Areas of interest
Machine Learning is a transverse field that aims at extracting information from data in order to make decisions and predictions. I currently work to developp new architectures ( ShaResNet ), apply them various application fields from classification to cartography to 3D scene understanding to space wearther prediction.
Earth Observation data is now massively available through satellite programs like Copernicus. This allows the emergence of new algorithms and techniques to cartography, analysis or restoration of multiple sensors active (SAR) or passive (optical) images. I am particularly interested in dealing with partially annoted data to create robust machine learning techniques abble to diggest multi-temporal data.
Computational Geometry and reconstruction was the core area of my thesis. Understanding 3D scenes, and reconstructing abstracted surfaces (BIM for buildings) is a major interest to me. I work on outdoor scenes ( Semantic3D ) as well as indoor data for robotics ( SUNRGB-D ).
DeLTA, deep machine learning for aerospace applications is a research project at ONERA. Among its objectives are the development and the promotion of innovative machine learning based approaches for aerospace applications.
Medusa for Big Data in Earth Observation. Multidate Earth observation Data for Urban Sprawl Aftercare. The new context of big data for Earth observation, particularly the field of urban development and smart cities, has led Onera to launch the research project Medusa.
Medusa project is designed to bring together and promote processing of remote sensing images in the current context of big data