SET04 - AISRIS: an automatic ichtyoplancton sampler, recorder, identifier and scaler

Obtaining a good knowledge of the larval life of each species is crucial, not only to study them (e.g. access to abyssal fish in adulthood) but also to protect them (e.g. protection of source areas). But currently, destructive capture methods by net or light trap allow only a very small number of larvae to be captured, rarely allowing for spatio-temporal monitoring of species. 

Our project aims to individually capture fish larvae, in order to be able, on one hand, to photograph them in high quality and in color for precise automatic identification by deep-learning (species and life stage). On other hand, it will also aim to identify them genetically by metabarcoding in order to capture not only the environmental DNA of the larvae but also of their symbionts and parasites. The individualization of the capture will also make it possible to capture the location, the time and all the associated environmental parameters.

We are currently building a small scale unit to demonstrate individual larvae capture during trawling. We will then develop an automatic larvae image recognition system.

A software to discriminate fish larvae from plankton using deep learning.

Software and electronic engineer
Data analyst
Marine biologist