Master Thesis NCC LiDAR to BIM – Segmentation and CAD Matching in Construction Environments Using Computer Vision Segmentation Models

Background
Automated Scan-to-BIM is a central part of the digitalization of the construction industry. However practical implementation is hindered by difficulties in both available technology and the lack of high-quality datasets in construction environments. Research Zbirovsky and Nezerka 2025 Mahmoud et al. 2024 Campagnolo et al. 2023 shows that deep learning-based methods can increase the degree of automation. Despite promising results robust pipelines that connect actual LiDAR scans to fully modeled CAD/BIM with sufficient precision are lacking.

The projects goal is to match identified objects in LiDAR scans with corresponding CAD/BIM objects (walls columns and slabs) and quantitatively evaluate the methods reliability. The results should provide an assessment of whether the method is practically usable

The project should result in a reproducible baseline on 3DSES (targeted >0.6-0.7 IoU for wall/column) a validated matching pipeline with NCCs LiDAR scans with results and an assessment of the methods practical usability in construction projects.

The project initially uses the 3DSES dataset Blanchard et al. 2024 which contains both CAD models and actual LiDAR scans of indoor environments. The project is based on point cloud models such as RandLA-Net where pretrained weights on large indoor datasets are advantageously used and transferred to 3DSES. Segmentation results are used to create objects that are then matched against the CAD/BIM model. If the desired accuracy is achieved the project proceeds to evaluate the methods accuracy on NCCs own scans and BIM models.

Your profile and Application
This Master Thesis is suitable for 2 students with an interest in applying advanced AI and data science techniques to real-world problems in the construction industry.

You are at the end of your studies in computer/data science/construction engineering or closely related and about to start your Master Thesis work for 30 HP. You have good programming skills in languages such as Python with experience in AI and machine learning frameworks or scikit-learn and/or experience in data analysis and visualization tools to handle and interpret complex your studies you should have completed coursework in AI and machine learning such as Introduction to Artificial Intelligence Machine Learning Data Mining and Analytics. A solid understanding of mathematics including probability statistics and linear algebra would be beneficial for developing and implementing AI algorithms.

Familiarity with construction CAD and BIM would be beneficial but not strictly required.

We provide the support and guidance you need to translate your theoretical knowledge into practical solutions. Join us and become a driving force behind NCCs technological advancements.

Location
Your industrial supervisor is placed in Göteborg but you can be a student at any Swedish university to apply

Time period
Spring 2026

Language
Swedish or English. This master thesis does not require fluency in Swedish.

Contact person

Last application day
November 25 2025.  Applications will be reviewed on a rolling basis and the call may close early if the position is filled.

References
Blanchard C. Vacelet T. Nahangi M. and Hammad A. (2024). 3DSES: 3D scan to existing structure dataset (lidar and cad models). Dataset containing paired CAD models and LiDAR scans for Scan-to-BIM benchmarking.

Campagnolo D. Camuffo E. Michieli U. Borin P. Milani S. and Giordano A. (2023). Fully automated scan-to-bim via point cloud instance 2023 IEEE International Conference on Image Processing (ICIP). IEEE.

Mahmoud M. Chen W. Yang Y. and Li Y. (2024). Automated BIM generation for large-scale indoor complex environments based on deep learning. Automation in Construction 162:105376.

Zbirovsky S. and Nezerka V. (2025). Cloud2bim: An open-source automatic pipeline for efficient conversion of large-scale point clouds into IFC format. arXiv preprint arXiv:. Preprint submitted to Automation in Construction; version v2 18 Mar 2025.

Ytterligare information :

Build with us - together with the best colleagues in the industry.
 
At NCC we are driven to continue developing whether its our construction projects or our personal expertise. Every day we drive the industrys most exciting and complex construction projects forward in close cooperation between our employees partners and customers. With us you are a significant employee who contributes to the landmarks that define our cities and communities. Here our employees take an active role in a corporate culture based on stable values and behaviors for a safe and secure workplace. Our collective expertise and diverse backgrounds make a difference. 
 
NCC - one of the leading companies in the Nordics. 

Remote Work :

Yes

Employment Type :

Intern