Humber, Isaac (2023) A Deep Learning Solution for the Detection of Health and Productivity Metrics in Sandalwood Forest Plantations Using Drone Imaging. [USQ Project]
![]() |
Text (Project – redacted)
Humber_I_Low_Redacted.pdf Download (3MB) |
Abstract
The production of Sandalwood Oil is a highly lucrative industry, and one in which Australia is currently the global leader. The processes currently applied to monitor tree health and predict the volume of marketable product at harvest are costly in both time and resources. Deep learning has shown promise for the automatic monitoring of health and volume in silviculture plantations using overhead imagery. However, this has never been done on the individual tree level for the 5 class health score or bole height and diameter at breast height applied by the industry. Nor has it been attempted within Sandalwood plantations. Thus, a two stage, deep learning based system was proposed, informed by the findings of relevant literature. The goal of the system was the of full automation of both general health monitoring and marketable volume prediction on the individual tree level, thus adding value to the imagery already recorded by the industry as part of yearly inventory.
The resulting system was able to detect trees within the plantations at an accuracy comparable to leading algorithms. Moreover, the system was able to classify the health scores used in industry-standard volume estimation calculations with an Average Precision of 0.97 and sample-weighted F1 score of 0.92, exceeding the performance of other tree health classifiers proposed in the literature.
![]() |
Statistics for this ePrint Item |
Item Type: | USQ Project |
---|---|
Item Status: | Live Archive |
Faculty/School / Institute/Centre: | Current – Faculty of Health, Engineering and Sciences - School of Engineering (1 Jan 2022 -) |
Supervisors: | Low, Tobias |
Qualification: | Bachelor of Engineering (Honours) (Computer Systems) |
Date Deposited: | 25 Sep 2025 02:15 |
Last Modified: | 25 Sep 2025 02:15 |
Uncontrolled Keywords: | drone imaging; forestation |
URI: | https://sear.unisq.edu.au/id/eprint/52957 |
Actions (login required)
![]() |
Archive Repository Staff Only |