Nazca Lines

A great use for machine learning in finding, and hopefully helping to protect, indigenous art covering miles of ground in southern Peru.

Gouged into a barren stretch of pampa in southern Peru, the Nazca Lines are one of archaeology’s most perplexing mysteries. On the floor of the coastal desert, the shallow markings look like simple furrows. But from the air, hundreds of feet up, they morph into trapezoids, spirals and zigzags in some locations, and stylized hummingbirds and spiders in others. There is even a cat with the tail of a fish. Thousands of lines jump cliffs and traverse ravines without changing course; the longest is bullet-straight and extends for more than 15 miles.

The vast incisions were brought to the world’s attention in the mid-1920s by a Peruvian scientist who spotted them while hiking through the Nazca foothills. Over the next decade, commercial pilots passing over the region revealed the enormousness of the artwork, which is believed to have been created from 200 B.C. to 700 A.D. by a civilization that predated the Inca.

The newly found images — an average of 30 feet across — could have been detected in past flyovers if the pilots had known where to look. But the pampa is so immense that “finding the needle in the haystack becomes practically impossible without the help of automation,” said Marcus Freitag, an IBM physicist who collaborated on the project.

To identify the new geoglyphs, which are smaller than earlier examples, the investigators used an application capable of discerning the outlines from aerial photographs, no matter how faint. “The A.I. was able to eliminate 98 percent of the imagery,” Dr. Freitag said. “Human experts now only need to confirm or reject plausible candidates.”

Source: The New York Times

Backup link: Archive Buttons