Wait, what?
Patterns in random images
From more than a million images, only 15 were picked.
Usually when we think about computers, algorithms, maths, we assume that those are flawless. Because how could a computer make a mistake in adding two numbers? That can only be because of human error. Although when talking about new age programming, those cannot exist without machine learning and that is a whole other topic. During this project, I made an experiment aimed at that particular branch of science
I programmed a software that automatically generated random images of a given size and tested it in search of face detection. As the feature detection algorithms are relatively slow, it took a couple of days to generate and check more than a million images. For the same reason face detection was chosen as it's one of the fastest models available.
Machine learning is known for being not 100% accurate. Was this a mistake or is there something that a human eye cannot see?
What's more to this?
After purchasing Digital Ghost NFT you will get additionally
Source code
Url for a Colab notebook so that you could generate your own Digital Ghost. That can take somewhere between a couple of minutes and hours
Original image
Detector can only find true Digital Ghosts on original images. Urls for those will be given for NFT owners. Additionally, detector source code in form on Colab notebook will be given.
Detector API
Detector
Curious about searching patterns around you?
Check out our detector API. You can try to find your own hidden ghost but be aware that true Digital Ghosts can be found only on original images and no others.