Using Computer Vision to Detect and Classify Archaeological Finds from Turku Market Square Excavation

Modern computer vision technologies allow for the automatic detection and classification of objects visible in digital images. Robust applications in the context of archaeological remote sensing have already been presented, both in Finland and abroad. However, using these methods to detect and classify smaller objects is still only partially explored. In this paper, I examine the potential of using a modern open-source YOLOv11 model for archaeological find detection and classification. Training material has been collected from the large-scale Turku Market Square (Kauppatori) excavation (conducted between 2018 and 2022) and includes a wide variety of find types from the Early Modern Period. The results show great potential for using automated object detection with archaeological find material. Practical applications of this method could lead to significant labor savings in processing find material from large-scale excavations.