⬆️ upgrade lerobot dataset to v3.0 (#60)

* update openx

* update libero

* update agibot

* update robomind

* update readme

* update return type

README.md

@@ -7,7 +7,7 @@
 [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/Tavish9/any4lerobot)
 [![Python versions](https://img.shields.io/pypi/pyversions/lerobot)](https://www.python.org/downloads/)
 [![LeRobot Dataset](https://img.shields.io/badge/dynamic/json?url=https://api.github.com/repos/tavish9/any4lerobot/commits?per_page=1&query=$[0].commit.committer.date&label=LeRobot&color=blue)](https://github.com/huggingface/lerobot)
-[![LeRobot Dataset](https://img.shields.io/badge/LeRobot%20Dataset-v2.1-ff69b4.svg)](https://github.com/huggingface/lerobot/pull/711)
+[![LeRobot Dataset](https://img.shields.io/badge/LeRobot%20Dataset-v3.0-ff69b4.svg)](https://github.com/huggingface/lerobot/pull/1412)
 [![License](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
 
 </div>
@@ -18,16 +18,17 @@
 
 A curated collection of utilities for [LeRobot Projects](https://github.com/huggingface/lerobot), including data conversion scripts, preprocessing tools, training workflow helpers and etc..
 
-## 🚀 What's New <a><img width="35" height="20" src="https://user-images.githubusercontent.com/12782558/212848161-5e783dd6-11e8-4fe0-bbba-39ffb77730be.png"></a>
+## 📣 What's New <a><img width="35" height="20" src="https://user-images.githubusercontent.com/12782558/212848161-5e783dd6-11e8-4fe0-bbba-39ffb77730be.png"></a>
 
+- **\[2025.09.28\]** We have upgraded LeRobotDataset from v2.1 to v3.0! 🔥🔥🔥
 - **\[2025.06.27\]** We have supported Data Conversion from LIBERO to LeRobot! 🔥🔥🔥
 - **\[2025.05.16\]** We have supported Data Conversion from LeRobot to RLDS! 🔥🔥🔥
 - **\[2025.05.12\]** We have supported Data Conversion from RoboMIND to LeRobot! 🔥🔥🔥
 - **\[2025.04.15\]** We add Dataset Merging Tool for merging multi-source lerobot datasets! 🔥🔥🔥
-- **\[2025.04.14\]** We have supported Data Conversion from AgiBotWorld to LeRobot! 🔥🔥🔥
 <details>
 <summary>More News</summary>
 
+- **\[2025.04.14\]** We have supported Data Conversion from AgiBotWorld to LeRobot! 🔥🔥🔥
 - **\[2025.04.11\]** We change the repo from `openx2lerobot` to `any4lerobot`, making a ​​universal toolbox for LeRobot​​! 🔥🔥🔥
 - **\[2025.02.19\]** We have supported Data Conversion from Open X-Embodiment to LeRobot! 🔥🔥🔥
 </details>

agibot2lerobot/agibot_h5.py

@@ -1,10 +1,7 @@
 import argparse
 import gc
 import shutil
-from concurrent.futures import (
+import tempfile
-    ThreadPoolExecutor,
-    as_completed,
-)
 from pathlib import Path
 
 import numpy as np
@@ -13,105 +10,13 @@ import torch
 from agibot_utils.agibot_utils import get_task_info, load_local_dataset
 from agibot_utils.config import AgiBotWorld_TASK_TYPE
 from agibot_utils.lerobot_utils import compute_episode_stats, generate_features_from_config
-from lerobot.datasets.compute_stats import aggregate_stats
+from lerobot.datasets.lerobot_dataset import LeRobotDataset
-from lerobot.datasets.lerobot_dataset import LeRobotDataset, LeRobotDatasetMetadata
+from lerobot.datasets.utils import validate_episode_buffer, validate_frame
-from lerobot.datasets.utils import (
-    check_timestamps_sync,
-    get_episode_data_index,
-    validate_episode_buffer,
-    validate_frame,
-    write_episode,
-    write_episode_stats,
-    write_info,
-)
-from lerobot.datasets.video_utils import get_safe_default_codec
 from ray.runtime_env import RuntimeEnv
 
 
-class AgiBotDatasetMetadata(LeRobotDatasetMetadata):
-    def save_episode(
-        self,
-        episode_index: int,
-        episode_length: int,
-        episode_tasks: list[str],
-        episode_stats: dict[str, dict],
-        action_config: list[dict],
-    ) -> None:
-        self.info["total_episodes"] += 1
-        self.info["total_frames"] += episode_length
-
-        chunk = self.get_episode_chunk(episode_index)
-        if chunk >= self.total_chunks:
-            self.info["total_chunks"] += 1
-
-        self.info["splits"] = {"train": f"0:{self.info['total_episodes']}"}
-        self.info["total_videos"] += len(self.video_keys)
-        if len(self.video_keys) > 0:
-            self.update_video_info()
-
-        write_info(self.info, self.root)
-
-        episode_dict = {
-            "episode_index": episode_index,
-            "tasks": episode_tasks,
-            "length": episode_length,
-            "action_config": action_config,
-        }
-        self.episodes[episode_index] = episode_dict
-        write_episode(episode_dict, self.root)
-
-        self.episodes_stats[episode_index] = episode_stats
-        self.stats = aggregate_stats([self.stats, episode_stats]) if self.stats else episode_stats
-        write_episode_stats(episode_index, episode_stats, self.root)
-
-
 class AgiBotDataset(LeRobotDataset):
-    @classmethod
+    def add_frame(self, frame: dict) -> None:
-    def create(
-        cls,
-        repo_id: str,
-        fps: int,
-        features: dict,
-        root: str | Path | None = None,
-        robot_type: str | None = None,
-        use_videos: bool = True,
-        tolerance_s: float = 1e-4,
-        image_writer_processes: int = 0,
-        image_writer_threads: int = 0,
-        video_backend: str | None = None,
-    ) -> "LeRobotDataset":
-        """Create a LeRobot Dataset from scratch in order to record data."""
-        obj = cls.__new__(cls)
-        obj.meta = AgiBotDatasetMetadata.create(
-            repo_id=repo_id,
-            fps=fps,
-            robot_type=robot_type,
-            features=features,
-            root=root,
-            use_videos=use_videos,
-        )
-        obj.repo_id = obj.meta.repo_id
-        obj.root = obj.meta.root
-        obj.revision = None
-        obj.tolerance_s = tolerance_s
-        obj.image_writer = None
-
-        if image_writer_processes or image_writer_threads:
-            obj.start_image_writer(image_writer_processes, image_writer_threads)
-
-        # TODO(aliberts, rcadene, alexander-soare): Merge this with OnlineBuffer/DataBuffer
-        obj.episode_buffer = obj.create_episode_buffer()
-
-        obj.episodes = None
-        obj.hf_dataset = obj.create_hf_dataset()
-        obj.image_transforms = None
-        obj.delta_timestamps = None
-        obj.delta_indices = None
-        obj.episode_data_index = None
-        obj.video_backend = video_backend if video_backend is not None else get_safe_default_codec()
-        return obj
-
-    def add_frame(self, frame: dict, task: str, timestamp: float | None = None) -> None:
         """
         This function only adds the frame to the episode_buffer. Apart from images — which are written in a
         temporary directory — nothing is written to disk. To save those frames, the 'save_episode()' method
@@ -130,11 +35,10 @@ class AgiBotDataset(LeRobotDataset):
 
         # Automatically add frame_index and timestamp to episode buffer
         frame_index = self.episode_buffer["size"]
-        if timestamp is None:
+        timestamp = frame.pop("timestamp") if "timestamp" in frame else frame_index / self.fps
-            timestamp = frame_index / self.fps
         self.episode_buffer["frame_index"].append(frame_index)
         self.episode_buffer["timestamp"].append(timestamp)
-        self.episode_buffer["task"].append(task)
+        self.episode_buffer["task"].append(frame.pop("task"))  # Remove task from frame after processing
 
         # Add frame features to episode_buffer
         for key, value in frame.items():
@@ -156,8 +60,7 @@ class AgiBotDataset(LeRobotDataset):
                 save the current episode in self.episode_buffer, which is filled with 'add_frame'. Defaults to
                 None.
         """
-        if not episode_data:
+        episode_buffer = episode_data if episode_data is not None else self.episode_buffer
-            episode_buffer = self.episode_buffer
 
         validate_episode_buffer(episode_buffer, self.meta.total_episodes, self.features)
 
@@ -170,11 +73,8 @@ class AgiBotDataset(LeRobotDataset):
         episode_buffer["index"] = np.arange(self.meta.total_frames, self.meta.total_frames + episode_length)
         episode_buffer["episode_index"] = np.full((episode_length,), episode_index)
 
-        # Add new tasks to the tasks dictionary
+        # Update tasks and task indices with new tasks if any
-        for task in episode_tasks:
+        self.meta.save_episode_tasks(episode_tasks)
-            task_index = self.meta.get_task_index(task)
-            if task_index is None:
-                self.meta.add_task(task)
 
         # Given tasks in natural language, find their corresponding task indices
         episode_buffer["task_index"] = np.array([self.meta.get_task_index(task) for task in tasks])
@@ -187,31 +87,46 @@ class AgiBotDataset(LeRobotDataset):
             episode_buffer[key] = np.stack(episode_buffer[key]).squeeze()
 
         for key in self.meta.video_keys:
-            video_path = self.root / self.meta.get_video_file_path(episode_index, key)
+            episode_buffer[key] = str(videos[key])  # PosixPath -> str
-            episode_buffer[key] = str(video_path)  # PosixPath -> str
-            video_path.parent.mkdir(parents=True, exist_ok=True)
-            shutil.copyfile(videos[key], video_path)
 
         ep_stats = compute_episode_stats(episode_buffer, self.features)
 
-        self._save_episode_table(episode_buffer, episode_index)
+        ep_metadata = self._save_episode_data(episode_buffer)
+        has_video_keys = len(self.meta.video_keys) > 0
+        use_batched_encoding = self.batch_encoding_size > 1
+
+        self.current_videos = videos
+        if has_video_keys and not use_batched_encoding:
+            for video_key in self.meta.video_keys:
+                ep_metadata.update(self._save_episode_video(video_key, episode_index))
 
         # `meta.save_episode` be executed after encoding the videos
         # add action_config to current episode
-        self.meta.save_episode(episode_index, episode_length, episode_tasks, ep_stats, action_config)
+        ep_metadata.update({"action_config": action_config})
+        self.meta.save_episode(episode_index, episode_length, episode_tasks, ep_stats, ep_metadata)
 
-        ep_data_index = get_episode_data_index(self.meta.episodes, [episode_index])
+        if has_video_keys and use_batched_encoding:
-        ep_data_index_np = {k: t.numpy() for k, t in ep_data_index.items()}
+            # Check if we should trigger batch encoding
-        check_timestamps_sync(
+            self.episodes_since_last_encoding += 1
-            episode_buffer["timestamp"],
+            if self.episodes_since_last_encoding == self.batch_encoding_size:
-            episode_buffer["episode_index"],
+                start_ep = self.num_episodes - self.batch_encoding_size
-            ep_data_index_np,
+                end_ep = self.num_episodes
-            self.fps,
+                self._batch_save_episode_video(start_ep, end_ep)
-            self.tolerance_s,
+                self.episodes_since_last_encoding = 0
-        )
 
-        if not episode_data:  # Reset the buffer
+        if not episode_data:
-            self.episode_buffer = self.create_episode_buffer()
+            # Reset episode buffer and clean up temporary images (if not already deleted during video encoding)
+            self.clear_episode_buffer(delete_images=len(self.meta.image_keys) > 0)
+
+    def _encode_temporary_episode_video(self, video_key: str, episode_index: int) -> Path:
+        """
+        Use ffmpeg to convert frames stored as png into mp4 videos.
+        Note: `encode_video_frames` is a blocking call. Making it asynchronous shouldn't speedup encoding,
+        since video encoding with ffmpeg is already using multithreading.
+        """
+        temp_path = Path(tempfile.mkdtemp(dir=self.root)) / f"{video_key}_{episode_index:03d}.mp4"
+        shutil.copy(self.current_videos[video_key], temp_path)
+        return temp_path
 
 
 def get_all_tasks(src_path: Path, output_path: Path):
@@ -221,7 +136,7 @@ def get_all_tasks(src_path: Path, output_path: Path):
         yield (json_file, local_dir.resolve())
 
 
-def save_as_lerobot_dataset(agibot_world_config, task: tuple[Path, Path], num_threads, save_depth, debug):
+def save_as_lerobot_dataset(agibot_world_config, task: tuple[Path, Path], save_depth):
     json_file, local_dir = task
     print(f"processing {json_file.stem}, saving to {local_dir}")
     src_path = json_file.parent.parent
@@ -252,7 +167,6 @@ def save_as_lerobot_dataset(agibot_world_config, task: tuple[Path, Path], num_th
 
     all_subdir_eids = sorted([int(Path(path).name) for path in all_subdir])
 
-    if debug or not save_depth:
     for eid in all_subdir_eids:
         if eid not in task_info:
             print(f"{json_file.stem}, episode_{eid} not in task_info.json, skipping...")
@@ -271,7 +185,8 @@ def save_as_lerobot_dataset(agibot_world_config, task: tuple[Path, Path], num_th
             continue
 
         for frame_data in frames:
-                dataset.add_frame(frame_data, task_instruction)
+            frame_data["task"] = task_instruction
+            dataset.add_frame(frame_data)
         try:
             dataset.save_episode(videos=videos, action_config=action_config)
         except Exception as e:
@@ -280,42 +195,6 @@ def save_as_lerobot_dataset(agibot_world_config, task: tuple[Path, Path], num_th
             continue
         gc.collect()
         print(f"process done for {json_file.stem}, episode_id {eid}, len {len(frames)}")
-    else:
-        with ThreadPoolExecutor(max_workers=num_threads) as executor:
-            futures = []
-            for eid in all_subdir_eids:
-                if eid not in task_info:
-                    print(f"{json_file.stem}, episode_{eid} not in task_info.json, skipping...")
-                    continue
-                futures.append(
-                    executor.submit(
-                        load_local_dataset,
-                        eid,
-                        src_path=src_path,
-                        task_id=task_id,
-                        save_depth=save_depth,
-                        AgiBotWorld_CONFIG=agibot_world_config,
-                    )
-                )
-
-            for raw_dataset in as_completed(futures):
-                eid, frames, videos = raw_dataset.result()
-                if not all([video_path.exists() for video_path in videos.values()]):
-                    print(f"{json_file.stem}, episode_{eid}: some of the videos does not exist, skipping...")
-                    continue
-                action_config = task_info[eid]["label_info"]["action_config"]
-                for frame_data in frames:
-                    dataset.add_frame(frame_data, task_instruction)
-                try:
-                    dataset.save_episode(videos=videos, action_config=action_config)
-                except Exception as e:
-                    print(
-                        f"{json_file.stem}, episode_{eid}: there are some corrupted mp4s\nException details: {str(e)}"
-                    )
-                    dataset.episode_buffer = None
-                    continue
-                gc.collect()
-                print(f"process done for {json_file.stem}, episode_id {eid}, len {len(frames)}")
 
 
 def main(
@@ -324,7 +203,6 @@ def main(
     eef_type: str,
     task_ids: list,
     cpus_per_task: int,
-    num_threads_per_task: int,
     save_depth: bool,
     debug: bool = False,
 ):
@@ -345,14 +223,10 @@ def main(
         tasks = filter(lambda task: task[0].stem in task_ids, tasks)
 
     if debug:
-        save_as_lerobot_dataset(agibot_world_config, next(tasks), num_threads_per_task, save_depth, debug)
+        save_as_lerobot_dataset(agibot_world_config, next(tasks), save_depth)
     else:
         runtime_env = RuntimeEnv(
-            env_vars={
+            env_vars={"HDF5_USE_FILE_LOCKING": "FALSE", "HF_DATASETS_DISABLE_PROGRESS_BARS": "TRUE"}
-                "HDF5_USE_FILE_LOCKING": "FALSE",
-                "HF_DATASETS_DISABLE_PROGRESS_BARS": "TRUE",
-                "LD_PRELOAD": str(Path(__file__).resolve().parent / "libtcmalloc.so.4.5.3"),
-            }
         )
         ray.init(runtime_env=runtime_env)
         resources = ray.available_resources()
@@ -363,9 +237,7 @@ def main(
         remote_task = ray.remote(save_as_lerobot_dataset).options(num_cpus=cpus_per_task)
         futures = []
         for task in tasks:
-            futures.append(
+            futures.append((task[0].stem, remote_task.remote(agibot_world_config, task, save_depth)))
-                (task[0].stem, remote_task.remote(agibot_world_config, task, num_threads_per_task, save_depth, debug))
-            )
 
         for task, future in futures:
             try:
@@ -385,7 +257,6 @@ if __name__ == "__main__":
     parser.add_argument("--eef-type", type=str, choices=["gripper", "dexhand", "tactile"], default="gripper")
     parser.add_argument("--task-ids", type=str, nargs="+", help="task_327 task_351 ...", default=[])
     parser.add_argument("--cpus-per-task", type=int, default=3)
-    parser.add_argument("--num-threads-per-task", type=int, default=2)
     parser.add_argument("--save-depth", action="store_true")
     parser.add_argument("--debug", action="store_true")
     args = parser.parse_args()

libero2lerobot/libero_h5.py

@@ -4,20 +4,27 @@ import re
 import shutil
 from pathlib import Path
 
-import numpy as np
 import pandas as pd
 import ray
 from datatrove.executor import LocalPipelineExecutor, RayPipelineExecutor
 from datatrove.pipeline.base import PipelineStep
+from lerobot.datasets.aggregate import (
+    aggregate_data,
+    aggregate_metadata,
+    aggregate_stats,
+    aggregate_videos,
+    validate_all_metadata,
+)
 from lerobot.datasets.lerobot_dataset import LeRobotDataset, LeRobotDatasetMetadata
 from lerobot.datasets.utils import (
-    write_episode,
+    DEFAULT_CHUNK_SIZE,
-    write_episode_stats,
+    DEFAULT_DATA_FILE_SIZE_IN_MB,
+    DEFAULT_VIDEO_FILE_SIZE_IN_MB,
     write_info,
-    write_task,
+    write_stats,
+    write_tasks,
 )
 from libero_utils.config import LIBERO_FEATURES
-from libero_utils.lerobot_utils import validate_all_metadata
 from libero_utils.libero_utils import load_local_episodes
 from ray.runtime_env import RuntimeEnv
 from tqdm import tqdm
@@ -63,178 +70,75 @@ class SaveLerobotDataset(PipelineStep):
         for episode_index, episode_data in enumerate(raw_dataset):
             with self.track_time("saving episode"):
                 for frame_data in episode_data:
-                    dataset.add_frame(
+                    frame_data["task"] = task_instruction
-                        frame_data,
+                    dataset.add_frame(frame_data)
-                        task=task_instruction,
-                    )
                 dataset.save_episode()
                 logger.info(f"process done for {dataset.repo_id}, episode {episode_index}, len {len(episode_data)}")
 
 
-class AggregateDatasets(PipelineStep):
+def create_aggr_dataset(raw_dirs: list[Path], aggregated_dir: Path):
-    name = "Aggregate Datasets"
-    type = "libero2lerobot"
-
-    def __init__(self, raw_dirs: list[Path], aggregated_dir: Path):
-        super().__init__()
-        self.raw_dirs = raw_dirs
-        self.aggregated_dir = aggregated_dir
-
-        self.create_aggr_dataset()
-
-    def create_aggr_dataset(self):
     logger = setup_logger()
 
-        all_metadata = [LeRobotDatasetMetadata("", root=raw_dir) for raw_dir in self.raw_dirs]
+    all_metadata = [LeRobotDatasetMetadata("", root=raw_dir) for raw_dir in raw_dirs]
 
     fps, robot_type, features = validate_all_metadata(all_metadata)
 
-        if self.aggregated_dir.exists():
+    if aggregated_dir.exists():
-            shutil.rmtree(self.aggregated_dir)
+        shutil.rmtree(aggregated_dir)
 
     aggr_meta = LeRobotDatasetMetadata.create(
-            repo_id=f"{self.aggregated_dir.parent.name}/{self.aggregated_dir.name}",
+        repo_id=f"{aggregated_dir.parent.name}/{aggregated_dir.name}",
-            root=self.aggregated_dir,
+        root=aggregated_dir,
         fps=fps,
         robot_type=robot_type,
         features=features,
     )
 
-        datasets_task_index_to_aggr_task_index = {}
+    video_keys = [key for key in features if features[key]["dtype"] == "video"]
-        aggr_task_index = 0
+    unique_tasks = pd.concat([m.tasks for m in all_metadata]).index.unique()
-        for dataset_index, meta in enumerate(tqdm(all_metadata, desc="Aggregate tasks index")):
+    aggr_meta.tasks = pd.DataFrame({"task_index": range(len(unique_tasks))}, index=unique_tasks)
-            task_index_to_aggr_task_index = {}
 
-            for task_index, task in meta.tasks.items():
+    meta_idx = {"chunk": 0, "file": 0}
-                if task not in aggr_meta.task_to_task_index:
+    data_idx = {"chunk": 0, "file": 0}
-                    # add the task to aggr tasks mappings
+    videos_idx = {key: {"chunk": 0, "file": 0, "latest_duration": 0, "episode_duration": 0} for key in video_keys}
-                    aggr_meta.tasks[aggr_task_index] = task
-                    aggr_meta.task_to_task_index[task] = aggr_task_index
-                    aggr_task_index += 1
 
-                task_index_to_aggr_task_index[task_index] = aggr_meta.task_to_task_index[task]
+    aggr_meta.episodes = {}
 
-            datasets_task_index_to_aggr_task_index[dataset_index] = task_index_to_aggr_task_index
+    for src_meta in tqdm(all_metadata, desc="Copy data and videos"):
+        videos_idx = aggregate_videos(
+            src_meta, aggr_meta, videos_idx, DEFAULT_VIDEO_FILE_SIZE_IN_MB, DEFAULT_CHUNK_SIZE
+        )
+        data_idx = aggregate_data(src_meta, aggr_meta, data_idx, DEFAULT_DATA_FILE_SIZE_IN_MB, DEFAULT_CHUNK_SIZE)
 
-        datasets_aggr_episode_index_shift = {}
+        meta_idx = aggregate_metadata(src_meta, aggr_meta, meta_idx, data_idx, videos_idx)
-        datasets_aggr_index_shift = {}
-        for dataset_index, meta in enumerate(tqdm(all_metadata, desc="Aggregate episodes and global index")):
-            datasets_aggr_episode_index_shift[dataset_index] = aggr_meta.total_episodes
-            datasets_aggr_index_shift[dataset_index] = aggr_meta.total_frames
 
-            # populate episodes
+        aggr_meta.info["total_episodes"] += src_meta.total_episodes
-            for episode_index, episode_dict in meta.episodes.items():
+        aggr_meta.info["total_frames"] += src_meta.total_frames
-                aggr_episode_index = episode_index + aggr_meta.total_episodes
-                episode_dict["episode_index"] = aggr_episode_index
-                aggr_meta.episodes[aggr_episode_index] = episode_dict
 
-            # populate episodes_stats
+    logger.info("write tasks")
-            for episode_index, episode_stats in meta.episodes_stats.items():
+    write_tasks(aggr_meta.tasks, aggr_meta.root)
-                aggr_episode_index = episode_index + aggr_meta.total_episodes
+
-                episode_stats["index"].update(
+    logger.info("write info")
+    aggr_meta.info.update(
         {
-                        "min": episode_stats["index"]["min"] + aggr_meta.total_frames,
+            "total_tasks": len(aggr_meta.tasks),
-                        "max": episode_stats["index"]["max"] + aggr_meta.total_frames,
+            "total_episodes": sum(m.total_episodes for m in all_metadata),
-                        "mean": episode_stats["index"]["mean"] + aggr_meta.total_frames,
+            "total_frames": sum(m.total_frames for m in all_metadata),
+            "splits": {"train": f"0:{sum(m.total_episodes for m in all_metadata)}"},
         }
     )
-                episode_stats["episode_index"].update(
-                    {
-                        "min": np.array([aggr_episode_index]),
-                        "max": np.array([aggr_episode_index]),
-                        "mean": np.array([aggr_episode_index]),
-                    }
-                )
-                df = pd.read_parquet(meta.root / meta.get_data_file_path(episode_index))
-                df["task_index"] = df["task_index"].map(datasets_task_index_to_aggr_task_index[dataset_index])
-                episode_stats["task_index"].update(
-                    {
-                        "min": np.array([df["task_index"].min()]),
-                        "max": np.array([df["task_index"].max()]),
-                        "mean": np.array([df["task_index"].mean()]),
-                        "std": np.array([df["task_index"].std()]),
-                    }
-                )
-                aggr_meta.episodes_stats[aggr_episode_index] = episode_stats
-
-            # populate info
-            aggr_meta.info["total_episodes"] += meta.total_episodes
-            aggr_meta.info["total_frames"] += meta.total_frames
-            aggr_meta.info["total_videos"] += len(aggr_meta.video_keys) * meta.total_episodes
-
-        logger.info("Write meta data")
-        aggr_meta.info["total_tasks"] = len(aggr_meta.tasks)
-        aggr_meta.info["total_chunks"] = aggr_meta.get_episode_chunk(aggr_meta.total_episodes - 1) + 1
-        aggr_meta.info["splits"] = {"train": f"0:{aggr_meta.info['total_episodes']}"}
-
-        # create a new episodes jsonl with updated episode_index using write_episode
-        for episode_dict in tqdm(aggr_meta.episodes.values(), desc="Write episodes info"):
-            write_episode(episode_dict, aggr_meta.root)
-
-        # create a new episode_stats jsonl with updated episode_index using write_episode_stats
-        for episode_index, episode_stats in tqdm(aggr_meta.episodes_stats.items(), desc="Write episodes stats info"):
-            write_episode_stats(episode_index, episode_stats, aggr_meta.root)
-
-        # create a new task jsonl with updated episode_index using write_task
-        for task_index, task in tqdm(aggr_meta.tasks.items(), desc="Write tasks info"):
-            write_task(task_index, task, aggr_meta.root)
-
     write_info(aggr_meta.info, aggr_meta.root)
 
-        self.datasets_task_index_to_aggr_task_index = datasets_task_index_to_aggr_task_index
+    logger.info("write stats")
-        self.datasets_aggr_episode_index_shift = datasets_aggr_episode_index_shift
+    aggr_meta.stats = aggregate_stats([m.stats for m in all_metadata])
-        self.datasets_aggr_index_shift = datasets_aggr_index_shift
+    write_stats(aggr_meta.stats, aggr_meta.root)
 
-        logger.info("Meta data done writing")
 
-    def run(self, data=None, rank: int = 0, world_size: int = 1):
+def delete_temp_data(temp_dirs: list[Path]):
     logger = setup_logger()
-
+    logger.info("Delete temp data_dir")
-        dataset_index = rank
+    for temp_dir in temp_dirs:
-        aggr_meta = LeRobotDatasetMetadata("", root=self.aggregated_dir)
+        shutil.rmtree(temp_dir)
-        meta = LeRobotDatasetMetadata("", root=self.raw_dirs[dataset_index])
-        aggr_episode_index_shift = self.datasets_aggr_episode_index_shift[dataset_index]
-        aggr_index_shift = self.datasets_aggr_index_shift[dataset_index]
-        task_index_to_aggr_task_index = self.datasets_task_index_to_aggr_task_index[dataset_index]
-
-        with self.track_time("aggregating dataset"):
-            logger.info("Copy data")
-            for episode_index in range(meta.total_episodes):
-                aggr_episode_index = episode_index + aggr_episode_index_shift
-                data_path = meta.root / meta.get_data_file_path(episode_index)
-                aggr_data_path = aggr_meta.root / aggr_meta.get_data_file_path(aggr_episode_index)
-                aggr_data_path.parent.mkdir(parents=True, exist_ok=True)
-
-                # update index, episode_index and task_index
-                df = pd.read_parquet(data_path)
-                df["index"] += aggr_index_shift
-                df["episode_index"] += aggr_episode_index_shift
-                df["task_index"] = df["task_index"].map(task_index_to_aggr_task_index)
-                df.to_parquet(aggr_data_path)
-
-            logger.info("Copy videos")
-            for episode_index in range(meta.total_episodes):
-                aggr_episode_index = episode_index + aggr_episode_index_shift
-                for vid_key in meta.video_keys:
-                    video_path = meta.root / meta.get_video_file_path(episode_index, vid_key)
-                    aggr_video_path = aggr_meta.root / aggr_meta.get_video_file_path(aggr_episode_index, vid_key)
-                    aggr_video_path.parent.mkdir(parents=True, exist_ok=True)
-                    shutil.copy(video_path, aggr_video_path)
-
-
-class DeleteTempData(PipelineStep):
-    name = "Delete Temp Data"
-    type = "libero2lerobot"
-
-    def __init__(self, temp_dirs: list[Path]):
-        super().__init__()
-        self.temp_dirs = temp_dirs
-
-    def run(self, data=None, rank: int = 0, world_size: int = 1):
-        logger = setup_logger()
-
-        logger.info(f"Delete temp data {self.temp_dirs[rank]}")
-        shutil.rmtree(self.temp_dirs[rank])
 
 
 def main(
@@ -244,8 +148,7 @@ def main(
     cpus_per_task: int,
     tasks_per_job: int,
     workers: int,
-    resume_from_save: Path = None,
+    resume_dir: Path = None,
-    resume_from_aggregate: Path = None,
     debug: bool = False,
     repo_id: str = None,
     push_to_hub: bool = False,
@@ -302,16 +205,9 @@ def main(
         **({"cpus_per_task": cpus_per_task, "tasks_per_job": tasks_per_job} if executor is RayPipelineExecutor else {}),
     }
 
-    executor(pipeline=[SaveLerobotDataset(tasks)], **executor_config, logging_dir=resume_from_save).run()
+    executor(pipeline=[SaveLerobotDataset(tasks)], **executor_config, logging_dir=resume_dir).run()
-    executor(
+    create_aggr_dataset([task[1] for task in tasks], aggregate_output_path)
-        pipeline=[DeleteTempData([task[1] for task in tasks])],
+    delete_temp_data([task[1] for task in tasks])
-        **executor_config,
-        depends=executor(
-            pipeline=[AggregateDatasets([task[1] for task in tasks], aggregate_output_path)],
-            **executor_config,
-            logging_dir=resume_from_aggregate,
-        ),
-    ).run()
 
     for task in tasks:
         shutil.rmtree(task[1].parent, ignore_errors=True)
@@ -340,8 +236,7 @@ if __name__ == "__main__":
     parser.add_argument("--cpus-per-task", type=int, default=1)
     parser.add_argument("--tasks-per-job", type=int, default=1, help="number of concurrent tasks per job, only used for ray")
     parser.add_argument("--workers", type=int, default=-1, help="number of concurrent jobs to run")
-    parser.add_argument("--resume-from-save", type=Path, help="logs directory to resume from save step")
+    parser.add_argument("--resume-dir", type=Path, help="logs directory to resume")
-    parser.add_argument("--resume-from-aggregate", type=Path, help="logs directory to resume from aggregate step")
     parser.add_argument("--debug", action="store_true")
     parser.add_argument("--repo-id", type=str, help="required when push-to-hub is True")
     parser.add_argument("--push-to-hub", action="store_true", help="upload to hub")

libero2lerobot/libero_utils/lerobot_utils.py

@@ -1,23 +0,0 @@
-import tqdm
-from lerobot.datasets.lerobot_dataset import LeRobotDatasetMetadata
-
-
-def validate_all_metadata(all_metadata: list[LeRobotDatasetMetadata]):
-    """
-    implemented by @Cadene
-    """
-    # validate same fps, robot_type, features
-
-    fps = all_metadata[0].fps
-    robot_type = all_metadata[0].robot_type
-    features = all_metadata[0].features
-
-    for meta in tqdm.tqdm(all_metadata, desc="Validate all meta data"):
-        if fps != meta.fps:
-            raise ValueError(f"Same fps is expected, but got fps={meta.fps} instead of {fps}.")
-        if robot_type != meta.robot_type:
-            raise ValueError(f"Same robot_type is expected, but got robot_type={meta.robot_type} instead of {robot_type}.")
-        if features != meta.features:
-            raise ValueError(f"Same features is expected, but got features={meta.features} instead of {features}.")
-
-    return fps, robot_type, features

openx2lerobot/openx_rlds.py

@@ -37,8 +37,8 @@ from pathlib import Path
 import numpy as np
 import tensorflow as tf
 import tensorflow_datasets as tfds
-from lerobot.constants import HF_LEROBOT_HOME
 from lerobot.datasets.lerobot_dataset import LeRobotDataset
+from lerobot.utils.constants import HF_LEROBOT_HOME
 from oxe_utils.configs import OXE_DATASET_CONFIGS, ActionEncoding, StateEncoding
 from oxe_utils.transforms import OXE_STANDARDIZATION_TRANSFORMS
 
@@ -147,8 +147,8 @@ def save_as_lerobot_dataset(lerobot_dataset: LeRobotDataset, raw_dataset: tf.dat
                     **image_dict,
                     "observation.state": traj["proprio"][i],
                     "action": traj["action"][i],
+                    "task": traj["task"][0].decode(),
                 },
-                task=traj["task"][0].decode(),
             )
         lerobot_dataset.save_episode()
 

robomind2lerobot/robomind_h5.py

@@ -8,18 +8,9 @@ from pathlib import Path
 import numpy as np
 import pandas as pd
 import ray
-import torch
 from lerobot.datasets.compute_stats import aggregate_stats
 from lerobot.datasets.lerobot_dataset import LeRobotDataset, LeRobotDatasetMetadata
-from lerobot.datasets.utils import (
+from lerobot.datasets.utils import flatten_dict, validate_episode_buffer, write_info, write_stats
-    check_timestamps_sync,
-    get_episode_data_index,
-    validate_episode_buffer,
-    validate_frame,
-    write_episode,
-    write_episode_stats,
-    write_info,
-)
 from lerobot.datasets.video_utils import get_safe_default_codec
 from ray.runtime_env import RuntimeEnv
 from robomind_uitls.configs import ROBOMIND_CONFIG
@@ -37,38 +28,31 @@ class RoboMINDDatasetMetadata(LeRobotDatasetMetadata):
         episode_length: int,
         episode_tasks: list[str],
         episode_stats: dict[str, dict],
-        action_config: dict[str, str | dict],
+        episode_metadata: dict,
     ) -> None:
+        episode_dict = {
+            "episode_index": episode_index,
+            "tasks": episode_tasks,
+            "length": episode_length,
+        }
+        episode_dict.update(episode_metadata)
+        episode_dict.update(flatten_dict({"stats": episode_stats}))
+        self._save_episode_metadata(episode_dict)
+
+        # Update info
         self.info["total_episodes"] += 1
         self.info["total_frames"] += episode_length
-
+        self.info["total_tasks"] = len(self.tasks)
-        chunk = self.get_episode_chunk(episode_index)
-        if chunk >= self.total_chunks:
-            self.info["total_chunks"] += 1
-
         if split == "train":
             self.info["splits"]["train"] = f"0:{self.info['total_episodes']}"
             self.train_count = self.info["total_episodes"]
         elif "val" in split:
             self.info["splits"]["validation"] = f"{self.train_count}:{self.info['total_episodes']}"
-        self.info["total_videos"] += len(self.video_keys)
-        if len(self.video_keys) > 0:
-            self.update_video_info()
 
         write_info(self.info, self.root)
 
-        episode_dict = {
+        self.stats = aggregate_stats([self.stats, episode_stats]) if self.stats is not None else episode_stats
-            "episode_index": episode_index,
+        write_stats(self.stats, self.root)
-            "tasks": episode_tasks,
-            "length": episode_length,
-            **({"action_config": action_config} if action_config else {}),
-        }
-        self.episodes[episode_index] = episode_dict
-        write_episode(episode_dict, self.root)
-
-        self.episodes_stats[episode_index] = episode_stats
-        self.stats = aggregate_stats([self.stats, episode_stats]) if self.stats else episode_stats
-        write_episode_stats(episode_index, episode_stats, self.root)
 
 
 class RoboMINDDataset(LeRobotDataset):
@@ -85,6 +69,7 @@ class RoboMINDDataset(LeRobotDataset):
         image_writer_processes: int = 0,
         image_writer_threads: int = 0,
         video_backend: str | None = None,
+        batch_encoding_size: int = 1,
     ) -> "LeRobotDataset":
         """Create a LeRobot Dataset from scratch in order to record data."""
         obj = cls.__new__(cls)
@@ -101,6 +86,8 @@ class RoboMINDDataset(LeRobotDataset):
         obj.revision = None
         obj.tolerance_s = tolerance_s
         obj.image_writer = None
+        obj.batch_encoding_size = batch_encoding_size
+        obj.episodes_since_last_encoding = 0
 
         if image_writer_processes or image_writer_threads:
             obj.start_image_writer(image_writer_processes, image_writer_threads)
@@ -113,53 +100,10 @@ class RoboMINDDataset(LeRobotDataset):
         obj.image_transforms = None
         obj.delta_timestamps = None
         obj.delta_indices = None
-        obj.episode_data_index = None
         obj.video_backend = video_backend if video_backend is not None else get_safe_default_codec()
         return obj
 
-    def add_frame(self, frame: dict, task: str, timestamp: float | None = None) -> None:
+    def save_episode(self, split, action_config: dict, episode_data: dict | None = None) -> None:
-        """
-        This function only adds the frame to the episode_buffer. Apart from images — which are written in a
-        temporary directory — nothing is written to disk. To save those frames, the 'save_episode()' method
-        then needs to be called.
-        """
-        # Convert torch to numpy if needed
-        for name in frame:
-            if isinstance(frame[name], torch.Tensor):
-                frame[name] = frame[name].numpy()
-
-        validate_frame(frame, self.features)
-
-        if self.episode_buffer is None:
-            self.episode_buffer = self.create_episode_buffer()
-
-        # Automatically add frame_index and timestamp to episode buffer
-        frame_index = self.episode_buffer["size"]
-        if timestamp is None:
-            timestamp = frame_index / self.fps
-        self.episode_buffer["frame_index"].append(frame_index)
-        self.episode_buffer["timestamp"].append(timestamp)
-        self.episode_buffer["task"].append(task)
-
-        # Add frame features to episode_buffer
-        for key, value in frame.items():
-            if key not in self.features:
-                raise ValueError(f"An element of the frame is not in the features. '{key}' not in '{self.features.keys()}'.")
-
-            if self.features[key]["dtype"] in ["video"]:
-                img_path = self._get_image_file_path(
-                    episode_index=self.episode_buffer["episode_index"], image_key=key, frame_index=frame_index
-                )
-                if frame_index == 0:
-                    img_path.parent.mkdir(parents=True, exist_ok=True)
-                self._save_image(value, img_path)
-                self.episode_buffer[key].append(str(img_path))
-            else:
-                self.episode_buffer[key].append(value)
-
-        self.episode_buffer["size"] += 1
-
-    def save_episode(self, split, action_config: dict, episode_data: dict | None = None, keep_images: bool = False) -> None:
         """
         This will save to disk the current episode in self.episode_buffer.
 
@@ -168,8 +112,7 @@ class RoboMINDDataset(LeRobotDataset):
                 save the current episode in self.episode_buffer, which is filled with 'add_frame'. Defaults to
                 None.
         """
-        if not episode_data:
+        episode_buffer = episode_data if episode_data is not None else self.episode_buffer
-            episode_buffer = self.episode_buffer
 
         validate_episode_buffer(episode_buffer, self.meta.total_episodes, self.features)
 
@@ -182,11 +125,8 @@ class RoboMINDDataset(LeRobotDataset):
         episode_buffer["index"] = np.arange(self.meta.total_frames, self.meta.total_frames + episode_length)
         episode_buffer["episode_index"] = np.full((episode_length,), episode_index)
 
-        # Add new tasks to the tasks dictionary
+        # Update tasks and task indices with new tasks if any
-        for task in episode_tasks:
+        self.meta.save_episode_tasks(episode_tasks)
-            task_index = self.meta.get_task_index(task)
-            if task_index is None:
-                self.meta.add_task(task)
 
         # Given tasks in natural language, find their corresponding task indices
         episode_buffer["task_index"] = np.array([self.meta.get_task_index(task) for task in tasks])
@@ -199,35 +139,32 @@ class RoboMINDDataset(LeRobotDataset):
             episode_buffer[key] = np.stack(episode_buffer[key]).squeeze()
 
         self._wait_image_writer()
-        self._save_episode_table(episode_buffer, episode_index)
         ep_stats = compute_episode_stats(episode_buffer, self.features)
 
-        if len(self.meta.video_keys) > 0:
+        ep_metadata = self._save_episode_data(episode_buffer)
-            video_paths = self.encode_episode_videos(episode_index)
+        has_video_keys = len(self.meta.video_keys) > 0
-            for key in self.meta.video_keys:
+        use_batched_encoding = self.batch_encoding_size > 1
-                episode_buffer[key] = video_paths[key]
+
+        if has_video_keys and not use_batched_encoding:
+            for video_key in self.meta.video_keys:
+                ep_metadata.update(self._save_episode_video(video_key, episode_index))
 
         # `meta.save_episode` be executed after encoding the videos
-        self.meta.save_episode(split, episode_index, episode_length, episode_tasks, ep_stats, action_config)
+        ep_metadata.update({"action_config": action_config})
+        self.meta.save_episode(split, episode_index, episode_length, episode_tasks, ep_stats, ep_metadata)
 
-        ep_data_index = get_episode_data_index(self.meta.episodes, [episode_index])
+        if has_video_keys and use_batched_encoding:
-        ep_data_index_np = {k: t.numpy() for k, t in ep_data_index.items()}
+            # Check if we should trigger batch encoding
-        check_timestamps_sync(
+            self.episodes_since_last_encoding += 1
-            episode_buffer["timestamp"],
+            if self.episodes_since_last_encoding == self.batch_encoding_size:
-            episode_buffer["episode_index"],
+                start_ep = self.num_episodes - self.batch_encoding_size
-            ep_data_index_np,
+                end_ep = self.num_episodes
-            self.fps,
+                self._batch_save_episode_video(start_ep, end_ep)
-            self.tolerance_s,
+                self.episodes_since_last_encoding = 0
-        )
 
-        if not keep_images:
+        if not episode_data:
-            # delete images
+            # Reset episode buffer and clean up temporary images (if not already deleted during video encoding)
-            img_dir = self.root / "images"
+            self.clear_episode_buffer(delete_images=len(self.meta.image_keys) > 0)
-            if img_dir.is_dir():
-                shutil.rmtree(self.root / "images")
-
-        if not episode_data:  # Reset the buffer
-            self.episode_buffer = self.create_episode_buffer()
 
 
 def get_all_tasks(src_path: Path, output_path: Path, embodiment: str):
@@ -288,8 +225,11 @@ def save_as_lerobot_dataset(task: tuple[dict, Path, str], src_path, benchmark, e
             if status and len(raw_dataset) >= 50:
                 try:
                     for frame_data in raw_dataset:
-                        dataset.add_frame(frame_data, task_instruction)
+                        frame_data["task"] = task_instruction
-                    dataset.save_episode(split, action_config.get(episode_path.parent.parent.name, {}))
+                        dataset.add_frame(frame_data)
+                    dataset.save_episode(
+                        split, action_config.get(episode_path.parent.parent.name, {"task_summary": None, "steps": None})
+                    )
                     logging.info(f"process done for {path}, len {len(raw_dataset)}")
                 except Exception:
                     # [HACK]: not consistent image shape...
@@ -324,11 +264,7 @@ def main(
         save_as_lerobot_dataset(next(tasks), src_path, benchmark, embodiments[0], save_depth)
     else:
         runtime_env = RuntimeEnv(
-            env_vars={
+            env_vars={"HDF5_USE_FILE_LOCKING": "FALSE", "HF_DATASETS_DISABLE_PROGRESS_BARS": "TRUE"}
-                "HDF5_USE_FILE_LOCKING": "FALSE",
-                "HF_DATASETS_DISABLE_PROGRESS_BARS": "TRUE",
-                "LD_PRELOAD": str(Path(__file__).resolve().parent / "libtcmalloc.so.4.5.3"),
-            }
         )
         ray.init(runtime_env=runtime_env)
         resources = ray.available_resources()