Detection of Anolis carolinensis using drone images and a deep neural network: an effective tool for controlling invasive species

Tomoki Aota, Koh Ashizawa, Hideaki Mori, Mitsuhiko Toda, Satoshi Chiba

Research output: Contribution to journalArticlepeer-review

Abstract

Invasive species greatly disrupt island ecosystems, risk assessment and the conservation of native ecosystems have therefore become pressing concerns. However, the cost of monitoring invasive species by humans is often high. In this study, we developed a system to detect an invasive lizard species, Anolis carolinensis, that threatens the native insect ecosystem of the Ogasawara Islands in Japan. Surveying these forest lizards requires specialized field observers, a challenge that prevents the government of Japan from efficient conservation and management of this ecosystem. The proposed system detects these lizards in drone images using a type of machine learning called deep neural network. Data were collected using a drone on Ani-jima in the Ogasawara Islands, and the trained network shows approximately 70% precision of detecting A. carolinensis. This study shows the combination of remote sensing and machine learning have the potential to contribute to an efficient and effective approach to conserving ecosystems.

Original languageEnglish
Pages (from-to)1321-1327
Number of pages7
JournalBiological Invasions
Volume23
Issue number5
DOIs
Publication statusPublished - 2021 May

Keywords

  • Conservation
  • Invasive species
  • Lizard
  • Machine learning
  • Remote sensing
  • Reptile

ASJC Scopus subject areas

  • Ecology, Evolution, Behavior and Systematics
  • Ecology

Fingerprint

Dive into the research topics of 'Detection of Anolis carolinensis using drone images and a deep neural network: an effective tool for controlling invasive species'. Together they form a unique fingerprint.

Cite this