Institute for Energy Transition
dedicated to Marine Renewable Energies
Know more

SEMMACAPE

Monitoring and study of marine megafauna by automatic characterisation in wind farms

Duration: 3 years

Context

The development of oofshore renewable energy (ORE) is booming in France. Between 2011 and 2017, the French government has launched 5 calls for tenders and expressions of interest for a total of 7 wind farms installed, 4 wind pilot farms and 2 tidal pilot farms. Over the next 5 years, 5 additional calls for tenders are scheduled, which could double the installed capacity at sea. According to the Environmental Code, project developers must produce environmental impact studies, particularly on marine megafauna (birds, mammals, turtles, big fish...).
 
The analysis of the impacts of an ORE project generally requires aerial observations of marine megafauna in order to better characterise the frequentation of species in the proposed areas. This includes monitoring during the construction, operation and decommissioning phases, a total of about 30 years.
 
These observations are classically based on aerial overflights by specialised naturalist observers. However, in the age of big data, recent scientific and technological developments offer new prospects for radically improving the cost-effectiveness of such monitoring.

Objective

Démontrer la pertinence des solutions logicielles de traitement et d’analyse des photographies aériennes pour assurer le recensement automatisé de la mégafaune marine.

Expected results

  • Demonstration of the feasibility of a fully automated image analysis solution for aerial tracking of marine megafauna at the scale of an MORE project area.
  • Proposal of a software solution adapted to the monitoring of the marine megafauna present in metropolitan France, and more particularly in the areas of future French and European offshore wind farms.
  • Guaranteeing the technical feasibility of aerial monitoring after the installation of offshore wind turbines, thanks to a combination of technologies allowing observations at an altitude imposed by safety constraints (300 m and more) and freeing the massive recourse to naturalist experts for their interpretation.

Scientific contents

  • Carrying out an aerial observation campaign of the megafauna (standardized visual method and very high-resolution digital image acquisition system) integrating the seasonal variability of species and environmental condition.
  • Development and qualification of 2 types of algorithms for the automated processing of aerial images, for the identification and classification of animals: 
    • Detection by deep neural network known as end-to-end, from the global image to the enclosing boxes in a direct way;
    • Anomaly detection by unsupervised deep learning.
  • Evaluation of the performance of each of the detection methods tested on the basis of indicators broken down by species or groups of species, as well as environmental conditions.

Partners

Coordinator: Université de Bretagne Sud - UMR IRISA

SEMMACAPE project partners

This project benefits from an ADEME grant under the "Energies Durables" call for research projects (2018-2019).

Logo Ademe.png