Applying deep learning in ecology: identifying vegetation and plant species

Monitoring and mapping species and their habitats are foundational tasks in ecology and natural resource management. However, the conventional method use to accomplish these tasks, i.e., field surveys, is highly labour intensive. Automated species identification from visual images is one of the most...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Hozen Seitaigaku Kenkyu = Japanese Journal of Conservation Ecology 2020/03/05, Vol.25(1), pp.1822
Hauptverfasser: Watanabe, Shuntaro, Onishi, Masanori, Minagawa, Mari, Ise, Takeshi
Format: Artikel
Sprache:jpn
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Monitoring and mapping species and their habitats are foundational tasks in ecology and natural resource management. However, the conventional method use to accomplish these tasks, i.e., field surveys, is highly labour intensive. Automated species identification from visual images is one of the most promising avenues to reducing the costs of monitoring and mapping species distributions. Deep learning technologies are becoming increasingly accurate for object detection and image classification. In this paper, we focus on deep learning methods that have shown excellent performance in object detection and image classification in recent years. First, we provide a brief overview of convolutional neural networks, one of the primary algorithms used in deep learning. Second, we review case studies that have applied deep learning to plant species identification and vegetation mapping and discuss future applications. We recommend that future work aim to provide a framework for systematically collecting labelled image data to enable the mapping and monitoring of biodiversity at high temporal resolutions and low cost.
ISSN:1342-4327
2424-1431
DOI:10.18960/hozen.1822