Skip to main content


Anomalies in satellite imagery

Trainee Achievements

Anomalies in satellite imagery

IGERT-TEECH Trainee Andew Huynh is training machines to look for anomalies in high-resolution, multi-spectral satellite imagery. To develop his machine-learning algorithm, he is using unique meta-data, consisting of over 2 million manually created tags, stemming from a crowd-sourcing project conducted in collaboration with the National Geographic Society. Using clustering and statistical methods, priors are created for the training of the machine-learning algorithm, allowing archeological sites of interest to be auto-detected thereafter. An initial round of human-centric analysis runs with on-site verification by a team in the field, resulted in the identification of over 50 archaeological sites in a two week period.