Improving New Zealand’s vegetation mapping using weakly supervised deep learning

The New Zealand Land Use Carbon Analysis System Land Use Map is a series of land use layers that map land use classes, including exotic and native forest, dating back to 1990 and updated every four years since 2008. This map is a rich resource, but the high effort required to update the map means errors may creep in without detection. We trialled whether a deep learning model trained on this imperfect data could be practically used to detect and correct errors, including missed exotic forest plantations and wilding conifer infestations. Our results suggest the approach is feasible.