Dual-Task Learning for Dead Tree Detection and Segmentation with Hybrid Self-Attention U-Nets in Aerial Imagery

2025-03-27

AI Summary

Detecting and instance-segmenting dead trees in aerial imagery remains challenging due to severe crown occlusion, spectral ambiguity between live and dead vegetation, and over-segmentation. To address these issues, this paper proposes a U-Net architecture enhanced with self-attention mechanisms, coupled with a novel adaptive filtering–watershed hybrid post-processing framework that jointly optimizes detection and instance segmentation. Evaluated on high-resolution aerial imagery of boreal forests, the method achieves a 41.5% improvement in instance segmentation accuracy and a 57% reduction in localization error compared to baseline approaches. Notably, it is the first method to enable accurate single-tree-level identification of dead trees beneath dense canopies, thereby enabling large-scale forest mortality mapping. This advancement provides a robust technical foundation for forest health monitoring, carbon sink assessment, and wildfire risk mitigation.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Unsupervised & Self-Supervised Learning
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

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 postprocessing framework that refines deep learning-based tree segmentation by integrating watershed algorithms with adaptive filtering, enhancing boundary delineation, and reducing false positives in complex forest environments. Tested on high-resolution aerial imagery from boreal forests, the framework improved instance-level segmentation accuracy by 41.5% and reduced positional errors by 57%, demonstrating robust performance in densely vegetated regions. By balancing detection accuracy and over-segmentation artifacts, the method enabled the precise 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 (identifying fuel accumulations), carbon stock estimation (tracking emissions from decaying biomass), and precision forestry (targeting salvage loggings). By bridging advanced remote sensing techniques with practical forest management needs, this work advances tools for large-scale ecological conservation and climate resilience planning.

Problem

• Detects and segments dead trees in aerial imagery
• Addresses spectral overlaps and over-segmentation errors
• Improves accuracy for ecological monitoring and wildfire risk

Innovation

• Hybrid self-attention U-Nets for dead tree detection
• Watershed algorithms with adaptive filtering integration
• Improved segmentation accuracy and reduced positional errors