What can machine learning tell us about the rock art in Arnhem Land?
South Australian researchers, led by Daryl Wesley of Flinders University, working with the Mimal and Marrku Traditional Owners of the Wilton River area in the Northern Territory, took a close look at the rock art in Arnhem Land to examine how the art transformed stylistically over time.
The team used machine learning to analyse images of different styles and subjects, to see how similar they were at a minute level, and what the chronology of art evolution was.
Key research points
- Machine learning analysed rock art in Arnhem Land
- Drawings of a similar age had a similar style
- This approach may be useful for future archaeological studies with big datasets
“One amazing outcome is that the machine learning approach ordered the styles in the same chronology that archaeologists have ordered them in by inspecting which appear on top of which,” says Jarrad Kowlesser, a researcher at Flinders University.
“This shows that similarity and time are closely linked in the Arnhem Land rock art and that human figures drawn closer in time were more similar to one another than those drawn a long time apart.
“For example, the machine learning algorithm has plotted Northern Running figures and Dynamic figures very close to one another on the graph it produces.
“This shows that these styles which we know are closer to each other in age are also closer to each other in appearance, which might be a very hard thing to notice without an approach like this.”
The team first taught the computer how to recognise different images by using an existing dataset of 14 million photos of animals and objects. This model was then applied to the rock art images. The results are published in Australian Archaeology. The code has been made available via GitHub.
“In total the computer saw more than 1000 different types of objects and learned to tell the difference between them just by looking at photos of them,” Dr Wesley explains.
“The important skill this computer developed was a mathematical model that has the ability to tell how similar two different images are to one another.”
This approach, called ‘transfer learning’, could remove possible bias in human evaluation, especially when tiny details are easily missed. It might also be a useful tool for future archaeological studies.
More on Indigenous rock art