FastUMI — Egocentric Data-Center Activity Demonstrations (LeRobot)
Overview
FastUMI — Egocentric Data-Center Activity Demonstrations (LeRobot) is a video dataset. It covers English. Typical applications include LLM pre-training, evaluation & benchmarking and robotics & embodied AI. Relevant sectors: Technology & IT, AI & Machine Learning and Manufacturing & Industrial. The listing mentions timestamps. Pricing is not published and is provided on request.
Primary applications
What the listing offers
- Listing mentions timestamps.
- A sample is indicated as available from the provider.
- Version 1.0 as published by the source.
Egocentric, multi-view human demonstrations of data-center / IT-infrastructure manipulation activities, captured with a FastUMI handheld egocentric rig (Universal Manipulation Interface), human operator. Each demonstration is packaged as a HuggingFace LeRobot v2.1 dataset (single episode per dataset) and is QC-complete. Four synchronized camera streams: head RGB (1280×720), left + right wrist RGB (1920×1080) in H.264, and a 16-bit head-depth stream (FFV1 lossless), at ~30 fps. The parquet carries frame timestamps + indices (video-centric), Useful for egocentric manipulation perception, multi-view (head+wrist) representation learning, VLA pretraining on human demonstrations, and bridging human-demonstration data into robot-learning pipelines.
Related datasets
More in Video & Physical AI →Wrapping a Gift - Egocentric Human Demonstration
A single continuous egocentric demonstration of a person wrapping a gift — handling wrapping paper, tape and scissors on a work surface.
- Language
- English
- Volume
- —
- Domain
- AI & Machine Learning +1
- Use case
- LLM pre-training