Résumé
At a time of massive deployment of cameras on the territory and the development of connected infrastructures, the OCAPI project aims to extend the uses of cameras deployed on existing infrastructures to the management of biodiversity in the vicinity of these infrastructures within the framework of the management of risks associated with collisions and the monitoring of environmental measures effectiveness. To this end, the OCAPI project contributed to the development of deep learning algorithms for the automatic identification of the large mammals most frequently encountered in accidents with vehicles (wild boar, roe deer and deer). This development was made possible by the deployment of an image annotation platform benefiting from the image contributions (photographs and videos) of major partners such as Vinci autoroute, the Lynx Network, or the Regional and Departmental Federations of Hunters. The recognition data produced by the automatic recognition systems are then processed using species distribution models (machine learning) to produce maps of the risks associated with the presence of these large mammals. The entire processing and analysis chain was designed and tested in a virtual environment based on a real case in the OCAPI project. Although developed with the main objective of adaptive management of the risk of collisions between vehicles and large mammals, the monitoring and analysis scheme for wildlife in the vicinity of infrastructures can be applied to other operational frameworks such as the monitoring of environmental measures (measures resulting from the implementation of the mitigation hierarchy, voluntary management measures, etc.) on the scale of infrastructures, but also to other contexts such as hunting management or the monitoring of biodiversity conservation programmes. Thus, the deployment of the solution proposed in the OCAPI project on 1) existing infrastructures, then 2) on infrastructure networks, is likely to provide territories with large-scale biodiversity monitoring systems based on existing infrastructures.