Annotating, detecting and counting wild animals has never been easier
AIDE in a nutshell © 2020 B. Kellenberger
Researchers of the ECEO laboratory in Sion collaborated with Microsoft AI4Earth to create an open source software helping ecologists to detect and count animals with drones.
The project, started during an internship at Microsoft AI4Earth in 2019, has led to a fully functional, open source software suite for animal ecology. The software, name AIDE for "Annotation Interface for Data‐driven Ecology" is based on deep learning, a type of artificial intelligence that can be used to identify animals in images taken from the ground, drones or satellites. It allows ecologists to upload their imaging campaigns, annotate them smartly and run modern deep learning models to detect and count animals without coding or advanced computer science background. AIDE also ses AI to prioritize images with high chances to contain animals and allows sharing models between campaigns.
AIDE is open for the community and already being used in several research institutes worldwide.
The project is an outcome of a research internship at Microsoft AI4Earth by Dr. Kellenberger
B. Kellenberger, D. Tuia, and D. Morris. AIDE: Accelerating image-based ecological surveys with artificial intelligence. Methods Ecol. Evol., 11(12):1716–1727, 2020.