File size: 1,753 Bytes
905612a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 | """Iterate local single-episode LeRobot datasets using episodes.csv."""
import argparse
import csv
from pathlib import Path
from lerobot.datasets.lerobot_dataset import LeRobotDataset
def iter_datasets(root, names=None, limit=4):
root = Path(root).resolve()
selected = set(names) if names else None
loaded = 0
with (root / "episodes.csv").open(newline="") as stream:
for row in csv.DictReader(stream):
name = row["episode"]
if selected is not None and name not in selected:
continue
path = root / row["episode_path"]
if not (path / "meta/info.json").is_file():
raise FileNotFoundError(f"Extract the corresponding shard first: {name}")
dataset = LeRobotDataset(repo_id=f"Crobotia/{name}", root=path,
video_backend="pyav")
yield name, dataset
loaded += 1
if limit is not None and loaded >= limit:
return
if loaded == 0:
raise ValueError("No selected episode was found in episodes.csv")
if __name__ == "__main__":
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--root", type=Path, default=Path("."))
parser.add_argument("--episodes", nargs="+", help="Episode names from episodes.csv")
parser.add_argument("--limit", type=int, default=4, help="Maximum episodes to load (default: 4)")
args = parser.parse_args()
if args.limit < 1:
parser.error("--limit must be positive")
for name, dataset in iter_datasets(args.root, args.episodes, args.limit):
sample = dataset[0]
print(name, dataset.num_frames, sample["task"], tuple(sample["observation.state"].shape))
|