Piecing together history from shards is hard. Now robots can make it easier.
Archaeologists at Northern Arizona University are using Machine Learning to separate different pottery shards from each other, allowing them to sort them into different categories, reducing the time needed to reassemble and analyze the samples.
“Now, using digital photographs of pottery, computers can accomplish what used to involve hundreds of hours of tedious, painstaking and eye-straining work by archaeologists who physically sorted pieces of broken pottery into groups, in a fraction of the time and with greater consistency,” said Leszek Pawlowicz, adjunct faculty in the Department of Anthropology.
Their research appears in the June issue of the Journal of Archaeological Science.
Pawlowicz and his team took thousands of pictures of pottery fragments and then worked with four experts to train the ML network. Finally, once the shards had been identified, they turned the AI loose on the images, asking it to identify the shards in the same way the human experts did.
“The results were remarkable,” Pawlowicz said. “In a relatively short period of time, the computer trained itself to identify pottery with an accuracy comparable to, and sometimes better than, the human experts.”
You can read more about the system here. The real question, however, is when a robot arm will be able to reassemble a crazy number of shards into a real pot, an idea that probably gives Indiana Jones goosebumps.