VoxelScape devkit

This repository contains helper scripts to open, visualize and process point clouds and annotations from the VoxelScape dataset.

• Link to VoxelScape dataset.

---

Dataset structure

After downloading and un-zipping each sequence folder from the above download link into /VoxelScape/dataset/ directory. The data should be organized as follows:

/VoxelScape/dataset/
          ├── 00/
          │   ├── bboxs/
          │   │     ├ xyz_iom_000000.pkl
          │   │     └ xyz_iom_000001.pkl
          │   ├── kitti_labels/
          │   │     ├ xyz_iom_000000.label
          │   │     └ xyz_iom_000001.label
          |   ├── orig_labels/
          |   |     ├ xyz_iom_000000_subl.label
          |   |     ├ xyz_iom_000000.label
          |   |     ├ xyz_iom_000001_subl.label
          |   |     ├ xyz_iom_000001.label
          │   └── velodyne/
          │         ├ xyz_iom_000000.bin
          │         └ xyz_iom_000001.bin
          ├── 01/
          ├── 02/
          .
          .
          .
          └── 99/

• velodyne contains the pointclouds for each scan in each sequence. Each .bin scan is a list of float32 points in [x,y,z,intensity] format.
• bboxs contains the 3D bounding boxes annotation of the 9 object classes in the dataset. Each .pkl file contains the 8-vertices of the 3D bounding box and the class labels for each object exist in the corresponding .bin scan.
• kitti_labels contains only 19 merged/subset semantic class labels, which correspond to the labels exist in the SemanticKITTI dataset, for each scan in each sequence. Each .label file contains a uint32 label for each point in the corresponding .bin scan.
• orig_labels contains the total 32 semantic class labels introduced in the VoxelScape dataset. Each _subl.label file contains a fine-grained uint32 label for differnet attributes of each object in the corresponding .bin scan. See sem_util.py for more information on the sub-labels.

The dataset has two main configuration files, namely config/semantic-voxelscape.yaml and config/bbox-voxelscape.yaml. As the name implies, the first config file corresponds to the semantic labels, while the second one corresponds to the 3D bbox annotations. There's also another config file config/semantic-kitti.yaml which has the semantic class labels from the SemanticKITTI dataset which can be used for visualising only the 19 subset labels from kitti_labels. In both the config/semantic-voxelscape.yaml and config/semantic-kitti.yaml files you will find:

• labels: dictionary which maps the numeric labels in _subl.label/.label files inside orig_labels/kitti_labels folder to a string class. Example: 256: "construction-cone"
• color_map: dictionary which maps numeric labels in _subl.label/.label files inside orig_labels/kitti_labels folder to a bgr color for visualization. Example 256: [79, 79, 47] # construction-cone, dark green-ish

Dependencies for the devkit:

System dependencies

$ sudo apt install python3-dev python3-pip

Python dependencies

$ sudo pip3 install -r requirements.txt

Scripts:

ALL OF THE SCRIPTS CAN BE INVOKED WITH THE --help (-h) FLAG, FOR EXTRA INFORMATION AND OPTIONS.

Semantic Labels

To visualize the full 32 semantic labels of the VoxelScape dataset, use the visualize_semantic.py script. It will open an interactive open3d visualization of the pointclouds coloured with the semantic labels colormap in the config/semantic-voxelscape.yaml file.

$ ./visualize_semantic.py --dataset /path/to/voxelscape/dataset/ --seq 00

where:

• dataset is the path to the VoxelScape dataset.
• seq is the sequence folder to be accessed.

In order to visualize only the 19 subset semantic labels (similar to the SemanticKITTI dataset) instead, pass the SemanticKITTI configuration file config/semantic-kitti.yaml to the --config option:

$ ./visualize_semantic.py --dataset /path/to/voxelscape/dataset/ --seq 00  --config config/semantic-kitti.yaml

3D Bbox Labels

To visualize the 3D bboxs annotations, use the visualize_bbox.py script. It will open an interactive open3d visualization of the 3D bounding boxes of each object (from our 9 object calsses) coloured according to its class name described in the config/bbox-voxelscape.yaml file.

$ ./visualize_bbox.py --dataset /path/to/voxelscape/dataset/ --seq 00 

where:

• dataset is the path to the VoxelScape dataset.
• seq is the sequence folder to be accessed.

In order to accelerate the visualization, the maximum number of 3D bboxs rendered for each frame is limited to 10 3D bbox per each scan. You can change this number by changing the --bbox_num option :

$ ./visualize_semantic.py --dataset /path/to/voxelscape/dataset/ --seq 00  --bbox_num 15

License

Our code is released under MIT license (see LICENCE.txt file).