Landscape-scale discovery democratised
Current factFinding sites required foot survey or costly aerial campaigns, capping searchable territory and concentrating discovery where teams could physically go.
Deep learning (CNNs) on airborne LiDAR and satellite imagery detects features across vast areas including under canopy, turning years of survey into weeks.
Landscape-scale remote prospection by smaller teams
MappingObservingInferring settlement systems from remote data
HypothesizingPattern-finding
- Maya settlement detection in NASA G-LiHT lidar
- Archaeoscape aerial-laser-scanning deep-learning benchmark
- CORONA-imagery site detection in Mesopotamia
Low on direction (established and expanding); moderate on magnitude (employment reshaping vs inventory expansion).
Detectors produce false positives needing ground-truthing; models inherit survey bias; finding is not digging or interpreting. Widens who can prospect, not who can excavate.