FFCS-Captcha

The program tries to find out captcha value without using Machine Learning and only Image Processing. The academics website of VIT Chennai has a captcha that prevents bots from logging into the website. But the capctha is simple and can only be broken down by using Image Processing.

In this project, I use OpenCV and PIL to guess the captcha value. The program uses OpenCV's Template Matching to break the captcha.

Requirements:

• Pillow (for image cropping)
• NumPy (for image conversion)
• OpenCV (for Template Matching)
• Matplotlib (for displaying image in Jupyter Notebook)

How to run:

First call the function get_random_captcha and pass the path of image you want to you want to break. By default it gets image from unlabelled data I have saved from the site.

Then call remove_noise and then detect_captcha functions respectively to get the captcha value.

Output:

The program correcty guesses the captcha

Note :You may need to install Jupyter Notebook for running the code.