Automatic 3D modeling and exploration of indoor structures from panoramic imagery

Surround-view panoramic imaging delivers extensive spatial cover- age and is widely supported by professional and commodity capture devices. Research on inferring and exploring 3D indoor models from 360-degree images has recently flourished, resulting in highly effective solutions. Nevertheless, challenges persist due to the complexity and variability of indoor environments and issues with noisy and incomplete data. This course provides an up-to-date integrative view of the field. After introducing a characterization of input sources, we define the structure of output models, the priors exploited to bridge the gap between imperfect input and desired output, and the main characteristics of geometry reasoning and data-driven approaches. We then identify and discuss the main sub-problems in indoor reconstruction from panoramas and review and analyze state-of-the-art solutions for indoor capture, room modeling, integrated model computation, visual representation generation, and immersive exploration. Relevant examples of implemented pipelines are described, focusing on deep-learning solutions. We finally point out relevant research issues and analyze research trends.

@InProceedings{PAG24,
  author       = {Pintore, G. and Agus, M. and Gobbetti, E.},
  title        = {Automatic 3D modeling and exploration of indoor structures from panoramic imagery},
  booktitle    = {SIGGRAPH Asia 2024 Courses (SA Courses '24)},
  month        = {december},
  year         = {2024},
  publisher    = {ACM Press},
  keywords     = {visual and data-intensive computing, indoor, panoramic, deep learning},
  doi          = {10.1145/3680532.3689580},
  url          = {https://publications.crs4.it/pubdocs/2024/PAG24},
}