Artificial intelligence discovers 4 mysterious shapes hidden in the Peruvian landscape -

YAMAGATA, Japan — Artificial intelligence (AI) has helped scientists uncover four enormous shapes concealed within the Peruvian landscape. The research focused on the Nazca Desert in southern Peru, known for the enigmatic Nazca Lines. These geoglyphs, some stretching up to 370 meters, were created between 500 BCE and 500 CE. According to a team in Japan, their creators etched shallow depressions or incisions into the desert floor, removed pebbles, and exposed differently colored dirt.

The study, undertaken by researchers at the Yamagata University Institute of Nasca and IBM Japan, details the deep learning AI model employed to survey these figurative geoglyphs. This led to the uncovering of four geoglyphs: an odd humanoid figure, a pair of legs, a fish, and a bird. The humanoid shape was revealed in 2019.

Before this, investigators would have had to examine aerial photographs of extensive areas manually, a method that was both time-consuming and inefficient. The research paper reveals that deep learning technology allows for the identification of potential geoglyphs approximately 21 times faster than what human capabilities permit.

five pictures side by side labeled A B C D EA is brown land with a lines through B is gray pic with carvings of what looks like a person with a headress. C is light brown with the image of what looks like a bird D is light tan with carving of a long nosed animal with two feet E is dark gray with that shape of long bodies animal with a long neck — Fig. 1. Types of figurative geoglyphs. There are two main types of figurative geoglyphs: line-type and relief-type. Both were created on the Nasca Pampa covered with black stones on top of a white layer of sand. (A) “Line-type figurative geoglyphs” were made by removing black stones in a linear pattern exposing the white sand underneath. (B to E) “Relief-type figurative geoglyphs” are often located on slopes and comprise a combination of black stone and white sand surfaces. This type of geoglyph can be categorized into subtypes 1–4 as follows: (B) “Relief-type 1” was made by removing black stones to expose the white sand at the motif. Some of the removed stones were reused to depict the details of the motif. (C) “Relief-type 2” was made by removing black stones around the motif. Some of the removed stones were piled on top of the motif to make it more prominent. (D) “Relief-type 3” was made by removing black stones along the outline of the motif and by piling the removed stones on top of the motif. (E) “Relief-type 4” is white because it was made by removing all the black stones from the inside of the motif.
(credit: Yamagata University)

“Due to the requirement of detecting unconfirmed geoglyph candidates, careful consideration and ingenuity were needed in order to train a deep learning object detection model using training data of very limited quality and quantity,” the team from Yamagata University explain in a media release.

peru symbol — AI discovers mysterious shapes hidden in Peruvian landscape (Credit: Yamagata University)

Following the success of this preliminary study, the researchers are now collaborating with IBM T. J. Watson Research Center to conduct a large-scale, AI-based survey of geoglyphs across the entire Nasca Pampa. They also intend to collaborate with Peru’s Ministry of Culture to initiate measures designed to safeguard the geoglyphs.

Peruvian geoglyph of a fish — Geoglyph of a fish. (Credit: SWNS)

Anthropologists, ethnologists, and archaeologists have endeavored to unravel the mystery of the lines by examining the ancient Nazca culture. One hypothesis is that the Nazca people created the lines to be seen by deities in the sky. While some Nazca lines form shapes best observed from the air at around 500 meters (over 1,600 feet), they can also be seen from surrounding foothills and other elevated vantage points.

The joint study is published in the Journal of Archaeological Science.

South West News Service writer Dean Murray contributed to this report.

Tags: ancient history, archaeology, artificial intelligence, machine learning, Peru