Automatic Tree Species Identification in a Cold Temperate Natural Broadleaf Mixed Forest Using High-Resolution UAV Imagery and Mask R-CNN

Puntos clave

• This research aims to enhance tree species identification in complex mixed forests using UAV imagery and Mask R-CNN.
• Utilized UAV-QField leaf-canopy validation approach for tree species validation.
• Employed multi-temporal UAV imagery for manual annotation of tree canopies.
• Implemented multi-class and species-specific Mask R-CNN models for tree detection.
• Mask R-CNN models demonstrated varying performance in detecting different tree species.
• The UAV-QField approach validated tree species effectively in closed canopy environments.