Master Thesis
Status: open
Supervisor: Martin Kampel
Automated Material and Damage Detection in Buildings using Computer Vision
The goal of this thesis is to develop and evaluate computer vision and machine learning methods for the digital inventory of buildings. The focus is on the automated recognition of building materials as well as the detection and assessment of typical structural damage (e.g., cracks, spalling, facade damage) based on image data.
The data used consists of 2D images (e.g., photos, 360° images) as well as optional 3D data (e.g., point clouds or laser scans). A particular emphasis is placed on investigating the extent to which the combination of 2D and 3D data improves the accuracy and robustness of detection.
The work includes an analysis of the state of the art, the preparation of suitable data, the development and implementation of detection algorithms, and their systematic evaluation. Finally, the results will be documented and assessed in terms of their practical application potential in the construction industry.
This thesis is carried out in collaboration with STRABAG AG. In the case of a successful collaboration, a research grant will be provided.
Tasks
- Conduct a literature review on computer vision methods for material and damage detection
- Analyse and prepare 2D and optional 3D building data
- Develop and implement models for material recognition and
- damage detection (e.g. cracks, spalling, facade defects)
- Integrate and evaluate multimodal (2D + 3D) approaches
- Assess performance in terms of accuracy and robustness
- Document results and discuss practical applicability
Contact
Martin Kampel (martin.kampel@tuwien.ac.at)