azithromycin/pseudolab_omx_f_PickUpDollWith2Cam

OMX-F Pick Up Doll With 2 Cameras TsFile

This is an Apache TsFile conversion of pseudolab/omx_f_PickUpDollWith2Cam, a LeRobot v2.1 OMX-F dataset containing demonstrations of picking up dolls with two cameras.

Source and provenance

• Original dataset: pseudolab/omx_f_PickUpDollWith2Cam
• Repository owner: pseudolab
• Uploader and commit author: LeoKimDowny
• License: Apache-2.0
• Paper/citation: not provided by the source repository
• Robot: omx_f; LeRobot codebase: v2.1
• Split: train; sampling frequency: 60 fps
• Task 0: Pick Up Dolls with 2 Cameras (4 episodes, 4,019 rows)
• Task 1: Pick up Dolls (90 episodes, 108,040 rows)
• Scale: 94 episodes, 112,059 rows, two task labels, 94 source Parquet shards
• Source data path: data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet

The source README embeds an obsolete two-episode info.json example. The counts above come from the pinned repository's current meta/info.json, all 94 Parquet footers, and the complete repository tree.

Schema

Column or group	TsFile role	Type	Notes
Time	TIME	INT64	round(timestamp * 1000) ms; restarts at zero per episode
episode_index	TAG	STRING	Original integer value stored as a TsFile device/TAG segment
task_index	TAG	STRING	Original integer value stored as a TsFile device/TAG segment
frame_index	FIELD	INT64	Original per-episode frame index
sample_index	FIELD	INT64	Renamed from source index
action_0 ... action_5	FIELD	FLOAT	Flattened source action[6]
observation_state_0 ... observation_state_5	FIELD	FLOAT	Flattened source observation.state[6]

Conversion details

• Time: TS_2DIFF + LZ4
• FLOAT/DOUBLE: GORILLA + LZ4
• INT32/INT64: TS_2DIFF + LZ4
• BOOLEAN: RLE + LZ4 when present; this source has no BOOLEAN field
• TAG values use the TsFile table-model device/tag mechanism.
• All train episodes are merged into one table and sorted by TAG columns and Time.
• Source timestamp is dropped because it is represented by Time / 1000 seconds.
• No numeric rows, action/state dimensions, frame indexes, episode indexes, or task indexes are dropped.
• Camera pixels are video assets rather than Parquet columns and are not embedded in TsFile.

Videos

Videos are not included in this TsFile dataset. They remain in the original repository under videos/chunk-000/ using this template: videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4.

Original video stream	Files	Bytes	Resolution	Codec
observation.images.camera1	94	604,862,192	640x480	H.264 (libx264)
observation.images.camera2	94	723,597,522	640x480	H.264 (libx264)

The two streams total 188 MP4 files and 1,328,459,714 bytes. Use episode_index and frame_index to align each TsFile row with both original camera streams at 60 fps.

Minimal read example

from tsfile import TsFileReader

reader = TsFileReader("data/pseudolab_omx_f_pick_up_doll_with_2_cam.tsfile")
columns = [
    "episode_index",
    "task_index",
    "frame_index",
    "sample_index",
    "action_0",
    "observation_state_0",
]
with reader.query_table("pseudolab_omx_f_pick_up_doll_with_2_cam", columns, batch_size=65536) as result:
    batch = result.read_arrow_batch()
    print(batch.to_pandas().head())
reader.close()

Citation

The source repository provides no formal citation. Cite the original Hugging Face dataset, repository owner pseudolab, and uploader LeoKimDowny when using this conversion.