Development of a neural network approach for automated recognition of prehistoric carvings at Stonehenge

Dense lichen covers roughly a quarter of the above-ground stone surfaces at Stonehenge, which renders them inaccessible to study. A to¬tal of 72 Early Bronze Age carvings have recently been found on the bare stone surfaces, prompt-ing concerns that lichen may be obscuring pre¬historic rock art. As a first step towards creat¬ing a technique for revealing carvings beneath lichen, photography-derived 3D modelling and machine learning were combined to create a method for identifying carvings on bare stone surfaces. Tasked with differentiating between areas of the stone surfaces with and without carvings, the method achieved 84.2% accuracy. With further development, this work could be used by rock art conservators and archaeolo¬gists to verify carving findings, search for pre¬viously unidentified carvings and eventually reveal carvings hidden by lichen.