Defeat facial recognition | Hacker News Reader

Defeat facial recognition

cvdazzle.com

79 pointsfomojola19 comments

One problem with this approach is that it would be relatively easy to train classifiers that are looking for people who are attempting to avoid detection.

The site seems to advocate that people use this as a opportunity for self-expression, which would hopefully cause there to be such diverse designs that the best you could do is train classifiers for specific high-profile individual people. Even then, those targeted individuals could just use a new, randomly generated dazzle every day.

Wear highly reflective object like a diamond on the face, preferably on the forehead. It will project light to the camera and create random pattern on the face. What do you think?

A reflective eye-patch would be a nice idea. Are the materials which interfere with cameras but are not too annoying to other people?

Seems like they not only defeat computer facial recognition, but human facial recognition as well.

How else would it work? What would make thing A be able to recognize a face and thing B inherently not? Computers can (and probably will) become as good (or better) as people at recognizing faces.

It might be worthwhile to consider that detecting a face and finding the identity of a face are not the same. From the "Eigenface" article on wikipedia,

"The pictures constituting the training set should have been taken under the same lighting conditions, and must be normalized to have the eyes and mouths aligned across all images."

I would be curious to see how well face identification works with various milder forms of camouflage.

While this can evade detection using full-facial classifiers, it's not difficult to detect individual facial features. If you can detect the eyes, nose, and mouth, and have them in a region in an orientation matching a face, you could assume that it's probably a face. While it is less efficient it's still a possibility.

>If you can detect the eyes, nose, and mouth, and have them in a region in an orientation matching a face, you could assume that it's probably a face.

Or a car.

Wouldn't a surgical mask and sunglasses do the same with less work? They have designer masks and sunglasses.

They would, but you'd be a bit more obvious: walking around with a surgical mask isn't really discreet.

And random black triangles on your face is discreet?

Maybe you could start a fashion trend.

That's what I was thinking and if it does I bet it will start in London due to all the security cameras.

You'd have to assume normal sunglasses are factored into current classifiers.

However, sunglasses coated in dazzle-style patterns...

Most likely, but they partnered with a (what looks like) fashion magazine for the images.

From the article:

- Sunglasses, for example, are a known occlusion which some algorithms account for.

- wearing masks in public can be illegal.

Wearing a centurion nosebridge cover can quickly become a known occlusion too. Dazzle is a short term fix.

this is like captcha with faces though.. good idea.

Reply on news.ycombinator.com