On: U.S. Army Camouflage Improvement Explained (2013)

[fho]

Question (from somebody who only skimmed the article): how much variance exists of that fabric? Is that pattern repeated after x meters if fabric?

If so: wouldn't knowing what to look for greatly increase the ability to automatically extract the pattern from image data? (Like how it is possible to extract signals from below the noise floor in LoRa communication?)

[fho → fho]

Or ... just detect this specific patch that seems to on every variation of the pattern: https://www.hyperstealth.com/camo-improvement/cadpat-us-vari... :-)

[yencabulator → fho]

Wanna bet such a thing is there purely for copyright reasons?

[ncmncm → fho]

My thought too: a cheap, simple NN attached to the raw image in scopes and sights, programmed to highlight anything it recognizes.

Now I imagine uniform fabric with adversarial patterns added, for those.

[fho → ncmncm]

Doesn't even have to be any fancy ML stuff. Just convolute the input images with the camo pattern.