pedroodb/glosl-how2sign

GloSL — How2Sign — American Sign Language (ASL) → English

Part of GloSL, a set of continuous sign language translation corpora in one format.

• Sign language: American Sign Language (ASL) (ase)
• Spoken language: English (en)
• Domain: instructional videos, green-screen studio
• Samples: 34,305 (61.15 h) — splits {'train': 30814, 'dev': 1527, 'test': 1964}
• Split policy: glosl-splits-v1, imported — How2Sign official train/val/test, video-disjoint (Duarte et al., CVPR 2021) (see card.json and diagnostics/)
• License: CC BY-NC 4.0 — public; source: https://how2sign.github.io/
• Cite: Duarte et al., How2Sign, CVPR 2021

Files

annotations.csv, card.json, poses/<NNNN>/<id>.pose, videos/<NNNN>/<id>.mp4 (same frames as the pose). card.json holds the full provenance.

annotations.csv columns (same in every GloSL dataset; empty when not applicable): id, split, text, gloss, gloss_type, signer, start, end, fps, n_frames, video_ref, pose_file, video_file.

Poses

mediapipe-holistic, 543 points (POSE_LANDMARKS 33, FACE_LANDMARKS 468, LEFT_HAND_LANDMARKS 21, RIGHT_HAND_LANDMARKS 21), 3D. Coordinates: pixels of the frame the pose refers to (header width x height, depth 0); z raw. Confidence: body = MediaPipe visibility (1.0 where the source shipped none); face/hands 1 when detected; missing point = (0,0,0) with confidence 0. Frame rate: native, per file (also in annotations.csv). Normalization (glosl-norm-v1) is applied at load time, not stored. Note: MediaPipe Holistic re-extracted from How2Sign green-screen FRONTAL sentence clips (source shipped only OpenPose); single centered signer so no YOLO crop; native fps; normalization applied at LOAD time (KB/05).