ShrubNet-JER: Deep Learning-Based Vegetation Segmentation and Classification of Creosote and Mesquite in the Jornada Experimental Range

Multi-class instance segmentation model that detects individual shrubs, classifies them by species (creosotebush, Larrea tridentata, vs. honey mesquite, Prosopis glandulosa), and delineates their crowns in very-high-resolution (3.7 cm) 5-band multispectral imagery from the USDA-ARS Jornada Experimental Range (JER), New Mexico, USA. Trained with the ArcGIS Pro Train Deep Learning Model tool (arcgis.learn 2.4.2).