Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Learning to clean heritage point clouds
Laser scanning technology enables accurate distance measurements between a scanner and the environment producing a grid of points in 3D space. Many laser scans can be registered together to produce a “point cloud”. This is commonly used in the Cultural Heritage domain to create highly detailed 3D mo...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English English |
| Published: |
Department of Computer Science
2026
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613280060571648 |
|---|---|
| access_status_str | Open Access |
| author | Hayward, Luc |
| author2 | Marais, Patrick |
| author_browse | Hayward, Luc Marais, Patrick |
| author_facet | Marais, Patrick Hayward, Luc |
| author_sort | Hayward, Luc |
| collection | Thesis |
| description | Laser scanning technology enables accurate distance measurements between a scanner and the environment producing a grid of points in 3D space. Many laser scans can be registered together to produce a “point cloud”. This is commonly used in the Cultural Heritage domain to create highly detailed 3D models. However, the raw point cloud often contains unwanted details and artefacts which require manual removal through a time-consuming process called point cloud cleaning. Whilst previous works have shown promising results, the application of deep learning in the cultural heritage domain has not been fully explored. In this study, we investigate the use of deep learning models for the binary segmentation of semantically varied cultural heritage sites using a few-shot fine-tuning approach. By relying only on learned geomet- ric information without additional point attributes such as colour, we demonstrate the e↵ective application of modern deep learning approaches in this domain. We present a method in which the user is presented with an unlabelled scan and asked to label a percentage between 2.5-50% of the scene. This labelled data is used to train the model and predict labels for the remainder of the scene. We compare the performance of three di↵erent deep learning models (Pointnet++, KPConv, and Point Transformer) against a baseline Random Forest model. We consider factors such as speed, pretraining, and the amount of hand labelling required. Our results show that minimal human labelling is needed to provide su�cient data for modern deep learning approaches and simple scenes are e�ciently labelled using Random Forests. We achieve up to 90% mean Intersection over Union on the remaining points, with as little as 5% of the scene labelled for training, in under two hours on consumer hardware. This translates to a reduction in manual labelling e↵ort of up to 90%. Among the models tested, KPConv reliably produces good results with minimal need for human labelling. The remaining models are slower to train and required more human labelling, while the Random Forest is e↵ective only on simple scans, where a machine learning model would be ine�cient to train. Our work demonstrates the potential of deep learning methods for label-e�cient and accurate point cloud cleaning in the cultural heritage domain, reducing the time spent on manual cleaning. |
| format | Thesis |
| id | oai:open.uct.ac.za:11427/43116 |
| institution | University of Cape Town (South Africa) |
| language | English eng |
| last_indexed | 2026-06-10T12:33:37.862Z |
| license_str | Not specified — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from UCTD — University of Cape Town Open Access Repository |
| publishDate | 2026 |
| publishDateRange | 2026 |
| publishDateSort | 2026 |
| publisher | Department of Computer Science |
| publisherStr | Department of Computer Science |
| record_format | dspace |
| source_str | UCTD — University of Cape Town Open Access Repository |
| spelling | oai:open.uct.ac.za:11427/43116 Learning to clean heritage point clouds Hayward, Luc Marais, Patrick Computer Science Cultural Heritage 3D space Laser scanning technology enables accurate distance measurements between a scanner and the environment producing a grid of points in 3D space. Many laser scans can be registered together to produce a “point cloud”. This is commonly used in the Cultural Heritage domain to create highly detailed 3D models. However, the raw point cloud often contains unwanted details and artefacts which require manual removal through a time-consuming process called point cloud cleaning. Whilst previous works have shown promising results, the application of deep learning in the cultural heritage domain has not been fully explored. In this study, we investigate the use of deep learning models for the binary segmentation of semantically varied cultural heritage sites using a few-shot fine-tuning approach. By relying only on learned geomet- ric information without additional point attributes such as colour, we demonstrate the e↵ective application of modern deep learning approaches in this domain. We present a method in which the user is presented with an unlabelled scan and asked to label a percentage between 2.5-50% of the scene. This labelled data is used to train the model and predict labels for the remainder of the scene. We compare the performance of three di↵erent deep learning models (Pointnet++, KPConv, and Point Transformer) against a baseline Random Forest model. We consider factors such as speed, pretraining, and the amount of hand labelling required. Our results show that minimal human labelling is needed to provide su�cient data for modern deep learning approaches and simple scenes are e�ciently labelled using Random Forests. We achieve up to 90% mean Intersection over Union on the remaining points, with as little as 5% of the scene labelled for training, in under two hours on consumer hardware. This translates to a reduction in manual labelling e↵ort of up to 90%. Among the models tested, KPConv reliably produces good results with minimal need for human labelling. The remaining models are slower to train and required more human labelling, while the Random Forest is e↵ective only on simple scans, where a machine learning model would be ine�cient to train. Our work demonstrates the potential of deep learning methods for label-e�cient and accurate point cloud cleaning in the cultural heritage domain, reducing the time spent on manual cleaning. 2026-04-22T11:24:21Z 2026-04-22T11:24:21Z 2023 2026-04-22T08:39:17Z Thesis / Dissertation Masters Masters http://hdl.handle.net/11427/43116 en eng application/pdf Department of Computer Science Faculty of Science University of Cape Town |
| spellingShingle | Computer Science Cultural Heritage 3D space Hayward, Luc Learning to clean heritage point clouds |
| thesis_degree_str | Master's |
| title | Learning to clean heritage point clouds |
| title_full | Learning to clean heritage point clouds |
| title_fullStr | Learning to clean heritage point clouds |
| title_full_unstemmed | Learning to clean heritage point clouds |
| title_short | Learning to clean heritage point clouds |
| title_sort | learning to clean heritage point clouds |
| topic | Computer Science Cultural Heritage 3D space |
| url | http://hdl.handle.net/11427/43116 |
| work_keys_str_mv | AT haywardluc learningtocleanheritagepointclouds |