GPT-6 Astra Tops EgoHome Benchmark for Egocentric Hand Understanding

1 reports archived1 independent sourcesupdated 11/11/2026, 05:57:06

Synthesis & Latest Updates

1 Sources Cross-Validated

Perceptron AI released EgoHome, an open benchmark for dense sub-second egocentric understanding of household-chore videos, with over 750 segments under 500ms and four annotation types: hand boxes, subtasks, hand actions and hand captions. GPT-6 Astra ranked first on the benchmark. The work targets the automatic labeling pipeline that underpins robot learning from human video, where mislabeled hand actions propagate errors into robot training.

LATEST/Perceptron AI released EgoHome, an open benchmark for dense sub-second egocentric understanding of household-chore videos, with over 750 segments under 500ms and four annotation types: hand boxes, subtasks, hand actions and hand captions. GPT-6 Astra ranked first on the benchmark. The work targets the automatic labeling pipeline that underpins robot learning from human video, where mislabeled hand actions propagate errors into robot training.

Hourly heat curve

2 fully observed hours

Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first

Rohan Paul (@rohanpaul_ai)T2Nov 11, 05:57·72 pts
GPT-6 Astra Tops EgoHome Benchmark for Egocentric Hand Understanding
- EgoHome targets sub-second egocentric understanding with over 750 segments under 500ms, far denser than typical video-understanding datasets
- Each clip carries four annotation types: hand boxes, subtasks, hand actions and hand captions; data, metrics and judge prompts are all open
- Its core use is validating auto-labeling quality for robot training video, where labeling errors propagate directly into wrong robot behavior
View Related Reports & Analysis Original Source