@@ -7,23 +7,30 @@ The goal of this task is to achieve semi-automatic (supervised) or automatic (un
...
@@ -7,23 +7,30 @@ The goal of this task is to achieve semi-automatic (supervised) or automatic (un
## Targets:
## Targets:
-Classify the building's LiDAR point cloud data into the following classes: wall, floor, ceiling, window, door, column, beam, stairs/ramps, roof, furniture, opening/passageway, and others. Alternatively, classify the exterior urban area's LiDAR data into the following classes: building, ground, road, sidewalk, signage, vegetation, vehicles, water, and others.
The relevance of point cloud data for capturing 3D environmental data in various scenarios, such as robotics, computer vision, autonomous driving, building and city digital twin models, is increasing rapidly. Despite its ability to accurately describe the geometry or shape of surface elements, it is still unstructured data, meaning it is only a group of points represented in 3D space with x, y, and z coordinates, as well as other features, such as color (R, G, B). Therefore, it is essential to provide meaning and structure to this data by segmenting or classifying the points into groups with specific semantic labels.
Semantic segmentation of point clouds is time-consuming, especially for large datasets.
The goal of this task is to achieve semi-automatic (supervised) or automatic (unsupervised) semantic segmentation of the LiDAR point cloud dataset of one of the stories of the HFT building (Bau 4) and/or its exterior urban areas (streets, signs, sidewalks, etc.).
## Targets:
-Visualize the data as separate, labeled point clouds (LAS, PTS, XYZ formats).
-Visualize the data as separate, labeled point clouds (LAS, PTS, XYZ formats).
-Visualize in map applications.
## Data Needs:
## Data Needs:
-[LiDAR scans of the HFT campus](https://hftstuttgartde.sharepoint.com/:f:/s/Informatic-weekend-2025/EmJmSuv8yL1Pq6lqgFEeh2oBdKe6v6bHmrcyL3zCeRhwpw?e=IQWEcL).
- LiDAR scans of the HFT campus.
- DEM/DTM alternative if LiDAR is unavailable.
- DEM/DTM alternative if LiDAR is unavailable.
- Reference: [HFT Data Wiki](https://wiki.hft-stuttgart.de/pages/viewpage.action?pageId=121441587&spaceKey=STRATINPUT&title=Datengrundlage%2B3D-Geb%C3%A4udemodell%2BHFT).
- Reference: [HFT Data Wiki] (https://wiki.hft-stuttgart.de/pages/viewpage.action?pageId=121441587&spaceKey=STRATINPUT&title=Datengrundlage%2B3D-Geb%C3%A4udemodell%2BHFT).
## Acceptance Criteria:
## Acceptance Criteria:
- Reasonable segmentation quality
- Reasonable segmentation quality
- A README with reproducible steps (workflow and code) for application to the rest of the HFT buildings for semantic segmentation.
- A README with reproducible steps (workflow and code) for application to the rest of the HFT buildings for semantic segmentation.
