@article{246563015f9b41f88db8091184758446,
title = "Deep learning enables satellite-based monitoring of large populations of terrestrial mammals across heterogeneous landscape",
abstract = "New satellite remote sensing and machine learning techniques offer untapped possibilities to monitor global biodiversity with unprecedented speed and precision. These efficiencies promise to reveal novel ecological insights at spatial scales which are germane to the management of populations and entire ecosystems. Here, we present a robust transferable deep learning pipeline to automatically locate and count large herds of migratory ungulates (wildebeest and zebra) in the Serengeti-Mara ecosystem using fine-resolution (38-50 cm) satellite imagery. The results achieve accurate detection of nearly 500,000 individuals across thousands of square kilometers and multiple habitat types, with an overall F1-score of 84.75% (Precision: 87.85%, Recall: 81.86%). This research demonstrates the capability of satellite remote sensing and machine learning techniques to automatically and accurately count very large populations of terrestrial mammals across a highly heterogeneous landscape. We also discuss the potential for satellite-derived species detections to advance basic understanding of animal behavior and ecology.",
keywords = "ITC-ISI-JOURNAL-ARTICLE, ITC-GOLD",
author = "Zijing Wu and Ce Zhang and Xiaowei Gu and Isla Duporge and Lacey Hughey and Jared Stabach and A.K. Skidmore and Grant Hopcraft and Stephen Lee and Peter Atkinson and Douglas McCauley and R. Lamprey and Ngene, {Shadrack M.} and Tiejun Wang",
note = "Funding Information: This collaboration was supported, in part, by grant no. 00138000039 from the Microsoft{\textquoteright}s AI for Earth Program ( https://www.microsoft.com/en-us/ai/ai-for-earth ). Z.W.{\textquoteright}s research was supported, in part, by funding from the Department of Natural Resources at the Faculty of Geo-Information Science and Earth Observation (ITC) of the University of Twente. This research was performed while I.D. held an NRC Research Associateship award at United States Army Research Laboratory. A.K.S.{\textquoteright}s research was partly supported by funding from the European Research Council (ERC) under the European Union{\textquoteright}s Horizon 2020 research and innovation program (grant agreement no. 834709 BIOSPACE). Funding Information: We thank Maxar Technologies (formerly DigitalGlobe) for providing very-fine-resolution commercial satellite images through the NextView Imagery End User License Agreement of the US National Geospatial-Intelligence Agency. We also thank the Army Departmental Requirements Office for acquiring the Maxar images for this work. We are grateful to Dr. Juan Lavista Ferres the Chief Scientist at the Microsoft AI for Good Research Lab for his support in our application for the Microsoft AI for Earth grant. Thanks also to Dr. Caleb Robinson and Dr. Anthony Ortiz for their assistance in transferring and storing satellite image data. We also wish to thank Ralph Mettinkhof for his support in using the Virtual Research Environment at the University of Twente. We are grateful to Dr. Olga Isupova and Dr. Xiaowen Dong for providing valuable comments on an early draft of this manuscript. Many thanks to Zeyu Xu for his support in testing YOLOv4 algorithm. L.F.H. acknowledges the support of the Ohrstrom Family Foundation. This collaboration was supported, in part, by grant no. 00138000039 from the Microsoft{\textquoteright}s AI for Earth Program (https://www.microsoft.com/en-us/ai/ai-for-earth). Z.W.{\textquoteright}s research was supported, in part, by funding from the Department of Natural Resources at the Faculty of Geo-Information Science and Earth Observation (ITC) of the University of Twente. This research was performed while I.D. held an NRC Research Associateship award at United States Army Research Laboratory. A.K.S.{\textquoteright}s research was partly supported by funding from the European Research Council (ERC) under the European Union{\textquoteright}s Horizon 2020 research and innovation program (grant agreement no. 834709 BIOSPACE). Funding Information: We thank Maxar Technologies (formerly DigitalGlobe) for providing very-fine-resolution commercial satellite images through the NextView Imagery End User License Agreement of the US National Geospatial-Intelligence Agency. We also thank the Army Departmental Requirements Office for acquiring the Maxar images for this work. We are grateful to Dr. Juan Lavista Ferres the Chief Scientist at the Microsoft AI for Good Research Lab for his support in our application for the Microsoft AI for Earth grant. Thanks also to Dr. Caleb Robinson and Dr. Anthony Ortiz for their assistance in transferring and storing satellite image data. We also wish to thank Ralph Mettinkhof for his support in using the Virtual Research Environment at the University of Twente. We are grateful to Dr. Olga Isupova and Dr. Xiaowen Dong for providing valuable comments on an early draft of this manuscript. Many thanks to Zeyu Xu for his support in testing YOLOv4 algorithm. L.F.H. acknowledges the support of the Ohrstrom Family Foundation. Publisher Copyright: {\textcopyright} 2023, The Author(s). Financial transaction number: 2500072683 ",
year = "2023",
month = dec,
doi = "10.1038/s41467-023-38901-y",
language = "English",
volume = "14",
journal = "Nature communications",
issn = "2041-1723",
publisher = "Nature Publishing Group",
}