It is not the eye of Shiva, it is much more powerful. It is capable of detecting on the ground a piece of ancient pottery, however tiny it may be. He is also capable of seeing phytocytes on the ground, the skeletons left by plants when they rot, the molds that the passage of time creates on the ground; he is able to see them, even if they measure a few millimeters. He not only sees on the ground, he also sees on old maps. He discovers in the lines of roads or walls or crops or mounds where there could have been some type of human settlement, some construction or town or city or road or temple. These are the new tools of computational archaeology, algorithms that, when well trained, are capable of seeing from satellites or drones or aircraft, but are also capable of passing their eyes through digitized old maps. They scan the territory and point: “There”.
That divine eye that is the archaeological algorithm has made it possible to discover some 6,000 potential places of interest in the Indus river valley, between Pakistan and India.
Both this discovery and the design and feeding of the algorithm are part of the doctoral thesis of Iban Berganzo, a telecommunications engineer with a doctorate in computational archeology and a researcher at the Institut Català d’Arqueologia Clàssica (ICAC), linked to the Rovira i Virgili University of Tarragona . It was directed by Héctor Orengo (ICAC) and Felipe Lumbreras, from the Computer Vision Center of the UAB.
“These are the lines of work today, the ones that combine specialties,” Berganzo explains to La Vanguardia. “It is not unreasonable to think that we can have a world archaeological cartography. And our teams are at the forefront of computational archeology in the world.”
The thesis describes how in an area of ??the brutal dimension of 475,000 km², along the valley of the Indus River, he has been able to point to some 6,000 possible archaeological objects. It is an unprecedented milestone in the reconstruction of the ancient landscape. Berganzo plans to extend the investigation to all of India.
This specialist has worked that territory because of the links between Orengo, his tutor, and Cambridge. The Cambridge University Library and the British Library preserve the highly detailed maps of the colonial era that have allowed –before and after the verification of the verified archaeological information– to train digital tools.
“We are launching artificial intelligence by setting in the algorithm the data of deposits of which you are certain: location, shapes… a mound, a ceramic. You have to go on validating what you are learning, until there comes a time when you detect what each thing is in real time”, says the engineer. “Until you get the positive detection to be statistically significant.” One of the technologies is Lidar (Light Detection and Ranging), which allows you to scan reliefs from the air.
One of the possibly most valuable functions is that these systems can allow (see opposite information) effective monitoring tools of the deposits in real time, to prevent looting and other threats.
This methodology not only seems like magic, it not only yields these spectacular results, analyzing old maps, but also opens up a practically infinite catalog of opportunities for archaeology. It is being applied in Syria and Lebanon, with maps of the French colonization, thus looking for ceramics in Greek sites and has been applied in Catalonia for the detection of walled structures in Iberian sites. A couple of years ago, in Galicia, a whopping 10,000 burial mounds or places of potential cultural interest were discovered using this method.
The excavations now begin from the air. With wide-ranging views from above that allow you to discover archaeological remains in thousands of square kilometers without having to tread the ground, without having to break a sweat, in search of three stones in a row that suggest an old wall. “We have calculated that to review the maps that our algorithms have reviewed in six hours and with no margin of error, we would need specialized archaeologists who would work 120 hours to detect the same thing,” he exemplifies. Twenty times less.
In this way, and thanks to the progressive training of the algorithm and artificial intelligence, any Lidar image from anywhere in the world can be used to detect reservoirs or objects. For Berganzo, however, this is a complement to the work on the dust: “We must never disdain the field work. We free the archaeologist from certain routine operations so that he can focus on the analysis of the sites”.
