The First EgoCross Challenge at EgoVis 2026: Cross-Domain Egocentric Video Question Answering

2026-08-05

AI Summary

This work investigates the generalization capabilities of multimodal large language models (MLLMs) in video question answering for non-egocentric, first-person scenarios—such as surgery, industrial assembly, extreme sports, and animal viewpoints—that diverge significantly from everyday human experience. To this end, we introduce EgoCross, the first cross-domain benchmark specifically designed for professional and unconventional settings, featuring two tracks: Source-Limited and Open-Source, which respectively restrict or permit the use of external training data to rigorously evaluate zero-shot and few-shot generalization. Leveraging MLLMs and video-text alignment techniques, we implement standardized evaluation on the Codabench platform, enabling integration of open-source models and few-shot learning. The benchmark attracted over 130 participants with more than 1,500 submissions, and the official leaderboard along with winning solutions’ code and data has been publicly released.

Abstract

EgoCross is a cross-domain egocentric video question answering benchmark designed to evaluate whether multimodal large language models can generalize beyond common daily-life scenarios. The first EgoCross Challenge was hosted at the Third EgoVis Workshop at CVPR 2026 and evaluated models on first-person videos from four target domains: surgery, industrial assembly, extreme sports, and animal perspectives. Each test example consists of an egocentric video clip, a question, and four candidate answers, from which the model must select the correct option. This technical report introduces the challenge task, benchmark resources, and two official Codabench tracks. The Source-Limited Track restricts participants to the official baseline model and a small support set, whereas the Open-Source Track permits broader choices of models and training data under rules that prohibit the manual construction of target-domain training data. In total, the challenge received more than 1,500 submissions from over 130 participants, with 19 teams participating in the Open-Source Track and 38 teams in the Source-Limited Track. We further present the official leaderboard results and summarize the winning solutions from both tracks. We hope that this report will serve as a useful technical reference for advancing cross-domain egocentric video understanding. All resources, including the challenge data, baseline implementation, and code released by the winning teams, are made publicly available.

Problem

• egocentric video
• cross-domain
• video question answering
• multimodal large language models
• generalization

Innovation

• cross-domain
• egocentric video
• video question answering
• multimodal large language models
• generalization