The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 80, in _split_generators
raise ValueError(
...<2 lines>...
)
ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
icf-capsule-polishing-simulation
Private research dataset of Blender-generated capsule polishing videos and synchronized rendered/ideal trajectories. Access is restricted to the repository owner and authorized collaborators. No redistribution permission or public license is granted by this card.
Contents and splits
5,765 complete exports; 4,022 train, 879 validation and 864 test. RGB videos are 1920×1080, 800 frames at 100 fps in the inspected exports. Configurations vary shell count (3/20), geometry (concentric/eccentric), groove RPM (21/48/78/108), pad RPM (0/50/100), slip, orientation, and visual randomization.
The frozen split groups related physical simulation families and shared render-cache keys before window extraction. Release preparation additionally checks complete video and trajectory SHA256 hashes for cross-partition duplicates. Repeated renders remain within a partition; counts represent exports, not unique videos. Checkpoint selection and threshold tuning use validation only. Test is reserved for the final frozen comparison.
Files:
manifest.csv: portable sample paths, run IDs, scientific configuration, partition and video hashes.split.csv: frozen group assignments and trajectory hashes.shards/{train,val,test}-XXXXX.tar: regular files undersamples/<run_id>/.- Each sample contains
video.mp4,trajectories.npz,parameters.json. checksums.json,shards.json: SHA256 hashes for every payload and shard.split_summary.json,release_summary.json,config.json: version and provenance summaries.
parameters.json omits the original workstation output directory. Blender .blend files, render caches, redundant CSV trajectories, logs and previews are not included. This release supplies learning/inference data, not the Blender scene generator.
Download and use
Code: https://github.com/shashankgalla/capsule-motion-tracking (requires collaborator access once published).
hf auth login
python scripts/share_dataset.py download \
--repo-id shashankgalla/icf-capsule-polishing-simulation \
--out data/icf-capsule-polishing-simulation --partitions train val
Download test only for final evaluation with --partitions test. The downloader verifies shard/payload hashes and safely extracts regular files. A full manifest remains available even when only selected partitions have been downloaded. Do not use datasets.load_dataset() directly on these custom NPZ/video shards; use the provided downloader and manifest loader. Keep the dataset revision returned by the upload alongside experimental results.
Trajectory schema
NPZ lengths use meters, velocities m/s, angles radians and angular rates rad/s. Main arrays:
| Key | Shape | Meaning |
|---|---|---|
time_s |
[T] |
acquisition times |
position_m |
[N,T,3] |
rendered centers in simulation coordinates |
quaternion_wxyz |
[N,T,4] |
rendered world-from-body orientations, scalar first |
image_center_px |
[N,T,2] |
projected centers, image x right/y down |
image_radius_px |
scalar | sphere radius in pixels |
angular_velocity_world_radps |
[N,T,3] |
ideal kinematic vector, not all visual perturbations |
omega_c |
[T] |
modeled circulation rate |
omega_b |
[T] |
modeled spin quantity; consult simulation convention |
Other arrays preserve ideal states, local quaternions, phase/randomizer events and contact diagnostics. Use finite differences of rendered quaternions for visual-motion scoring; ideal labels can differ due to wobble, jitter and randomization. Coordinate frames must be aligned before comparing axes.
Limits
The renderer uses an orthographic camera and approximate motion blur. Exported in_frame is a bounds flag, not visibility or ownership. Reconstructed material-point visibility based on sphere geometry and foreground masks is approximate; no depth/object-ID pass is included. Synthetic geometry, textures and lighting do not guarantee transfer to real polishing videos. Grouped splits prevent the declared and content-identical leakage routes, but do not establish absence of every possible statistical relationship.
Contact residuals describe the ideal kinematic state. The visualization is not a collision/contact-force simulation or an independently validated laboratory frame. Validate real-video calibration and timing separately.
- Downloads last month
- 116