StreetSigns-et-al.-reCaptcha-solver.git |
owner | qamirror@cock.li |
last change | Fri, 21 Sep 2018 03:05:42 +0000 (20 23:05 -0400) |
last refresh | Sat, 27 Apr 2024 05:50:06 +0000 (27 07:50 +0200) |
mirror URL | git://repo.or.cz/StreetSigns-et-al.-reCaptcha-solver.git |
https://repo.or.cz/StreetSigns-et-al.-reCaptcha-solver.git | |
ssh://git@repo.or.cz/StreetSigns-et-al.-reCaptcha-solver.git | |
bundle info | StreetSigns-et-al.-reCaptcha-solver.git downloadable bundles |
content tags |
A machine learning exercise to solve certain road based images on the no javascripts recaptcha. Contains a data aquisition folder, data processing and a data usage folder. Done in Python, PHP and Typescript.
Method uses a personally built 3 layer CNN to solve no-javascript captchas with a 1 in 10 success rate. Not perfect, but given enough IPs a person could get around this.
Google has since taken messures to make my method less useful and most websites don't enable javascript captcha.
Data usage is hidden anyways so random people can't rework an easy captcha solver.
Documentation files for server component in php, client in ts and learning in py are outdated.
5 years ago | master | logtree |