Four New Nazca Lines Identified by Artificial Intelligence in Peru

Using a form of artificial intelligence, researchers have discovered four new Nazca geoglyphs in Peru, according to a new study recently published in the Journal of Archaeological Science.

The Nazca lines are UNESCO-protected geoglyphs dating from between 100 BCE and 300 CE that depict humans, camelids, birds, orcas, felines, and snakes. There are a range of theories surrounding their purpose, including that the Nazca geoglyphs depict deities, are a form of irrigation, or are a calendar with astrological alignments, though it remains unclear.

Listen beautiful relax classics on our Youtube channel.

Discovered about 100 years ago with the advancement of aerial technology, the geoglyphs are nearly impossible to decipher from the ground.

Researchers at Yamagata University in Japan developed and trained the AI-driven deep learning technology to spot the lines from satellite images.

DL, as the researchers explain, “is a particular kind of machine learning that achieves great power and flexibility in pattern analysis of images, speech, language, and more. In archaeology, DL is used in the analysis of the iconography, text, and writing of excavated objects.”

While the variability in design and limited subset posed a challenge with training the AI program, researchers were able to develop a new approach to detecting geoglyphs based on similarities to other known designs.

Four new geoglyphs were identified during this process. Of the four, they depict a human-like figure holding a club, a pair of legs or hands stretching more than 250 feet, a fish, and a bird. The new designs were subsequently verified during on-site visits.

“We could identify new geoglyph’s candidates approximately 21 times faster than with the naked eye alone,” said researchers. “The approach would be beneficial for the future of archaeology in a new paradigm of combining field survey and AI.”

Late last year, the team discovered 168 new geoglyphs using aerial photography and drone images.


No votes yet.
Please wait...