Cod Captcha

We made everyone's favourite part of the internet better. We spend a lot of time solving CAPTCHAs, what's a little more time? With Cod Captcha, using the internet is now HARDER than it needs to be!

Comment 1

banner

Inspiration

Captcha. No like seriously have you tried solving one, they're getting ridiculously difficult. Am I a robot?

We hate captcha, we hate life. Let's make it even harder to browse the internet to promote going outside. Evidently, we hate writing rhymes too.

What it does

  1. Every 10 seconds we get the aggregate of your network logs
  2. For every megabyte of data sent/received, you will have to solve a CAPTCHA to prove you are a human
  3. A pop up takes over your entire device and hides all other running applications (this forces you to solve the CAPTCHA, we can't have robots take over the world)
  4. Solve a series of CAPTCHAs
  5. Selecting images: Inspired by the CAPTCHAs found online, but these images are all generated with Stable Diffusion.
  6. Pupil verification: We honestly thought it would be really funny if people stared at their webcams with a 2-3cm distance. Can confirm, it was very funny.
  7. Scissors Paper Stone: Use your camera to play a game of scissors paper stone to verify you're a human! I mean, obviously only humans can win at that game. We're honestly surprised how accurate this hand recognition model we trained in less than an hour is.
  8. Where's Waldo: Select the square where Waldo is in. It's a fun game :)
  9. Failing 2 CAPTCHAs in a single session will reveal a special surprise 🎶

Make solving captcha and proving you are a human even harder than it needs to be.

For every megabyte queried on your system, the user would have to solve 1 captcha.

Good luck playing games or watching YouTube videos :)

How we built it

  • Stable Diffusion - Captcha Image Dataset generation & Waldo
  • Swift - MacOS Target device, UI takes over control until user completes captcha.
  • Python (OpenCV) - Iris / Pupil Eye verification captcha
  • Bash - Netstat (logging of network received / sent) - determines no. of captcha.

Challenges we ran into

  • Sleep deprivation: Running on 0 to 2 hours of sleep! Yippie! I have absolutely no idea what I am typing here! (and I have no idea what I'm proofreading)
  • The cold: Feels like we're coding in a freezer, we gathered round the warmth of stable diffusion to keep warm through the harsh dining hall environment.
  • Integration hell: "It works on my device", but it didn't work on mine :(.

Accomplishments that we're proud of

  • Making captcha harder and more annoying
  • Making a borderline malware

What we learned

  • Importance of documentation
  • OpenCV Iris / Pupil detection
  • Creating CAPTCHAs
  • Image manipulation
  • Photoshop Generative Fill (we quite literally asked it for an image of Waldo and then generative expand-ed the environment)
  • Training a machine learning model based on a pre-trained model with metal acceleration
  • Utilizing Generative AI for dataset generation (could be the future of captcha)

What's next for Cod Captcha

  • Cod Captcha was built with noble intentions, to encourage everyone to take a step back from the internet and explore the world around them.
  • A lot of what has been done with technology has been about dismantling roadblocks and making it more accessible. We feel otherwise. We believe in a world full of roadblocks and challenges, otherwise where's the fun in all of that.
  • Also this new gen ai thing has been pretty scary, with our innovative CAPTCHAs, we can stay ahead of whatever Sam Altman is planning to do next. We even have eyeball scanning. We're coming for you Sam.
  • Hey look there's this Dead Internet Theory thing! It has a Wikipedia article, it's a real problem.

Built With

Share this project:

Updates