Deep Learning-Based Automatic Classification for Aquatic Ecosystem Bioindication through Image Analysis – BIOINDIC-IA
PI : Martin Laviale (LIEC)
PI@CEREGE : Thibault de Garidel
Aquatic ecosystems are subject to numerous anthropogenic pressures. This calls for the development of innovative ecological diagnostic tools to support robust management measures. Current tools can assess ecological status based on taxonomic characteristics (i.e., species lists) and/or functional traits (i.e., individual phenotypic attributes) of a given group of organisms. However, these approaches often rely on morphological criteria that can be difficult to characterize. In this context, the objective of BIOINDIC-IA is to enhance the biomonitoring of aquatic ecosystems by using artificial intelligence to automatically analyze taxonomy (species) and morphological traits from microscopy images. The project focuses on two model organisms: benthic freshwater diatoms, recognized as bioindicators in the European Water Framework Directive, and marine benthic foraminifera, which are rarely included in ecological assessments despite their acknowledged potential for coastal water biomonitoring.