Wildlife Insights: A platform to understand and share biodiversity information from in-situ passive sensors

Abstract:

Biodiversity is an essential component of all ecosystems. Most places on the globe do not have local, timely and accurate in-situ information on which species occur or how their populations are changing over time and space. Passive sensors (camera traps, acoustic sensors, eDNA analyzers) have become a popular way to collect in-situ species data. However, data collection has increased at a much faster rate than the development of tools to manage, process and analyze these data. Without these tools, biodiversity managers and other stakeholders are overwhelmed with all this information to effectively manage, understand and monitor species populations. Among passive sensors, camera traps are one of the most popular types currently deployed. We identify four barriers that are hindering the widespread use of camera trap data for science and conservation. We propose specific solutions to remove these barriers via a modern technology platform called Wildlife Insights. We present specific technology solutions, an architecture for this platform and describe its main components, including novel integration of deep learning models for species classification to identify images without animals and predict species in images. We recognize and discuss the potential risks of publishing shared biodiversity data and a framework to mitigate those risks. Finally, we discuss a strategy to ensure platforms like Wildlife Insights are sustainable and have an enduring impact on the conservation of wildlife.