Finding Sites in the Amazon Forest: AI-Based Deep Learning Analysis of Satellite Imagery from the Upper Xingu Basin, Brazil
Author(s): Wetherbee Dorshow
Year: 2021
Summary
This is an abstract from the "Archaeology in the Xingu River Basin: Long-Term Histories, Current Threats, and Future Perspectives" session, at the 86th annual meeting of the Society for American Archaeology.
This paper summarizes preliminary results of an AI-based analysis that identifies potential precolumbian Amazonian archaeological site locations based on the presence of clusters of a specific species of palm tree. The study uses Landsat-8, Sentinel-2, and Planet satellite imagery as input to Deep Leaning Object Detection AI analysis, delineating potential for the presence of disturbed plaza areas, trash mounds, and Terra Preta / Egepe zones (anthropogenic soils) indicative of large precolumbian sites in the Upper Xingu Basin.
Cite this Record
Finding Sites in the Amazon Forest: AI-Based Deep Learning Analysis of Satellite Imagery from the Upper Xingu Basin, Brazil. Wetherbee Dorshow. Presented at The 86th Annual Meeting of the Society for American Archaeology. 2021 ( tDAR id: 467202)
This Resource is Part of the Following Collections
Keywords
Geographic Keywords
South America: Amazonia and Orinoco Basin
Spatial Coverage
min long: -81.914; min lat: -18.146 ; max long: -31.421; max lat: 11.781 ;
Record Identifiers
Abstract Id(s): 32323