Datasets:

ppoitier
/

lsfb-cont

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 81, in _split_generators
                  first_examples = list(islice(pipeline, self.NUM_EXAMPLES_FOR_FEATURES_INFERENCE))
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 32, in _get_pipeline_from_tar
                  fs: fsspec.AbstractFileSystem = fsspec.filesystem("memory")
                                                  ~~~~~~~~~~~~~~~~~^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/fsspec/registry.py", line 302, in filesystem
                  cls = get_filesystem_class(protocol)
                File "/usr/local/lib/python3.14/site-packages/fsspec/registry.py", line 239, in get_filesystem_class
                  raise ValueError(f"Protocol not known: {protocol}")
              ValueError: Protocol not known: memory
              
              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 71, 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.

LSFB-CONT: Continuous French Belgian Sign Language Dataset

LSFB-CONT is the continuous (untrimmed) subset of the LSFB corpus: full-length recordings of signers, with pose sequences and, for part of the data, temporal sign/translation annotations. It is built for Continuous Sign Language Recognition (CSLR), Sign Language Segmentation (SLS), and Sign Language Translation (SLT).

The isolated, per-sign counterpart of this dataset is lsfb-isol, where each sample is a single pre-segmented sign clip extracted from these recordings.

Dataset structure

The dataset is distributed as WebDataset .tar shards, in two configs:

annotated (8 shards) — recordings with temporal annotations (glosses, lemmas, translations).
unannotated (12 shards) — additional recordings with pose only, no annotations. Useful for self-supervised / semi-supervised pretraining.

Each sample is keyed by a recording id (e.g. CLSFBI0301A_S007_B) and contains:

pose.<body_part>.npy — keypoint sequence for the whole recording, for one body part (upper_pose, left_hand, right_hand, lips, left_eye, right_eye, left_eyebrow, right_eyebrow, left_iris, right_iris)
signer.txt — anonymized signer id (e.g. S007)
language.txt — always lsfb

annotated samples additionally contain:

annotations.left_hand.json / annotations.right_hand.json / annotations.both_hands.json — lists of {start_ms, end_ms, start_frame, end_frame, gloss, lemma, sign_type, specifier, variation} segments, one list per annotated hand/tier
annotations.translation.json — list of {start_ms, end_ms, start_frame, end_frame, text} French translation segments

This release does not include raw video; only pose sequences and annotations are provided.

Usage

This dataset is designed to be loaded with sign-language-data-loading (sldl, PyPI), which reads the WebDataset shards directly and handles pose decoding, windowing and target encoding:

pip install "sign-language-data-loading[torch]"

from sldl import SignLanguageDataset
from sldl.configs import LSFBContConfig

dataset = SignLanguageDataset.from_config(
    LSFBContConfig(
        root="path/to/lsfb-cont",  # local clone or download of this repo
        split="training",
        load_videos=False,  # no video files in this release
    )
)

sample = dataset[0]
print(sample["annotations"]["both_hands"].head())
print(sample["poses"]["upper_pose"].shape)

Pose and annotation transforms (e.g. converting sign boundaries to frame labels for CSLR) are provided by sign-language-tools (PyPI), which sldl's built-in target encoders (e.g. FrameLabelsTarget) build on.

License

This dataset is distributed under CC BY-NC-SA 4.0.

Related projects

sign-language-data-loading (PyPI) — PyTorch-friendly data loading for sign language datasets stored as WebDataset shards; this dataset's intended loader.
sign-language-tools (PyPI) — pose and annotation transforms for sign language processing.
lsfb-isol — the isolated (single-sign) counterpart of this dataset, useful for Isolated Sign Language Recognition (ISLR).

Downloads last month: 42

Total file size:

70.5 GB