| Authors | Manner and Lundström |
| Country | Skogforsk (Research Modelte) |
| Paper (PDF) | View paper |
| Category | Harvesting & Extraction › Quality & optimisation |
as forked trees reduce harvesting efficiency, an early removal (e.g. during thinning) would be benefitial. techModellogy to solve this problem?
0
Detection of forked trees is an emerging application of computer vision and machine learning in precision forestry, aimed at automatically identifying trees with bifurcated or multi-stemmed growth forms before or during harvesting operations. A forked tree — one in which the main stem divides into two or more co-dominant leaders — represents a significant timber quality defect, reducing log recovery rates, complicating harvester head processing, and increasing the risk of mechanical damage and operator injury. Traditionally, identification of forked stems relied on manual cruising by field foresters or was discovered only at the time of felling, creating operational inefficiencies. Automated detection approaches use a combination of aerial and terrestrial imagery, LiDAR point clouds, and machine learning classifiers — including convolutional neural networks (CNNs), random forests, and instance segmentation models such as Mask R-CNN — to flag forked individuals within a stand prior to harvest planning. LiDAR-based methods analyse crown geometry and stem divergence in point cloud data, while image-based approaches leverage canopy texture and branching patterns visible in high-resolution drone or aerial imagery. In Australian plantation forestry, where eucalypt species such as Eucalyptus nitens and E. globulus are prone to forking due to wind damage, frost events, and genetic factors, automated forking detection has direct economic value by informing selective thinning prescriptions and harvest sequencing. Detection models are typically trained on labelled datasets of manually annotated tree inventories and validated against field measurements. Outputs can be integrated into forest management information systems, enabling harvester operators to pre-load forking flags into machine control systems, adjust head settings, or manually process problem stems, thereby reducing downtime and improving overall harvesting efficiency.
Improves harvester efficiency by identifying problematic stems early, with moderate risk-reduction benefits related to avoiding cycle-time delays and quality issues.
| Field | Value |
|---|---|
| Company | Manner and Lundström |
| Country | Skogforsk (Research Modelte) |
| Type | Paper |
| Year | 2024 |
| Remarks | 0.0 |
| Case Studies | 0 |