Toward automatic identification of transformation products in Steels on EBSD maps
1 year contract-Full time
LEM3 – Université de Lorraine – CNRS - METZ - France
In collaboration with Industeel-ArcelorMittal, the aim is to develop a database of EBSD microstructures as reference metallurgical states and to use it to train an algorithm for recognition of transformation products in Steels.
The postdoctoral work has several tasks:
- Acquisition of EBSD maps of metallurgical states provided by ArcelorMittal
- Applying codes developed at LEM3 to identify the different phases (Labeling)
- Literature review on Machine Learning and Convolutional Neuronal Networks to determine the most appropriate approaches to the problem.
- Implementing an algorithm for phase recognition in Steels based on EBSD maps, using existing libraries and developing partnerships with specialized teams in the field if and developing partnerships with specialized teams in the field if necessary.