Deep Fake: Catch Me if You Can

deep fake

deep fake


It all started so beautifully innocent: Let’s train a network in generating fake images, while another network will recognize whether the image/video is genuine or a fake one and will tell the first one why its reasoning ==> GANs.

These days, Google and MS and building heavily weighted cannons to recognize such “fake” things. The mouse and a cat game went to the frontier, where the scientists said that the only way to identify the fake video is by looking for signs of blood pulse rhythm in the small blood vessels on the face. If there are such recognized, then the video is “for now, almost positively,” not a fake one.

However, just to be aligned with the modern fakers:

Deepfake videos of Tom Cruise show the technology’s threat to society is very real (inputmag.com)

Have a nice day!

semantic search
Are you a robot, m?

Deep Learning

deep fakeDeep Learning

Leave a Reply

Your email address will not be published. Required fields are marked *

18 + five =