Looting historical heritage is an illicit phenomenon that affects archaeological areas worldwide, particularly where surveillance is low, and access proves difficult. Carried out with different methods (by hand or mechanical support) and rates (systematic or casual), every looting enterprise leaves distinct marks on the terrain and damages the involved areas to a varying degree. A looted area could therefore display just a couple of holes or reveal a ‘moonlike’ scenario, full of deep and large craters, as in Dura Europos or Apamea (Syria).
Finding ways to detect, monitor, and tackle looting activities and the subsequent illicit trafficking of antiquities is currently one of the major challenges in Cultural Heritage preservation. The most efficient approach relies on Earth Observation (EO) data exploitation. In critical areas, such as conflict zones, country borders, desert areas, and remote regions, this technique can often be the only one available to determine when, how, and where looting happens. By exploiting the potential of time-series images, this method also enables the retrieval of information related to the shapes and patterns of the looting pits.
ALCEO project, in collaboration with and co-financed by the European Space Agency, aims to develop next-generation Artificial Intelligence methods to automatise the detection of looted sites on time series of EO data by building innovative Machine-Learning algorithms to fully exploit the large amount of data produced by satellite-based sensors and made available through platforms like Copernicus and USGS. By measuring the dissimilarities between consecutive satellite imagery, the system will be able to automatically detect the relevant ‘anomalies’ recognising typical patterns and features of looting activities.