Open-H-Embodiment: A Large-Scale Dataset for Enabling Foundation Models in Medical Robotics
Paper • 2604.21017 • Published
Dataset Viewer
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
episode_index: int64
tasks: list<item: string>
child 0, item: string
length: int64
segment3_cache_layout: struct<path_template: string, fields: struct<identity: string, episode: string, starts: string, vali (... 170 chars omitted)
child 0, path_template: string
child 1, fields: struct<identity: string, episode: string, starts: string, valid_lengths: string, normalized: string>
child 0, identity: string
child 1, episode: string
child 2, starts: string
child 3, valid_lengths: string
child 4, normalized: string
child 2, episodes: int64
child 3, chunks: int64
child 4, source_combined_sha256: string
child 5, prediction_arrays_unchanged: bool
child 6, segment4_files_unchanged: bool
release_revision: string
segment3_cache_update: struct<previous_sha256: string, reason: string, chunks: int64, inference_precision: string>
child 0, previous_sha256: string
child 1, reason: string
child 2, chunks: int64
child 3, inference_precision: string
metadata_snapshot: struct<dataset: string, subset: string, revision: string>
child 0, dataset: string
child 1, subset: string
child 2, revision: string
model_repository: struct<repo_id: string, repo_type: string>
child 0, repo_id: string
child 1, repo_type: string
dataset_repository: struct<repo_id: string, repo_type: string>
child 0, repo_id: string
child 1, repo_type: string
files: struct<goal_points.csv: string, examples/episode_001440_endoscope_left.png: string, openh_metadata/i (.
...
, predictions/segment3/act_1308.npz: string
child 8, predictions/segment3/act_1309.npz: string
child 9, predictions/segment3/act_1310.npz: string
child 10, predictions/segment3/act_1440.npz: string
child 11, predictions/segment3/act_1441.npz: string
child 12, predictions/segment3/act_1442.npz: string
child 13, predictions/segment3/act_1443.npz: string
child 14, predictions/segment3/act_1444.npz: string
child 15, predictions/segment3/act_1445.npz: string
child 16, predictions/segment3/act_1446.npz: string
child 17, predictions/segment3/act_1447.npz: string
child 18, predictions/segment3/act_1448.npz: string
child 19, predictions/segment3/act_1450.npz: string
sources: struct<suturebot_code: string, paper: string, raw_annotations: string, openh_revision: string, raw_d (... 24 chars omitted)
child 0, suturebot_code: string
child 1, paper: string
child 2, raw_annotations: string
child 3, openh_revision: string
child 4, raw_dataset_revision: string
preprocessing_id: string
model_files: struct<dot_best_6311.ckpt: string, checkpoint_normalization.json: string, config.json: string>
child 0, dot_best_6311.ckpt: string
child 1, checkpoint_normalization.json: string
child 2, config.json: string
goal_example: struct<episode: int64, frame_index: int64, camera: string, dataset_revision: string, format: string>
child 0, episode: int64
child 1, frame_index: int64
child 2, camera: string
child 3, dataset_revision: string
child 4, format: string
to
{'files': {'goal_points.csv': Value('string'), 'examples/episode_001440_endoscope_left.png': Value('string'), 'openh_metadata/info.json': Value('string'), 'openh_metadata/tasks.jsonl': Value('string'), 'openh_metadata/episodes.jsonl': Value('string'), 'predictions/segment3/act_1306.npz': Value('string'), 'predictions/segment3/act_1307.npz': Value('string'), 'predictions/segment3/act_1308.npz': Value('string'), 'predictions/segment3/act_1309.npz': Value('string'), 'predictions/segment3/act_1310.npz': Value('string'), 'predictions/segment3/act_1440.npz': Value('string'), 'predictions/segment3/act_1441.npz': Value('string'), 'predictions/segment3/act_1442.npz': Value('string'), 'predictions/segment3/act_1443.npz': Value('string'), 'predictions/segment3/act_1444.npz': Value('string'), 'predictions/segment3/act_1445.npz': Value('string'), 'predictions/segment3/act_1446.npz': Value('string'), 'predictions/segment3/act_1447.npz': Value('string'), 'predictions/segment3/act_1448.npz': Value('string'), 'predictions/segment3/act_1450.npz': Value('string')}, 'model_files': {'dot_best_6311.ckpt': Value('string'), 'checkpoint_normalization.json': Value('string'), 'config.json': Value('string')}, 'model_repository': {'repo_id': Value('string'), 'repo_type': Value('string')}, 'dataset_repository': {'repo_id': Value('string'), 'repo_type': Value('string')}, 'release_revision': Value('string'), 'preprocessing_id': Value('string'), 'sources': {'suturebot_code': Value('string'), 'paper': Value('string'), 'raw_annotations': Value('string'), 'openh_revision': Value('string'), 'raw_dataset_revision': Value('string')}, 'goal_example': {'episode': Value('int64'), 'frame_index': Value('int64'), 'camera': Value('string'), 'dataset_revision': Value('string'), 'format': Value('string')}, 'metadata_snapshot': {'dataset': Value('string'), 'subset': Value('string'), 'revision': Value('string')}, 'segment3_cache_update': {'previous_sha256': Value('string'), 'reason': Value('string'), 'chunks': Value('int64'), 'inference_precision': Value('string')}, 'segment3_cache_layout': {'path_template': Value('string'), 'fields': {'identity': Value('string'), 'episode': Value('string'), 'starts': Value('string'), 'valid_lengths': Value('string'), 'normalized': Value('string')}, 'episodes': Value('int64'), 'chunks': Value('int64'), 'source_combined_sha256': Value('string'), 'prediction_arrays_unchanged': Value('bool'), 'segment4_files_unchanged': 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 544, 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 3052, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, 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 558, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, 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 361, 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 2381, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
episode_index: int64
tasks: list<item: string>
child 0, item: string
length: int64
segment3_cache_layout: struct<path_template: string, fields: struct<identity: string, episode: string, starts: string, vali (... 170 chars omitted)
child 0, path_template: string
child 1, fields: struct<identity: string, episode: string, starts: string, valid_lengths: string, normalized: string>
child 0, identity: string
child 1, episode: string
child 2, starts: string
child 3, valid_lengths: string
child 4, normalized: string
child 2, episodes: int64
child 3, chunks: int64
child 4, source_combined_sha256: string
child 5, prediction_arrays_unchanged: bool
child 6, segment4_files_unchanged: bool
release_revision: string
segment3_cache_update: struct<previous_sha256: string, reason: string, chunks: int64, inference_precision: string>
child 0, previous_sha256: string
child 1, reason: string
child 2, chunks: int64
child 3, inference_precision: string
metadata_snapshot: struct<dataset: string, subset: string, revision: string>
child 0, dataset: string
child 1, subset: string
child 2, revision: string
model_repository: struct<repo_id: string, repo_type: string>
child 0, repo_id: string
child 1, repo_type: string
dataset_repository: struct<repo_id: string, repo_type: string>
child 0, repo_id: string
child 1, repo_type: string
files: struct<goal_points.csv: string, examples/episode_001440_endoscope_left.png: string, openh_metadata/i (.
...
, predictions/segment3/act_1308.npz: string
child 8, predictions/segment3/act_1309.npz: string
child 9, predictions/segment3/act_1310.npz: string
child 10, predictions/segment3/act_1440.npz: string
child 11, predictions/segment3/act_1441.npz: string
child 12, predictions/segment3/act_1442.npz: string
child 13, predictions/segment3/act_1443.npz: string
child 14, predictions/segment3/act_1444.npz: string
child 15, predictions/segment3/act_1445.npz: string
child 16, predictions/segment3/act_1446.npz: string
child 17, predictions/segment3/act_1447.npz: string
child 18, predictions/segment3/act_1448.npz: string
child 19, predictions/segment3/act_1450.npz: string
sources: struct<suturebot_code: string, paper: string, raw_annotations: string, openh_revision: string, raw_d (... 24 chars omitted)
child 0, suturebot_code: string
child 1, paper: string
child 2, raw_annotations: string
child 3, openh_revision: string
child 4, raw_dataset_revision: string
preprocessing_id: string
model_files: struct<dot_best_6311.ckpt: string, checkpoint_normalization.json: string, config.json: string>
child 0, dot_best_6311.ckpt: string
child 1, checkpoint_normalization.json: string
child 2, config.json: string
goal_example: struct<episode: int64, frame_index: int64, camera: string, dataset_revision: string, format: string>
child 0, episode: int64
child 1, frame_index: int64
child 2, camera: string
child 3, dataset_revision: string
child 4, format: string
to
{'files': {'goal_points.csv': Value('string'), 'examples/episode_001440_endoscope_left.png': Value('string'), 'openh_metadata/info.json': Value('string'), 'openh_metadata/tasks.jsonl': Value('string'), 'openh_metadata/episodes.jsonl': Value('string'), 'predictions/segment3/act_1306.npz': Value('string'), 'predictions/segment3/act_1307.npz': Value('string'), 'predictions/segment3/act_1308.npz': Value('string'), 'predictions/segment3/act_1309.npz': Value('string'), 'predictions/segment3/act_1310.npz': Value('string'), 'predictions/segment3/act_1440.npz': Value('string'), 'predictions/segment3/act_1441.npz': Value('string'), 'predictions/segment3/act_1442.npz': Value('string'), 'predictions/segment3/act_1443.npz': Value('string'), 'predictions/segment3/act_1444.npz': Value('string'), 'predictions/segment3/act_1445.npz': Value('string'), 'predictions/segment3/act_1446.npz': Value('string'), 'predictions/segment3/act_1447.npz': Value('string'), 'predictions/segment3/act_1448.npz': Value('string'), 'predictions/segment3/act_1450.npz': Value('string')}, 'model_files': {'dot_best_6311.ckpt': Value('string'), 'checkpoint_normalization.json': Value('string'), 'config.json': Value('string')}, 'model_repository': {'repo_id': Value('string'), 'repo_type': Value('string')}, 'dataset_repository': {'repo_id': Value('string'), 'repo_type': Value('string')}, 'release_revision': Value('string'), 'preprocessing_id': Value('string'), 'sources': {'suturebot_code': Value('string'), 'paper': Value('string'), 'raw_annotations': Value('string'), 'openh_revision': Value('string'), 'raw_dataset_revision': Value('string')}, 'goal_example': {'episode': Value('int64'), 'frame_index': Value('int64'), 'camera': Value('string'), 'dataset_revision': Value('string'), 'format': Value('string')}, 'metadata_snapshot': {'dataset': Value('string'), 'subset': Value('string'), 'revision': Value('string')}, 'segment3_cache_update': {'previous_sha256': Value('string'), 'reason': Value('string'), 'chunks': Value('int64'), 'inference_precision': Value('string')}, 'segment3_cache_layout': {'path_template': Value('string'), 'fields': {'identity': Value('string'), 'episode': Value('string'), 'starts': Value('string'), 'valid_lengths': Value('string'), 'normalized': Value('string')}, 'episodes': Value('int64'), 'chunks': Value('int64'), 'source_combined_sha256': Value('string'), 'prediction_arrays_unchanged': Value('bool'), 'segment4_files_unchanged': Value('bool')}}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Segment 3 SutureBot workshop cache
Small metadata snapshots, matched goal annotations, one teaching still, and cached ACT predictions for OpenHTutorial_SutureBotACT_Inference.ipynb. Research and educational purposes only; not for clinical use. This is the local staging copy for dataset repository njb15/corl2026-seg3-cache; no upload is implied.
Files
manifest.json: SHA-256 hashes, the required model configuration/checkpoint/statistics hashes, pinned source revisions, and cache schema/provenance.goal_points.csv: insertion/exit coordinates and verified source-episode mappings for five evaluation clips and two training examples.examples/episode_001440_endoscope_left.png: lossless frame-zero teaching still from the pinned OpenH stream.openh_metadata/{info.json,tasks.jsonl,episodes.jsonl}: metadata snapshots from that same OpenH revision.predictions/segment3/act_<episode>.npz: one compressed NumPy file per evaluation clip; 15 clips and 332 observation chunks total.
Each NPZ contains identity (JSON provenance), episode (scalar ID), starts (N observation frame indices, stride 20), valid_lengths (N unpadded lengths, at most 60), and normalized (N×60×20 FP32 hybrid-action predictions). These arrays do not require pickle; load with allow_pickle=False. The checkpoint and prediction arrays are unchanged by the separation of the repositories.
The model checkpoint, normalization, and config.json come from model repository njb15/corl2026-suturebot-act. manifest.json binds this cache to those exact model assets. Checkpoints are not duplicated here.
Notebook use
The notebook has independent MODEL_REPO_ID / MODEL_REVISION and CACHE_REPO_ID / CACHE_REVISION settings. Use the reviewed corl2026-v1 tags or exact upload commits. Public participant downloads set token=False. The intended release tags are not created by local staging or the review-branch upload script.
USE_CACHED_RESULTS = True skips checkpoint loading and inference. Live mode computes missing predictions from the saved checkpoint. GOAL_POINTS_SOURCE selects either the prepared CSV or original source ZIP members, using the verified mapping and checking identical coordinates.
Both modes still retrieve the selected original OpenH Parquets/videos for reference poses, evaluation targets, and playback. Raw videos, episode Parquets, prepared camera observations, instructor reports, and model checkpoints are not included in this upload bundle.
Predictions use measured starting poses and the authors' reanchoring convention. The notebook currently evaluates against measured poses/jaw angles and exports that convention with its metrics. Training-split overlap is unknown; these are offline teaching comparisons, not a benchmark or clinical validation.
Sources and terms
See SOURCES.md for the original dataset and annotation sources and their pinned revisions. This bundle does not grant a new license for source material. Retain the source terms when redistributing derived data.
Citation: Haworth et al., SutureBot, NeurIPS 2025 Datasets and Benchmarks (arXiv:2510.20965); Open-H-Embodiment report.
- Downloads last month
- 46