Mapping the distribution of invasive tree species using deep one-class classification in the tropical montane landscape of Kenya

Some invasive tree species threaten biodiversity and cause irreversible damage to global ecosystems. The key controlling monitoring the propagation of is detect their occurrence as early possible. In this regard, one-class classification (OCC) shows potential in forest areas with abundant richness since it only requires a few positive samples be mapped, instead all species. However, classical OCC method remote sensing heavily dependent on manually designed features, which have limited ability complex distributions. Deep learning based methods mostly focus multi-class classification, there been studies deep paper, unlabeled framework—ITreeDet—is proposed for identifying Eucalyptus spp. (eucalyptus) Acacia mearnsii (black wattle) Taita Hills southern Kenya. ITreeDet framework, an absNegative risk estimator train robust model by fully using massive data. Compared state-of-the-art methods, represents great improvement detection accuracy, F1-score was 0.86 0.70 eucalyptus black wattle, respectively. study area covers 100 km2 Hills, where, according our findings, total wattle 1.61 3.24 km2, respectively, represent 6.78% 13.65% covered trees forest. addition, both are located higher elevations, extensive spread around confirms its tendency. maps generated use algorithm will help local government develop management strategies these two