Captcha Solver Automation
Reads text captchas on portals automatically, with image cleanup, OCR and a Death By Captcha API fallback, so automated workflows keep running.
The problem
Many portals, including government websites for tenders, registrations and status checks, put a text captcha in front of every search or form. When a team has to check hundreds of records a day, typing captchas by hand turns a simple data job into hours of clicking, and automated workflows stop at the first captcha.
What I built
A captcha-reading module that plugs into scraping and form-automation workflows:
- Image cleanup – background noise, dots and strike-through lines are removed and the text is straightened.
- Character splitting – the cleaned image is split into single characters.
- Recognition – an OCR model tuned to each portal's captcha style reads every character and scores how sure it is.
- Death By Captcha fallback – when a character is unclear, the image goes to the Death By Captcha API, so the workflow keeps running instead of failing.
- Retry and logging – a wrong answer gets a fresh captcha and another try, and every attempt is logged.
How it works
The solver is written in Python with OpenCV for image cleanup and Tesseract OCR for reading. It runs inside Selenium workflows and falls back to the Death By Captcha API for hard cases.
Try it
Use the live demo above: pick a captcha style, type it yourself, or press Solve automatically to watch each step.
Want something similar?
I build data-collection and form-automation workflows for portals your team is allowed to automate, with the captcha step handled for you.