Reproduce “EgoIntrospect: An Egocentric Dataset and Benchmark for User-Centric Internal State Reasoning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zeyu Wang, Chang Liu, Eduardus Tjitrahardja, Yuntao Wang, Borislav Pavlov, Fangfei Gou, Jose Manuel Davila, Dai Shi and 17 more
EgoIntrospect introduces a dataset and benchmark for testing whether wearable assistants can infer wearers’ feelings, intentions, and memory needs. It pairs 180 hours of video, audio, gaze, motion, and physiological data from 60 participants with wearer labels, and evaluates 21 models on 15 participants. On four-choice recommendation selection, Gemini-3.1-pro-preview achieves 55.81% accuracy versus 25% chance, showing limits, not real-world readiness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.