Datasets:
Sample Pack v0.1 — video + depth now available
Quick update on our kitchen manipulation preview dataset 👋
We upgraded the Hugging Face preview from static screenshots to Sample Pack v0.1 — actual video + depth for evaluation.
What's in the pack:
• task_01_cutting — egocentric (1080p) + side view (60fps) + overhead RGB + depth.hdf5 (float32 meters, RealSense D435I)
• task_02_stir_fry — egocentric (1080p) + side view (1080p)
Captured in a real commercial Chinese restaurant (Zhongshan), professional chef, multi-view aligned for manipulation / IL / VLA prototyping.
License: CC-BY-NC-4.0 (preview). Full 4K sources, longer clips, .bag, RLDS/LeRobot export — available on request.
Would love early feedback if anyone tries loading the depth or multi-view streams. andy@dynamicnova.com
(dataset · imitation-learning · kitchen · rgb-d)