Panoptic Multi-TSDFs Data
Datasets for the Panoptic Multi-TSDFs project, including a synthetic indoor “flat” dataset rendered in Unreal Engine and supplementary panoptic data for RIO sequences to run the panoptic mapping pipeline.
| Code: panoptic_mapping • Papers: IEEE | ArXiv |
Flat Dataset
- Two trajectories in a changing indoor environment with panoptic annotations
- Per frame: color PNG, depth TIFF (m), panoptic GT labels, Detectron2 predictions and metadata, pose, timestamps
- Ground-truth structure:
flat_<RunNo>_gt_10000.ply - Utility files: label maps, detectron labels, change log, intrinsics
- Simulator configuration (airsim.yaml), corrections, and waypoint logs included
RIO Demo Data
- Additional panoptic labels and point clouds for RIO scan IDs:
0cac7578-8d6f-2d13-8c2d-bfa7a04f8af3,20c9939d-698f-29c5-85c6-3c618e00061f,2451c041-fae8-24f6-9213-b8b6af8d86c1,ddc73793-765b-241a-9ecd-b0cebb7cf916,ddc73795-765b-241a-9c5d-b97744afe077,f62fd5f8-9a3f-2f44-8b1e-1289a3a61e26 - Per sequence: panoptic labels (GT and Detectron2), per-sequence ground-truth point clouds, combined scene point clouds, label maps
- Original RIO data must be downloaded separately: RIO dataset
Downloads
- Flat dataset (392.7 MB): https://doi.org/10.3929/ethz-c-000788335
- RIO panoptic add-ons (87.8 MB): https://doi.org/10.3929/ethz-c-000788335
Citation
L. Schmid et al.,
“Panoptic Multi-TSDFs: a Flexible Representation for Online Multi-resolution Volumetric Mapping and Long-term Dynamic Scene Consistency”, IEEE International Conference on Robotics and Automation (ICRA), 2022.
Tags: panoptic, mapping, tsdf, synthetic, indoor