Data-driven Techniques for Immersive WebXR Exploration of Indoor Spaces from Monocular 360 Input

Capturing and exploring common indoor environments, such as homes, offices, or public buildings, is essential across fields like architecture, civil engineering, urban planning, and real estate. In this context, 360-degree imagery is attracting lots of interest, since it offers a cost-effective solution for wide scene coverage in a single shot and, when viewed through Head-Mounted Displays (HMDs), it naturally supports 3D exploration via head movements. However, a single panorama remains a flat and static representation. To overcome these limitations, recent efforts in artificial intelligence (AI) have been devoted to the problem of generating editable, immersive 3D representations from a single 360-degree image. This tutorial presents state-of-the-art techniques for AI-driven scene reconstruction, virtual styling and staging, and their integration into immersive applications using WebXR, supporting stereoscopic vision and motion parallax on standard VR headsets.

@InProceedings{APG25,
  author       = {Agus, M. and Pintore, G. and Gobbetti, E.},
  title        = {Data-driven Techniques for Immersive WebXR Exploration of Indoor Spaces from Monocular 360 Input},
  booktitle    = {Proc. ACM Web3D - Tutorial track},
  year         = {2025},
  note         = {To appear},
  keywords     = {visual computing},
  url          = {https://publications.crs4.it/pubdocs/2025/APG25},
}