chenwei-zhao/captcha-recognizer

What it solves

This project provides a way to automatically identify the gap location in slider-based CAPTCHAs. It solves the problem of finding the exact coordinates and confidence level of the missing piece in a background image, which is typically used to bypass slider verification systems.

How it works

The library uses a deep learning-based general gap detection model to analyze images. It can process either background images alone or full CAPTCHA images that include both the slider piece and the background. It also provides tools to handle common real-world challenges like initial slider offsets and image scaling issues.

Who it’s for

Developers who need to automate the interaction with slider CAPTCHAs or those studying computer vision for gap detection.

Highlights

• General Gap Detection: Uses a deep learning model to identify gaps across different types of slider CAPTCHAs.
• Flexible Input: Supports background images, full images containing both the slider and background, and various data types (strings, paths, bytes, and numpy arrays).
• Onetime Setup: Easy installation via PyPI.
• Offset Handling: Includes a specific method (identify_offset) to determine the initial position of the slider piece to calculate the actual distance needed to move.
• Cross-Platform: Works on Linux, Windows, and MacOS.