The PalmsCNN repository provides the code for training a model designed to segment and detect the crowns of three palm species in Amazonian forests using RGB imagery captured by drones. This work is part of the research conducted by Tagle et al. 2025 "Effective integration of drone technology for mapping and managing palm species in the Peruvian Amazon."

The proposed approach combines drone-based data with an architecture that integrates ecoCNN for data generation, DeepLabv3+ for semantic segmentation, and DWT (Deep Watershed) for image processing. This combination enables precise and efficient detection of palm crowns in drone imagery.

Species Detected The model is trained to detect three ecologically and economically important palm species