Over 300 New 'Nazca Lines' Geoglyphs Have Been Revealed by AI
thedebrief.org
thedebrief.org
Thanks for the share.
It does make me wonder if these models suffer from pareidolia. If so, does that say anything about how similar they are to human intelligence? Or are they basically "inheriting" humanity's evolutionary biases as a function of their having been trained on human-generated content?
1. Collected data by flying aircraft over the area. Used a land classification mask to restrict the are to ~ 600 sq km
2. Make image patches of 11m by 11m. I believe there is some overlap in the patches. Sharpen the images for contrast.
3. The training data comes from previously known glyphs. Positive label patches are ones with a glyph. Negative label patches are randomly sampled from the vicinity of the glyph.
4. It looks like they fine tuned resnet 50 with these labels
5. Ran inference on other patches. They had false positives
6. Manually verified these AI predicted glyphs by ground surveys
I couldn't figure out how they drew the outlines in the pictures. I guess it was manually done
They may have tried and accidentally destroyed all the other test planets with their death Star.
My main pet peeve is that AI is a blurry term almost to the point it has no definitions. It's basically a marketing term. AI has been applied, historically, to many trendy and novel facets of computer science, until they are not novel anymore and then they stop being AI. Expert systems, voice generation, image processing, genetic algorithms...
My take on the article: this is cool. I don't think I would call it AI though, if I could avoid it.