Characterisation of 3D microstructures by distance transform and deep learning

L'École nationale supérieure des mines de Saint-Étienne, aussi dénommée Mines Saint-Étienne, est l'une des 204 écoles d'ingénieurs françaises accréditées au 1er septembre 2020 à délivrer un diplôme d'ingénieur. Elle figure parmi les plus prestigieuses écoles d'ingénieurs de France et se distingue également dans les classements internationaux.

Créée en 1816, elle est membre de l'Institut Mines-Télécom, sous tutelle du ministère chargé de l'industrie (aujourd'hui, ministère de l'Économie, de l'industrie et du numérique).

École d'ingénieurs de haut niveau, nous formons des ingénieurs civils des MINES, des docteurs et ingénieurs spécialisés, alliant sciences et management, dont les compétences sont recherchées dans le milieu industriel.

Nous avons structuré notre recherche en six grands domaines pour répondre aux préoccupations sociétales et industrielles actuelles.

Internship
In the search for solutions to climate change, the issue of energy is of central importance. Indeed, energy conversion and storage are of paramount interest for the future. One of the challenges related
to these objectives is to increase our knowledge of the materials' behaviour, particularly in relation to energy conversion and storage. One method of achieving this objective is to develop innovative
numerical methodologies for extracting the structural features of materials, with a view to improving their efficiency.
The structure of materials is connected to their physicochemical properties, which are the key features to develop more efficient materials. Consequently, investigating the structure or morphology is of paramount interest for a broad range of applications. Image processing represents a complementary approach to experimental characterisation of materials and is a promising avenue for finding solutions to climate change.
The internship will focus on numerical methodologies based on image processing and more specifically distance transforms, to propose innovative solutions to extract critical structural features from images of materials. Furthermore, the potential of deep learning in enhancing the characterisation of microstructures in materials will be explored. The objective is to propose novel algorithms and to compare them to existing methodologies.

Profile
Successful candidates will be final-year engineering students or Master's degree students specialising in image processing, applied mathematics, bioinformatics or related fields. They should have
knowledge of mathematics and computer programming in Python or another scientific programming language.