Urb3DCD-cls
Notes on the Dataset
- In addition to per-point change labels, the majority change per point cloud is specified through the folder structure.
- Some point clouds contain very few points (even only one point in some cases). For our cases, we only included point clouds with more than 10 points (excluding 15 scenes).
Detailed Information
| General Information | |
| Name | Urb3DCD-cls |
| Release Year | 2023 |
| Terms of Use | CC BY 4.0 |
| Access Requirements | Create account |
| Dataset Size | 323.9 MB |
| Documentation | Multi-paragraph description of the dataset, e.g., in the form of a paper section or a comprehensive readme file |
| Code | Yes |
| Applications | Built environment change detection and classification |
| Detailed Applications | Urban point cloud change classification |
| Acquisition | |
| Number of Scenes | 7830 | One coherent scene is recorded, but it is split into multiple non-overlapping parts |
| Number of Epochs per Scene (minimum) | 2 |
| Number of Epochs per Scene (median) | 2 |
| Number of Epochs per Scene (maximum) | 2 |
| General Scene Type | Urban |
| Specific Scene Type | Object in context |
| Location | Lyon (France) |
| Acquisition Type | Airborne laser scanning | Synthetic scan |
| Acquisition Device | Simulated ALS |
| Acquisition Platform | Simulated (aircraft) |
| Scan Interval | Undefined |
| Acquisition Months | |
| Representation | |
| Data Representation | Unstructured, globally aligned (e.g., point cloud or ray cloud) |
| Specific Data Representation | Point cloud |
| File Format/Encoding | PLY |
| Raw Data | - |
| Additional Data | - |
| Coordinate System | 9828 |
| Quality and Usability | |
| Registration | Finely registered |
| Number of Partial Epochs | - |
| Unusable Data Reason | Small (less than ten points) |
| Splits | Yes |
| Per-Point Attributes | |
| Intensity/Reflectivity | No |
| Color | No |
| Semantic Labels | 4 |
| Instance Labels | No |
| Change Labels | 7 |
| Statistics | |
| Number of Points per Epoch (minimum) | 11 |
| Number of Points per Epoch (median) | 4K |
| Number of Points per Epoch (maximum) | 8K |
| Avg. Point Spacing (minimum) | 1.6dm |
| Avg. Point Spacing (median) | 3.3dm |
| Avg. Point Spacing (maximum) | 9.7dm |
| Avg. Change Points per Epoch | 17.76% |
Paper Reference 1
@article{degelis2023dataseturb3dcdv2,
author = {Iris de~G\'elis and Sébastien Lefèvre and Thomas Corpetti},
title = {Siamese KPConv: 3D multiple change detection from raw point clouds using deep learning},
journal = {ISPRS Journal of Photogrammetry and Remote Sensing},
volume = {197},
pages = {274--291},
year = {2023},
issn = {0924-2716},
doi = {10.1016/j.isprsjprs.2023.02.001},
}
Dataset Reference 1
@data{degelis2021dataseturb3dcd,
doi = {10.3390/rs13132629},
author = {de Gélis, Iris and Lefèvre, Sébastien and Corpetti, Thomas},
publisher = {IEEE Dataport},
title = {Urb3DCD : Urban Point Clouds Simulated Dataset for 3D Change Detection},
year = {2021},
}