Student Assistant for Image Analysis with Machine Learning (all genders)

As part of a research project, a process for damage detection and analysis on elastomer samples is being developed. The goal of the project is to create a model trained by machine learning that automatically performs damage analysis based on imaging and additional information (e.g., Shore hardness) and independently assesses the causes of damage. Preliminary student work has created an initial machine learning pipeline for object recognition and image data analysis, which will be adapted over the course of the project and supplemented with explainable AI methods. Further work on synthetic data generation is planned. Literature research and comparisons to Machine learning methods for object detection and classification as well as for the integration of non-image-based information into machine learning pipelines Methods for synthetic image generation xAI methods in the context of object recognition and image classification (saliency maps, etc.) Adaptation of an existing machine learning pipeline for object recognition and classification of damage patterns on elastomers Integration of additional (non-image-based) information into the machine learning pipeline Integration of xAI processes Documentation and presentation of the procedure and results

The Fraunhofer-Gesellschaft ( www.fraunhofer.de ) currently operates 76 institutes and research facilities in Germany and is one of the leading organizations for application-oriented research. Approximately 32,000 employees generate an annual research volume of 3.4 billion euros.