Dual-task learning for dead tree detection and segmentation with hybrid self-attention U-Nets in aerial imagery

Abstract

Mapping standing dead trees is critical for assessing forest health, monitoring biodiversity, and mitigating wildfire risks, for which aerial imagery has proven useful. However, dense canopy structures, spectral overlaps between living and dead vegetation, and over-segmentation errors limit the reliability of existing methods. This study introduces a hybrid post-processing framework that refines deep learning-based tree segmentation by integrating watershed algorithms with adaptive filtering to enhance instance separation and boundary precision. Leveraging a dual-task learning architecture with a Self-Attention U-Net, the framework simultaneously predicts segmentation masks, centroid heatmaps, and hybrid boundary maps, optimizing for both pixel-level accuracy and instance-level detection. Tested on high-resolution aerial imagery from boreal forests, the framework, compared to the U-Net baseline, improved instance-level segmentation accuracy by 41.5% (Tree IoU of 0.3810 vs. 0.2694) and reduced positional errors by 57% (centroid error of 3.70 pixels vs. 5.10 pixels), demonstrating robust performance in the densely vegetated boreal forest regions tested. By balancing detection accuracy (F1-score of 0.5895) and over-segmentation artifacts, the method enabled the accurate identification of individual dead trees, which is critical for ecological monitoring. The framework’s computational efficiency supports scalable applications, such as wall-to-wall tree mortality mapping over large geographic regions using aerial or satellite imagery. These capabilities directly benefit wildfire risk assessment, carbon stock estimation, and precision forestry. This work advances tools for large-scale ecological conservation and climate resilience planning.