Home Science Artificial Intelligence to Identify Roman Gold Mine Remains

Artificial Intelligence to Identify Roman Gold Mine Remains

Artificial Intelligence to Identify Roman Gold Mine Remains

He northwestern peninsula Ibérica has one of the largest Roman gold mining facilities the world, mainly exploited between the 1st and 3rd centuries AD. C. According to Latin authors such as Pliny the Elder, who described the situation in what is now Galicia, Asturias and León, more than 6.5 tons of the precious metal were mined during this period.

In this last province there are The Marks, one of the best preserved Roman mining remains in Europe. This old gold mine, declared a UNESCO World Heritage Site in 1997, has a Hydraulic system Consisting of a network of channels carved into the rock with a length of more than 700 km. The water was used to collapse the mountain and wash the materials to extract the golden metal.

New technologies have already been used in the exploration of this entire 2,000-year-old mining complex, such as: Drones and airborne LiDAR technology, but now researchers at the University of León have integrated the Artificial Intelligence (AI) identify and map ancient remains in the landscape.

Machine learning algorithms and drone imagery are combined to identify Roman mining sites and their features, such as hydraulic channels

The work, published in the journal Applied Intelligence“combines machine learning algorithms (deep learning) and georeferenced drone images to identify mining sites and other elements of the Roman framework, such as channels of the hydraulic system,” the lead author tells SINC. Daniel Fernandez Alonso.

“To do this,” he continues, “we trained this intelligent system on images with similar geometric patterns that could easily be confused with mining remnants (such as roads, paths and other elements that make up the landscape), and so adapted them. ” system until a 95% “% correctness in identifying the different structures of the old gold mines” was achieved.

In the area deep learningwhich allows automatic recognition without the intervention of human subjectivity, the authors have used the so-called Convolutional Neural Networks, capable of learning to improve the characteristics that best fit the elements defined during training. For example, distinguishing a road from a canal like that of Peña Aguda, a nearly 43 km long structure within Las Médulas.

“This type of neural network perfects a series of filters that, when applied to the images, highlight the parts with the elements that we are trying to find and distinguish, in our case mining remains and intersections,” says the co- Author. Maria Teresa García Ordas“and also has a layer of connected neurons that correctly classifies the images generated after applying the filters.”

Useful tool for archaeologists

The researchers recall that mountain areas such as those studied are in many cases difficult to access and that the landscape has changed over the centuries due to the increase in vegetation and anthropogenic activities. This makes it difficult to identify ancient mining infrastructures like the Roman ones, but the new methodology provides a useful tool to solve the problem and help archaeologists, according to the authors.

another of them, Javier Fernandez Lozanoalso points out that it is the first time that AI has been used to identify canals and ponds. In addition, “it is modeling that will help in the future in locating additional Roman gold mining elements and even in the discovery of new gold deposits or deposits.”

In the future, this method could be implemented to include additional images to identify patterns of new gold deposits.

“To do this,” he clarifies, “our method would have to be implemented and also include other types of images, such as those taken with multispectral and hyperspectral cameras, in order to recognize characteristic patterns of gold deposits that it can then deal with.” be compared based on a study.

Examples with a road intersection and a landscape as well as a canal used to train the machine. / J. Fernández Lozano et al.

This method can also be used to identify and reduce the risks of abandoned mines, thereby preventing accidents and economic losses.

He also emphasizes that this method, “for which we have already applied for a patent,” allows us to identify Remains of a more modern mining industry present in the landscape.

“The province of León – he gives an example – has, in addition to gold, a long history of coal, iron and tungsten mining, among other things. Recognizing the traces they have left on the territory makes it easier proper management by the administrationsThis will avoid personal accidents and costs associated with abandoning wells and other mining infrastructure.”

In fact, the researchers conclude: “This novel application of deep learning can be implemented to reduce the potential risks posed by abandoned mines (particularly unmarked mines), which can result in significant human and economic losses worldwide.

Reference:

Daniel Fernández-Alonso et al. “Convolutional neural networks for accurate identification of mining remains from UAV-derived images“. Applied Intelligence2023.

The departments Systems Engineering and Automatics (SECOMUCI Research Group) and Mining Prospecting and Research (GEOINCA Research Group) at the University of León are involved.

No Comments

Leave A Reply

Please enter your comment!
Please enter your name here

Exit mobile version