Skip to main content

dexdata

Data recording and replay utilities for Dexmate robots. Reads and writes MCAP episode data (robot state/action, RGB cameras, depth) with Protobuf serialization, and encodes cameras to video.

Install

pip install -e .            # core: MCAP read/write
pip install -e ".[convert]"   # adds torch + torchcodec for video_writer

Requires Python 3.11.

Layout

  • dexdata.mcap_utils — TeleopMcapWriter, TeleopMcapReader, Episode, EpisodeMetadata. Self-contained.
  • dexdata.video_writer — torch + torchcodec-backed parallel/serial video encoder (opt-in via [video] extra).

Quick read

from pathlib import Path
from dexdata.mcap_utils import TeleopMcapReader

reader = TeleopMcapReader(Path("/path/to/episode_dir"))
episode = reader.load_episode()
data = episode.to_numpy_dict()   # flat {key: np.ndarray} for training
fps = reader.metadata.record_rate_hz

An episode directory contains state_action.mcap, zero or more camera_*.mcap / depth_*.mcap, and info.json.

Scripts

  • scripts/example_mcap_read.py <episode_dir> — dump an episode to its numpy-dict form and print key shapes/dtypes.
  • scripts/mcap_to_video.py --episode_path <dir> — render a 2×2 camera grid to concatenated.mp4. FPS defaults to robot.record_rate_hz from info.json.
  • scripts/mcap_viz.py — Rerun-based episode visualization helper (RerunViz).

Protobuf schemas

Schemas live in src/dexdata/mcap_utils/proto/. Compiled *_pb2.py modules are checked in; regenerate with:

cd src/dexdata/mcap_utils/proto && ./compile_proto.sh

Metadata

Release files for dexdata 0.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distribution (wheel)

Table of built distributions (wheels) for dexdata 0.1.0
File Interpreter ABI Platform
dexdata-0.1.0-py3-none-any.whl Python 3 none any Details

Release files / dexdata-0.1.0-py3-none-any.whl

Download URL dexdata-0.1.0-py3-none-any.whl
Size 113.0 kB
Tags Python 3
SHA-256 checksum
How to use checksums
da4c716d181ea113678107a1d04fbcec6258b25026039473a1e98de02b5f6d8c
BLAKE2b-256 checksum
How to use checksums
788530b287e97af9e1d8fddc523ee93c59c0e8a5604541734beb4464ab437fa0
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on May 29, 2026.

Transparency log

Release history Release notifications | RSS feed

This release

0.1.0 This release

1 release file

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page