Job description

The project aims to develop key technologies required for fast, accurate, and automated detection of internal anomalies and defects directly from the imaging sensor data available in the high-throughput industrial workflow. To facilitate high-throughput inspection, decisions about the presence of defects or anomalies must be taken directly based on the limited, raw data measured by an in-line sensor system. Direct interpretation of such sensor data is challenging, and modern automated analysis tools based on machine learning require annotated training data. Two key problems must be tackled to facilitate such a high-throughput, data-driven workflow: 1. The lack of available ground truth data to support the use of supervised machine learning techniques, and 2. The limited signal information available in the high-speed sensor data, which may not be sufficient for certain types of internal anomalies. During the project, data-driven physics-aware work...