DIY Captcha Solver Using Tesseract Engine - Part 1 - Installation, Creating Caller App in C++

Опубликовано: 07 Май 2026
на канале: artbyrobot
1,692
10

In this video, I demonstrate how to install the Tesseract OCR (optical character recognition) engine on Windows and then call it using the console as well as using a calling desktop application as the caller. Tesseract is a deep learning AI engine developed by HP and Google that is able to read text from an image. It has been used to convert books to digital pdfs. I am using it to read captcha images containing text so that my software can bypass captchas! I am able to demonstrate that the engine is capable of solving captchas, however, I will need to train the engine to solve captchas with the fonts that I am up against. I will also need to later add in a series of additional image manipulations using other engines that will clear up the image and clean up all the distortions, background noise, and skewing of the captcha lettering prior to feeding that lettering into Tesseract for identification. I will be demonstrating these additional filtering steps in later videos in this series! The end goal will be to have a desktop application of mine be able to encounter captchas occasionally, copy the region of the screen containing the captcha into memory, saving that memory to a .jpg file, then running that jpg file through a series of filtering engines via a series of commands. After cleaning up and preparing the captcha, making it as legible and clean as possible, it will then feed the image to Tesseract engine which will read it and output the text within it to a .txt file. My program will then read in that txt file, pulling the text from it into a string, and then be able to type that string into the captcha form in order to "prove" it is a human!

Note that there are many services out there where I can pay $0.07/1000 captcha's solved, but I am too cheap to pay that! Plus, this project is a great learning experience for me. I am basically, for the first time, collaborating with 3rd party open source deep learning and neural network software developers by using their tools to aid my software in real time! So basically, without having to code crazy complex systems to solve captchas, I'm able to utilize various tools online to solve them by combining them all into my project. The idea of using 3rd party software, especially deep learning AI software in my projects is pretty powerful. I can always custom code my own AI to replace the 3rd party tools later, however, using their tools will save me time and in some cases, 3rd party tools may do a better job at some tasks than my in house AI. Why reinvent the wheel? For some stuff, I WILL reinvent the wheel to learn more and improve the wheel, but using other people's wheels here and there doesn't hurt either!



http://www.artbyrobot.com

  / artbyrobot  
https://plus.google.com/+artbyrobot1
  / realistic-humanoid-robot  
  / artbyrobot  
  / artbyrobot  
  / www  
  / artbyrobot  

To send me a one-time donation: https://streamtip.com/t/artbyrobot

To send me an automated monthly donation:   / artbyrobot