Neural network to help find hidden archaeological treasures

Researchers hope their neural network will be able to spot clues in satellite images. Google Earth

Researchers are developing a way to use artificial intelligence to detect ancient burial grounds and settlements.

A team from the University of Warsaw have come up with a system that will use a neural network, a series of algorithms that tries to recognise underlying relationships in a set of data by mimicking how the human brain operates.

Team leader Piotr Wroniecki said: “If the system works as expected, it could successfully revolutionise archaeology – as the introduction of radiocarbon dating did in the first half of the 20th century.”muzeum.bialystok.pl

For this, the team which is made up of archaeologists, geologists and IT experts, will need to feed a very large amount of data into the system, in the form of satellite images.

Archeologists have long been analysing these images, but, done by humans, the process is time consuming and subjective.

The neural network uses a series of algorithms that tries to recognise underlying relationships in a set of data by mimicking how the human brain operates.Google Earth

Team leader Piotr Wroniecki said: “We hope that artificial intelligence will be able to mark them quickly, efficiently and objectively.

“If the system works as expected, it could successfully revolutionise archaeology – as the introduction of radiocarbon dating did in the first half of the 20th century.”

Archeologists have long been analysing these images, but, done by humans, the process is time consuming and subjective.Google Earth

The neural network will be designed to spot clues in the vegetation in the satellite images that suggest that there were once buildings there.

For example, the grass or crops grow slightly differently on land were there used to be buildings, compared to the surrounding landscape.

After identifying areas of interest, the system will mark them on the map for archaeologists.Google Earth

After identifying these areas, the system will mark them on the map for the archaeologist.

This will speed up the process significantly: whereas it might take archaeologists weeks or months to analyse images of the land in this way (depending on the size of the area), the use of the neural networks can cut this time down to a matter of hours.