IMITATION

Image interpretation by taught simulation using neural networks

Preliminary results of AI segmentation with simulated defects

About the project

IMITATION is an internally funded project that aims at developing AI based 3D image analysis tools tailored to DTU Energy’s characterization techniques and technologies.
Within IMITATION we apply a simulation approach to training deep neural networks. Instead of manually annotating the ground truth we seek to simulate the training data using rapid simulation approaches covering the three main imaging probes, x-rays, electrons and neutrons. 

Type of project

Internal project

People from the SAM section involved

Peter Stanley Jørgensen

Contact

Peter Stanley Jørgensen

Peter Stanley Jørgensen Senior Researcher Department of Energy Conversion and Storage Mobile: +45 93511607