TUMVIE#

class tonic.datasets.TUMVIE(save_to: str, recording: Union[str, List[str]], transform: Optional[Callable] = None, target_transform: Optional[Callable] = None, transforms: Optional[Callable] = None)[source]#

TUM-VIE

TUM-VIE is an event camera dataset for developing 3D perception and navigation algorithms. It contains handheld and head-mounted sequences in indoor and outdoor environments with rapid motion during sports and high dynamic range. TUM-VIE includes challenging sequences where state-of-the art VIO fails or results in large drift. Hence, it can help to push the boundary on event-based visual-inertial algorithms.

The dataset contains:

  • Stereo event data Prophesee Gen4 HD (1280x720 pixels)

  • Stereo grayscale frames at 20Hz (1024x1024 pixels)

  • IMU data at 200Hz

  • 6dof motion capture data at 120Hz (beginning and end of each sequence)

Timestamps between all sensors are synchronized in hardware.

Note

To be able to read this dataset, you will need hdf5plugin and PIL packages installed.

Note

Use bike-night instead of bike-dark in the list of recordings if you want that specific one.

@string{iros="International Conference on Intelligent Robots and Systems (IROS)"}
@inproceedings{klenk2021tumvie,
 author = {S Klenk and J Chui and N Demmel and D Cremers},
 title = {TUM-VIE: The TUM Stereo Visual-Inertial Event Dataset},
 eprint = {2108.07329},
 eprinttype = {arXiv},
 eprintclass = {cs.CV},
 booktitle = {International Conference on Intelligent Robots and Systems (IROS)},
 year = {2021},
 keywords = {tumvie, event camera, dynamic vision sensor, SLAM, vslam},
}
Parameters:
  • save_to (string) – Location to save files to on disk. Will save files in a sub folder ‘davis_dataset’.

  • recording (string) – Use the name of the recording or a list thereof to download it, for example ‘skate-hard’ or [‘skate-hard’, ‘bike-easy’]. See project homepage for a list of available recordings. Can use ‘all’ to download all available recordings.

  • transform (callable, optional) – A callable of transforms to apply to events and/or images.

  • target_transform (callable, optional) – A callable of transforms to apply to the targets/labels.

  • transforms (callable, optional) – A callable of transforms that is applied to both data and labels at the same time.

__getitem__(index)[source]#
Returns:

tuple of (data, target), where data is a dictionary of (events_left, events_right, imu) and targets is a dictionary of (images_left, images_right, mocap).