Insect populations are currently undergoing a massive decline of around 10% per decade. Given the vital importance of this group in many ecosystem functions (e.g. pollination), there is an urgent need to better understand the causes of this decline, and above all to identify conservation methods to limit it. Traditional entomology research is ill-equipped to carry out non-invasive (i.e. no capture or killing), high-frequency spatio-temporal studies of the thousands of species that characterize this hyper-diverse class of the animal kingdom - a prerequisite for assessing the effectiveness of any restoration or protection measures.

Participatory sciences offer a fantastic opportunity to multiply entomological surveys, but they are limited by the difficulty of identifying the many, often cryptic, species of arthropods (e.g. SPIPOLL). The use of artificial intelligence, in particular deep learning techniques, applied to macrophotographs, makes it possible to overcome these limitations (e.g. iNaturalist).

In the INSECTA project, we propose to combine a standardized arthropod inventory protocol (macrophotographic transect) adaptable to all types of habitat and taxonomic groups, with semi-automated identification of the images collected.