Files and links (2)
Comparing random forests and convoluted neural networks for mapping ghost crab burrows using imagery from an unmanned aerial vehicle2.52 MB
Accepted VersionCC BY-NC-ND V4.0, Open Access
Journal article
Comparing random forests and convoluted neural networks for mapping ghost crab burrows using imagery from an unmanned aerial vehicle
Estuarine, Coastal and Shelf Science, Vol.224, pp.84-93
2019
Abstract
Sandy beaches are important ecosystems that line a third of the world's ice-free coastlines. Unfortunately, these environments and the life they support are often threatened by various anthropogenic and natural factors. Monitoring sandy beach health is important to aid in appropriate management decisions. One such method of quantifying environmental health is using bioindicators. Ghost crabs are a commonly used sandy beach bioindicator. Current techniques for assessing ghost crab abundance and distribution data involve manually counting each individual burrow opening, which can intrusive and timely for a large area. The aim of this study was to assesses the use of imagery from an unmanned aerial vehicle and machine-learning algorithms as an alternative approach to monitoring ghost crab burrows. The accuracy and transferability of random forest (RF) and convolutional neural networks (CNN) classifiers within an Object-Based Image Analysis (OBIA) framework were tested using hyper-resolution (0.04 m) orthomosaics from four different dates. CNN was a more robust classifier with higher accuracies (max F-score 0.84). Transferability of rule sets and models was limited for both classifiers, particularly when applied to sub-optimal imagery. Overall, we present a feasible workflow that provides ecologist and environmental managers with a cost-effective and less invasive alternative to mapping ghost crab burrows.
Details
- Title
- Comparing random forests and convoluted neural networks for mapping ghost crab burrows using imagery from an unmanned aerial vehicle
- Authors
- Rachel Bycroft (Author) - University of the Sunshine Coast - School of Science and EngineeringJavier X Leon (Author) - University of the Sunshine Coast - School of Science and EngineeringDavid S Schoeman (Author) - University of the Sunshine Coast - School of Science and Engineering
- Publication details
- Estuarine, Coastal and Shelf Science, Vol.224, pp.84-93
- Publisher
- Academic Press
- Date published
- 2019
- DOI
- 10.1016/j.ecss.2019.04.050
- ISSN
- 0272-7714
- Copyright note
- Copyright © 2019. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/
- Organisation Unit
- School of Science and Engineering - Legacy; University of the Sunshine Coast, Queensland; School of Science, Technology and Engineering; Sustainability Research Cluster
- Language
- English
- Record Identifier
- 99450666902621
- Output Type
- Journal article
Metrics
72 File views/ downloads
268 Record Views
InCites Highlights
These are selected metrics from InCites Benchmarking & Analytics tool, related to this output
- Collaboration types
- Domestic collaboration
- International collaboration
- Web Of Science research areas
- Marine & Freshwater Biology
- Oceanography
UN Sustainable Development Goals (SDGs)
This output has contributed to the advancement of the following goals:
Source: InCites