Open-source ecosystem for egocentric robot data: raw video in, a V-JEPA 2.1 model predicts every step, a review app lets people approve or fix them, and approved steps become training data that retrains the model. 92.5% frame accuracy on held-out video; review time per clip 40-50min → 5min
computer-vision robotics video-annotation data-pipeline annotation-tool robot-learning human-in-the-loop egocentric data-labeling physical-ai v-jepa model-assisted-labeling egocentric-10k robot-training
-
Updated
Oct 19, 2026 - Python