Cloud-based analysis of aerial imagery for unveiling ancient archaeological patterns

Image processing and artificial intelligence techniques represent new and effective tools for supporting archaeological research to bring ancient finds to light. They can help archaeologists to discover remains that are difficult to identify using traditional approaches. The design and development of such applications, which aim at processing large amounts of data to cover extended areas, requires the use of Cloud paradigms for exploiting Cloud elasticity and scaling with the problem size. This paper presents an original methodology that integrates deep learning, computer vision, and optimization models to identify archaeological remains from aerial images. Results demonstrate how the proposed approach can search for the remains of Centuriation, which is an ancient Roman system for dividing the land over a large area, and evaluate the scalability of a map-reduce implementation in the Cloud.