GPT-6 Astra Tops EgoHome Benchmark for Egocentric Hand Understanding
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.
Related Entities
Full Report
Rohan Paul (@rohanpaul_ai) · 2026-11-10 21
GPT-6 Astra came out on top in EgoHome, a new benchmark from Perceptron AI, that tests whether AI can follow what human hands do in chore videos, the skill robots need to learn from people.
AI can tell you someone is washing dishes, but not what their hands are doing, and robots need the second part.
This kind of benchmarks are so important because Robotics training now happen on large amounts of human video, and AI is used to label those videos automatically (“left hand picks up sponge, right hand turns tap”). If those labels are wrong, the robots learn wrong.
↳ 引用 @perceptroninc (Perceptron AI): Introducing EgoHome: a new open benchmark for dense, sub-second egocentric understanding.
Human-annotated household-chore video with 4 annotation types on every clip: hand boxes, subtasks, hand actions, hand captions.
It features over 750 segments under 500ms.
Data, metrics, and judge prompts are all open. 🧵
