2026/01/12

Previous weeks

• DAIC is out of maintenance and I could check that I have access.
• Idea: use the directions of edges of the existing footprints/roofprints as likely first good guesses
• Wrote the “Project Proposal”
• Started working with CGAL on extending roof planes and computing distance from point cloud to buildings

Discussion

• We discussed things to add/modify in the “Project Proposal”
• If datasets are too dense, it could be possible to reduce the density by simply picking every n-th point and the distribution will be fine if the order is the original one
• We could produce fake ground-truth data by simulating LiDAR on top of LoD 2.3 buildings, with tools like helios or equivalents at the IGN, which could be useful for results assessment

Next week

• Things to add to the “Project Proposal”:
  • Roof overhangs are the distinction between LoD 2.2 and 2.3
  • Add images and tables for the datasets mentioned in the literature review
  • Modify the research question:
    • Talk about characteristics instead of situation
    • Be more progressive (less binary)
    • Mention outlines as a possibility, and add a sub-question about it
  • Mention that we could use existing outlines which provide information about the shape, the distances and the angles, even if they are translated
  • Add a subtitle
  • Talk about what I am currently working on.
  • Add an illustration for footprint and roofprint
• Look for data about roof overhangs in general in France, in the Netherlands and more generally in Europe. This could be data about actual buildings, guidelines for the construction of buildings, or anything else.