Deep learning for automatic image-based biomonitoring of aquatic ecosystems – BIOINDIC-IA

Aquatic ecosystems are facing many anthropogenic pressures, highlighting the urgent need for developing innovative ecological diagnostic tools supporting the selection of robust management responses. Current diagnostic tools are able to estimate the ecological status from taxonomy-based (i.e. list of species) and/or trait-based (i.e. list of individual phenotypic attributes) characteristics of a given key biological compartment. Yet, this often relies on morphological criteria that are sometimes difficult to characterize. In this context, the goal of BIOINDIC-IA is to improve trait-based biomonitoring of aquatic ecosystems by using artificial intelligence for the automatic processing of microscope images in order to analyse the taxonomy and morphological traits. The project will focus on two model organisms: freshwater benthic diatoms, a relevant biological compartment already implemented within the European Water Framework Directive, and marine benthic foraminifera, which are seldomly included in ecological assessments despite their recognized potential for biomonitoring of coastal waters.