data/processed/sample_sentence_clips.jsonl

{"clip_id": "team_hello_want_drink_001", "split": "train", "source": "team-recorded-placeholder", "gloss": ["HELLO", "WANT", "DRINK"], "english": "Hello, I want a drink.", "video_path": "data/raw/team_hello_want_drink_001.mp4", "sampled_frames": [], "landmarks_path": null, "notes": "Metadata-only sample. Do not commit raw video or extracted frames."}
{"clip_id": "team_please_help_001", "split": "train", "source": "team-recorded-placeholder", "gloss": ["PLEASE", "HELP"], "english": "Please help.", "video_path": "data/raw/team_please_help_001.mp4", "sampled_frames": [], "landmarks_path": null, "notes": "Metadata-only sample. Add frame paths after Trey extracts frames locally."}
{"clip_id": "team_thank_you_finished_001", "split": "val", "source": "team-recorded-placeholder", "gloss": ["THANK_YOU", "FINISHED"], "english": "Thank you, I am finished.", "video_path": "data/raw/team_thank_you_finished_001.mp4", "sampled_frames": [], "landmarks_path": null, "notes": "Validation placeholder for the sentence manifest contract."}
{"clip_id": "team_no_stop_001", "split": "test", "source": "team-recorded-placeholder", "gloss": ["NO", "STOP"], "english": "No, stop.", "video_path": "data/raw/team_no_stop_001.mp4", "sampled_frames": [], "landmarks_path": null, "notes": "Test placeholder for CNN versus VLM comparison."}