Premium
Cover Picture and Issue Information
Publication year - 2019
Publication title -
methods in ecology and evolution
Language(s) - English
Resource type - Reports
SCImago Journal Rank - 3.425
H-Index - 105
ISSN - 2041-210X
DOI - 10.1111/2041-210x.13044
Subject(s) - odocoileus , cover (algebra) , rhinoceros , computer science , white (mutation) , geography , field (mathematics) , cartography , data science , artificial intelligence , information retrieval , world wide web , ecology , biology , biochemistry , mathematics , pure mathematics , gene , mechanical engineering , engineering
This month’s cover image shows a composite of four photos taken by camera traps from four different field studies. A buffalo herd ( Syncerus caffer caffer ) in the Serengeti National Park in Tanzania, an African forest elephant ( Loxodonta cyclotis ) in Gabon, a white rhinoceros ( Ceratotherium simum ) in South Africa, and a white‐tailed deer ( Odocoileus virginianus ) in Wisconsin, USA. Camera‐trapping is an increasingly popular method to collect data for ecological analyses that yields datasets with up to millions of images. The associated article describes how to annotate such large datasets in a timely manner by teaching powerful deep learning algorithms to identify animals in the images. The study finds that models accurately identify most species and reliably indicate cases with low confidence. It successfully demonstrated the application of such models on a citizen science platform ( www.zooniverse.org ) by combining machine‐ and human opinions in real‐time to reduce overall human effort. The proposed approach significantly reduces the lag between data collection and ecological analysis and allows for conducting ever larger camera trapping studies. The software used to train the deep learning models is publicly available and is currently in use. Photo credit: © snapshotserengeti; © Panthera; © A W Cardoso, University of Oxford; © Snapshot Wisconsin, Wisconsin Department of Natural Resources.