CHAS - Cultural Heritage Architectural Segmentation dataset

CHAS is a point cloud dataset from cultural heritage aimed to provide data for semantic segmentation techniques. The data in this repository were generated by terrestrial laser scanning and pictures from aerial survey using UAV. This dataset comprises relevant buildings representing religious and colonial Brazilian architecture. We hope to allow future research in point cloud segmentation and automatic Building Information Modeling.

We use the .7z file format to compact the clouds, providing high compression ratio. File names with "raw" suffix refers to RGB registered cloud. The "gt" suffix stands for ground truth clouds.

Raw point cloud description

Name                                      Architectural Style         Construction Year            TLS         UAV         Points*      ID

Good Death's Church                Neoclassical                       1867                         Yes          Yes          210      boa_morte
Central Mill                                   Industrial                           1881                         Yes          Yes           26        engenho
Ball House                                    Modern                             1940                         Yes           No          370          baile
Quilombo Farm                             Colonial                           1891                         Yes          Yes           55        quilombo                         St. Francis of Assisi Church          Modern                          1943                        Yes           Yes           46           curia

* millions 

Equipment

• FARO Focus3D X 330 HDR*
• DJI Inspire 2 with Gimbal DJI Zenmuse X4S 4K | H264 | F2.8 - F11 | 20MP**
• DJI Spark 12 MP Full HD GPS **

*Terrestrial Laser Scanner (TLS)

**Unmanned Aerial Vehicle (UAV)

Software

The following third-party software have been used to register the clouds.: FARO SCENE, CloudCompare and Pix4D Mapper. For a better understanding of the hybrid point cloud registration approach used in this dataset we recommend the following article: BIM for Heritage Documentation An ontology-based approach.

Ground-truth

Ground-Truth in this dataset was manually created with the Interactive Segmentation Tool on CloudCompare. Distinct colors represent different architectural elements. An “optimal” segmentation algorithm must partition the raw point clouds into equivalent single-color ground-truth clusters.    

How to cite this dataset?

Paiva, P. V. V., Cogima, C. K., Dezen-Kempter, E. and Carvalho, M. A. G. "Historical building point cloud segmentation combining hierarchical watershed transform and curvature analysis." Pattern Recognition Letters 135 (2020): 114-121. DOI: https://doi.org/10.1016/j.patrec.2020.04.010

Paiva, P. V. V., Cogima, C. K., Dezen-Kempter, E. and Carvalho, M. A. G. "CHAS - Cultural Heritage Architectural Segmentation dataset". Zenodo, March 26 2019. DOI: https://doi.org/10.5281/zenodo.2609498