{
  "glosl_version": "0.1",
  "id": "lsat",
  "name": "LSA-T — Continuous Argentinian Sign Language",
  "task": "continuous-slt",
  "sign_language": {
    "code": "aed",
    "name": "Lengua de Señas Argentina (LSA)"
  },
  "spoken_language": {
    "code": "es",
    "name": "Spanish"
  },
  "license": "repo MIT (code only); dataset data unlicensed",
  "access": {
    "redistributable": true,
    "gated": false,
    "reupload_ok": "public",
    "video_rehost": "permitted",
    "video_rehost_note": "The source video originates from the Canal Sordos Argentina (CN Sordos) YouTube channel and is republished here WITH the channel's explicit permission. Pose + video are hosted publicly on this basis.",
    "obtain_url": "https://midusi.github.io/LSA-T/"
  },
  "modalities": [
    "video",
    "pose",
    "text"
  ],
  "pose": {
    "estimator": "mediapipe-holistic",
    "source_keypoints": 543,
    "canonical_subset": "glosl-body-hands-v1",
    "normalization": "glosl-norm-v1",
    "dims": 3,
    "format": ".pose",
    "note": "reused from slt_datasets MediaPipe-Holistic .npy (KB/12); normalization applied at LOAD time (KB/05)"
  },
  "annotation_types": [],
  "splits": {
    "dev": 1354,
    "train": 5413,
    "test": 1692
  },
  "signers": null,
  "provenance": {
    "converted_with": "glosl-datasets build_and_upload.py",
    "source_paper": "Dal Bianco et al., LSA-T, IBERAMIA 2022 (arXiv:2211.15481)",
    "source_harmonized": "/mnt/disk3Tb/slt-datasets/LSAT",
    "public_needs_author_auth": false,
    "public_auth_note": "Video republished with the explicit permission of the CN Sordos (Canal Sordos Argentina) YouTube channel; no further author authorization is required for the public hosting of this dataset."
  },
  "video": {
    "track": "signer-clip",
    "cropped_signer_bbox": false,
    "note": "LSA-T videos are the dataset's own per-utterance signer clips (CN Sordos, signer-centric; ~1 person/frame at 1920x1080). Stored at source framing; not additionally YOLO-bbox-cropped in v1. A deterministic active-signer crop (glosl_datasets.pipeline.extract.crop_signer_video) is available if a tighter crop is wanted later."
  }
}