Hugging Face's logo

Datasets:
physer
/
videomanip-reproduction

ArXiv:
Tags:
robotics
grasping
dexterous-manipulation
videomanip
dro-grasp
isaaclab
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
object: string
num_envs: int64
thresh_m: double
force_N: double
mass_kg: double
rand_dirs: int64
force_scale: double
success_rate: double
per_env: list<item: struct<env: int64, trial: int64, disp_m: double, success: bool>>
  child 0, item: struct<env: int64, trial: int64, disp_m: double, success: bool>
      child 0, env: int64
      child 1, trial: int64
      child 2, disp_m: double
      child 3, success: bool
to
{'object': Value('string'), 'num_envs': Value('int64'), 'thresh_m': Value('float64'), 'force_N': Value('float64'), 'mass_kg': Value('float64'), 'success_rate': Value('float64'), 'per_env': List({'env': Value('int64'), 'trial': Value('int64'), 'disp_m': Value('float64'), 'success': Value('bool')})}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 489, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2818, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2355, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2380, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2369, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              object: string
              num_envs: int64
              thresh_m: double
              force_N: double
              mass_kg: double
              rand_dirs: int64
              force_scale: double
              success_rate: double
              per_env: list<item: struct<env: int64, trial: int64, disp_m: double, success: bool>>
                child 0, item: struct<env: int64, trial: int64, disp_m: double, success: bool>
                    child 0, env: int64
                    child 1, trial: int64
                    child 2, disp_m: double
                    child 3, success: bool
              to
              {'object': Value('string'), 'num_envs': Value('int64'), 'thresh_m': Value('float64'), 'force_N': Value('float64'), 'mass_kg': Value('float64'), 'success_rate': Value('float64'), 'per_env': List({'env': Value('int64'), 'trial': Value('int64'), 'disp_m': Value('float64'), 'success': Value('bool')})}
              because column names don't match

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.

VideoManip Reproduction (IsaacLab) — data & checkpoints

Unique artifacts produced by the unofficial sim-only reproduction of VideoManip (arXiv:2602.09013) in IsaacLab 2.3.2 / Isaac Sim 5.1. Code, docs, protocol and full result tables: https://github.com/physercoe/videomanip-reproduction

Layout 

checkpoints/
  mixed3x/           epoch_{5..40}.pth   DRO run on HaMeR+ContactOpt data  (6-obj sim mean 47.0% @ e5)
  mixed3x_handflow/  epoch_{1..20}.pth   DRO run on HandFlow+ContactOpt data (50.0% @ e16; 65.0% on paper-20)
  mixed3x_union/     epoch_{1..20}.pth   DRO run on the union dataset      (69.0% @ e2)
datasets/
  CMapDataset_{videomanip,handflow,union,mixed,mixed3x,mixed3x_handflow,mixed3x_union,selfdistill}/
  PointCloud_videomanip/   512x6 (xyz+normal) object point clouds, 6 objects
hand_records/<obj>/{hand,handflow}/*.npy   per-frame hand records (HaMeR / HandFlow, depth-corrected)
predictions/<obj>/predicted_grasps*.npz    DRO predictions + wrench-refined variants (object-at-origin q19)
derived_meshes/<obj>/{hand,object}.ply     extracted from the paper's predicted-grasp GLBs
eval_results/<obj>/*.json                  every sim-eval result (100-trial disturbance protocol)
eval_results/_global/                      loss/oscillation curves, trackio db, eval logs

6 own objects: spraybottle, bottle, can, bulb, hat, jengabox. pref_* = the paper's 20 objects evaluated with our models/harness. q19 convention = (dummy, x,y,z, roll,pitch,yaw, 12 finger joints) — see repo AGENTS.md.

Checkpoints & headline numbers

checkpoint 6 own objects paper's 20 objects
mixed3x/epoch_5.pth 47.0% 56.65%
mixed3x_handflow/epoch_16.pth 50.0% 65.0% (paper reports 63.75%)
mixed3x_union/epoch_2.pth 69.0% 62.05%

Note the checkpoint-oscillation finding (see repo docs/REPORT.md): later epochs are NOT better — always select checkpoints by sim evaluation, not by loss.

Usage 

from huggingface_hub import snapshot_download
p = snapshot_download(repo_id="physer/videomanip-reproduction", repo_type="dataset")
# or selectively: allow_patterns=["checkpoints/mixed3x_union/epoch_2.pth", "datasets/*"]

Datasets load with the patched DRO-Grasp in the GitHub repo (DRO_DATASET_DIR=data/<name>); inference: scripts/run_dro_inference.py.

Provenance & license

  • CMapDataset_mixed* contain samples derived from the DRO-Grasp authors' released grasp data (https://github.com/zhenyuwei2003/DRO-Grasp) mixed with grasps reconstructed by this project — credit both.
  • derived_meshes/ come from the VideoManip authors' predicted-grasp GLBs (https://github.com/videomanip/videomanip.github.io) — credit the paper's authors.
  • Everything else was produced by this reproduction and is released under MIT.
  • If you use these artifacts, cite the original VideoManip paper (bibtex in the GitHub README).
Downloads last month
50
Total file size:
2.84 GB

Paper for physer/videomanip-reproduction