Building a scalable wildlife monitoring system by integrating remote camera sampling and artificial intelligence with Essential Biodiversity Variables (EBVs)
We aim to develop a cutting-edge coordinated wildlife monitoring system underpinned by the EBVs. We will combine camera trapping, citizen science and artificial intelligence for the automated production of species populations and community structure EBVs.
Why?
Why a monitoring system?
The increase in anthropogenic activities has caused significant changes in the composition and structure of ecosystems, directly affecting their functioning. Tackling the ongoing biodiversity crisis requires accurate knowledge of species and animal communities, and their dynamics and population trends.
How can we do it?
In recent years, camera trapping, combined with novel citizen science platforms and machine learning tools, has revolutionized wildlife surveys and holds promise for implementing cost-effective monitoring programs.
For what?
The EBVs are conceptualized to be as the minimum set of biological state variables that are needed to monitor biodiversity change. EBVs enable modeling and early detection of changes in biodiversity and provide a framework for promoting interoperability among monitoring programs.
.
Project goals
Work Packages
Work Package 1: Harmonisation of image collation among study areas
This WP deals with the choice of study designs, harmonisation of data collection, and the development of a field protocol as well as a web application for online image collection and processing.
Work Package 2: Developing an automated AI system for species identification and individual counts in biogeographical regions of Europe
WP2 will focus on providing the infrastructure for automatic image conversion to analysable data and analysis outputs, storage, and online access for registered users. We will use computer vision to automatically identify the species and the number of individuals from images. Extracting this information is a three-step process, each represented by a single task.
Work Package 3: Automated estimation of Essential Biodiversity Variables and biodiversity change indicators
In this WP we will generate standardised Species Populations and Community EBVs workflows, and derived biodiversity change indicators. We will pilot a standardised production of modelled EBVs datasets from harmonised EBVs-ready data collections of species camera records. The workflows will result in two different types of modelled EBVs products. The first are EBVs products measuring species occurrence probabilities, relative abundances, and derived community-level metrics directly at the camera locations, accounting for imperfect species detectability and classification. The second are downstream EBVs products using model-based data integration with environmental covariates.
Work Package 4: Developing a European stakeholder-led biodiversity monitoring program - upscaling and transferability
This WP conveys different resources to effectively disseminate the results, mobilise the data to global infrastructures and upscale the automated monitoring system for biodiversity to other areas and regions in collaboration with different stakeholders. WildINTEL aims at ensuring the long-term maintenance and functioning of the monitoring system.