Bypassing reCAPTCHAV2 from Google Using Supply Chain of Neural Networks and Machine Learning

Abstract

Until recently, the search engine did not understand what exactly was depicted in the photo, which it gave out in the results, but only focused on the words that were found in the text next to this image or that were written in its alt or title attributes (img tag). Modern algorithms of search engines allow finding not only text files, and files marked with text tags, but also similar images. For these purposes, there are several services on the supply chain of network. Search by the user of the Internet of all images from one series or search for an analogue of an object on a photo by driving words into the search string will not be successful. Currently, search on the sample image (photos or any other image) is supported by both search engines leading in Russia - Google and Yandex. But as an alternative, there are services that structure the Internet for easy retrieval of information that the user needs. And at the moment there are social bookmarks, catalogs, torrent trackers, forums, specialized search engines, file sharing. The authors implemented a program that will take a link to the image at the input, and in return give a response from the API in json format.