Digital animal conservation

Managing animal populations and understand behavior are important topics in conservation and, more broadly, in animal ecology. The usage of digital technologies (from the camera trap to the drone to the satellite) can help rangers and scientists to accelerate surveys and scale findings beyond a localized small region.

The ECEO lab is active in several projects aiming at digitalizing animal conservation with image processing and machine learning. We are active in Namibia (Kuzikus reserve) and Eastern Africa, where we develop technology for semi-automatic censuses of ungulates and migratory birds, respectively.


  • Tuia, Kellenberger, Beery, Costelloe et al. (2022). Perspectives in machine learning for wildlife conservation. Nature Communications, 13. (paper)
  • Kellenberger, Veen, Folmer, Tuia (2021). 21 000 birds in 4.5 h: efficient large‐scale seabird detection with machine learning. Remote Sensing in Ecology and Conservation. (infoscience)
  • Kellenberger, Marcos, Lobry, Tuia (2019). Half a percent of labels is enough: efficient animal detection in UAV imagery using deep CNN and active learning, IEEE Transactions on Geoscience and Remote Sensing, 57: 9524-9533 (arxiv).
  • Kellenberger, Marcos, Tuia (2018). Detecting mammals in UAV images: Best practices to address a substantially unbalanced dataset with deep learning, Remote Sensing of Environment, 216: 139-153 (arxiv)
We develop deep learning based models to detect wildlife from drone imagery (from Kellenberger et al., 2019)