Detection of Pottery Sherds Using sUAS-Based Multispectral Imagery with Rule-Based and Machine-Learning Methods: An Experimental Archaeology on Horse Island in Branford, CT

Summary

This is an abstract from the "SAA 2025: Individual Abstracts" session, at the 90th annual meeting of the Society for American Archaeology.

From the increasing affordability and availability of commercialized drone products in the recent decade, drone imaging has wide application in various disciplines including archaeology. The full potential of multispectral drone imaging has yet to be explored for remote sensing in archaeological projects. In March 2023, the team from the Yale Ancient Pharmacology Program conducted an experimental archaeology project to collect multispectral drone imaging data from the modern pottery sherds spread over the land cover on Horse Island in Branford, CT, stewarded by the Yale Peabody Museum. Under the approval of Department of Environmental Health and Safety of Yale University, the project was fully operated by licensed sUAV pilots. This study uses three algorithms - thresholding, unsupervised, and supervised machine learning - to evaluate the respective effectiveness as detection methods. The reflectance spectrum of pottery sherds demonstrates that the Near-Infrared band is the most effective for detection methods. While all methods cannot fully exclude non-pottery features, the unsupervised method has the most capability in batch-processing in the field. Through the a priori distribution of pottery sherds, this technique has implications in optimizing the research design of archaeological surveys. Further studies will develop a normalized index for pottery for remote sensing.

This Resource is Part of the Following Collections

• Society for American Archaeology 90th Annual Meeting, Denver, CO (2025)   •
• SAA 2025: Individual Abstracts

Keywords

Record Identifiers

Abstract Id(s): 53053