This deep learning model is designed to detect and segment individual trees in high-resolution drone and aerial imagery. Tree detection supports a wide range of applications, including vegetation management, forestry monitoring, and urban planning. The high spatial and temporal resolution of drone and aerial imagery makes it ideal for accurately identifying and analysing tree coverage across various landscapes.

The model architecture is based on DeepForest and trained using annotated data from the National Ecological Observatory Network (NEON). It also integrates the Segment Anything Model (SAM) , developed by Meta, to enhance segmentation accuracy and generalization across diverse environments.