Selected annotated instance segmentation sub-volumes from a large scale CT data-set of a historic aircraft

Area of Science:

• Aerospace Engineering
• Digital Heritage
• Computer Vision

Background:

• The Messerschmitt Me 163 Komet, a World War II fighter aircraft, is housed in the Deutsches Museum, Munich.
• Computed Tomography (CT) scanning offers a non-destructive method to analyze historical artifacts like aircraft.
• Detailed analysis of CT data requires advanced segmentation techniques, which are currently lacking for large-scale aerospace datasets.

Purpose of the Study:

• To establish an initial interactive data annotation process for segmenting large-scale CT scan data of the Me 163 aircraft.
• To lay the groundwork for developing automated or semi-automated segmentation tools for similar datasets.
• To explore potential applications in digital heritage, non-destructive testing, and machine learning.

Main Methods:

• Acquisition of a complete CT scan of the Me 163 using an industrial CT scanner.
• Development and implementation of an interactive data annotation process.
• Annotation of seven 512x512x512 voxel sub-volumes of the CT dataset.

Main Results:

• Successfully acquired comprehensive CT data of the Me 163.
• Established a functional interactive data annotation workflow.
• Generated annotated sub-volumes suitable for further research and development.

Conclusions:

• The interactive annotation process is a crucial first step towards automated segmentation of large-scale aerospace CT data.
• The annotated data holds significant potential for digital preservation, structural analysis, and machine learning model training.
• Challenges in data interpretation and handling were identified, guiding future research directions.