A robot demonstration can lose its most useful half-second

A robot demonstration can lose its most useful half-second if a short tracking gap occurs during a critical task phase like cable insertion. While MEgoVista's evaluation penalizes missed detections, the impact of these gaps varies significantly based on their timing. The MEgo framework, which includes hand detection and temporal tracking before MANO-based 3D reconstruction, highlights the importance of examining coverage alongside motion quality. For small teams, evaluating pose error, detection coverage, and the longest tracking gap for each task phase (approach, contact, withdrawal) is recommended to ensure the usability of manipulation datasets.