Author(s: Daksh Trhan Cybersecurity, Machine Learning and How to Crack CAPTCHA with Machine Learning I am kind of amazed by the technology, sometimes, it hooks me to weird-yet-interesting short videos, other times, it asks me to prove, ‘I’m a human!’ You book Flight Tickets, you face CAPTCHA. You create accounts, you face CAPTCHA. Check for plagiarism in your article by using CAPTCHA once more! Sometimes I just want to shout, “YES!” I’m a robot. (Of course, I’m a robot) Sometimes I wonder who wins all the mountains/bikes/firehydrants/cycles on their first attempt. Captcha: What’s it? What is Captcha? And why do we use it. Is it getting more difficult? CAPTCHA is the Completely Automated Public Turing Test To Tell Humans and Computers Apart. Yahoo! was a young 21stcentury company. When Yahoo! was just starting to blossom, people were concerned that spammers would create fake accounts and write code. To stop spammers, there must be a way to distinguish human users from scripts. It should not be possible for computers to crack the required mechanism, but they need to be capable of grading that test. I told you technology is weird-yet-interesting. Due to the weakness of computers, less exposure and Python, machines were unable to recognize text. On the other side, humans were skilled in text recognition because we only read texts every day. Luis Von Ahn invented CAPTCHA. Computers would receive a randomly generated image of text and its answer. The text would then be warped to make it difficult for computers to comprehend. Marija Zaric, Unsplash Photo. The test helps to distinguish between users and humans. It wasn’t enough to last. Computers soon learned how to read the text in distorted form and became more proficient. These same issues arose because computers became too intelligent to pass the test. Now, with increased traffic, it was necessary for a stronger mechanism. Re-CAPTCHA This was a similar system to CAPTCHA but instead of providing one word, the CAPTCHA now contains two words. The first word is known by computers, but computer can guess the second word. However, the word for the second word was randomly pulled from any book or article. Assumed that humans would know the answer to the first word correctly, it is possible for another word to be correct as well. The second word is usually the one that computers use to verify the validity of the CAPTCHA. However, computers soon outperformed this technique and were able to crack ReCAPTCHA. The method was so successful that only 33% humans could defeat Re-CAPTCHA (v2) according to Google. However, AI beat it with a precision of 99.8%. This time the approach was new, as humans were to be expected to teach machines real-world entities. Photo by Dedy Kurniawan, Unsplash. We are all familiar with the Fire Hydrants and Buses, Cyclists, and Bikes tests. We are trying to show the machine how a real-world entity should look by choosing the right image. Our input is captured and used to help self-learning cars better understand the entities. Guess what, AI is getting better at it! It is also getting more accurate! Humans have given up on trying to make a reliable test with Re-CAPTCHA (v3). Based on user behavior, we now verify identity. Users are not aware that this invisible test is being performed. This invisible test is hidden behind your webpages to find out if you are human or bot. Privacy is not a reality. It can monitor your clicking habits, typing speed and workflow. It then attempts to determine based on this information. You may display unusual behavior such as writing 100s many words per second or clicking very often. You will be asked to confirm your identity by prompting Re-CAPTCHA (v2) Machine Learning cracked CAPTCHA Machine Learning cracked CAPTCHA. You just need the data and a basic OCR model. Github has the training data. The dataset is 1040 pictures. Visualizing the data Model Training our model Predicting output The code is at: Solving CAPTCHA with ML This article should have given an overview of CAPTCHAs. This work is intended to be an educational/fun project. It should not be used in any way that could harm or cause damage. Refer to: *) OCR Model used for CAPTCHA reading. Find me on Web: www.dakshtrehan.com Follow me at LinkedIn: www.linkedin.com/in/dakshtrehan Read my Tech blogs: www.dakshtrehan.medium.com Connect with me at Instagram: www.instagram.com/_daksh_trehan_ Want to learn more? YouTube uses AI to suggest videos Thanks for the Cheers! CAPTCHAs and MACHINES: An Bitter Rivalry This story was first published on Medium by . People are commenting on it and highlighting the details. Published via
We monitors and writes about new technologies in areas such as technology, innovation, digitization, space, Earth, IT and AI.