OpenEQA

OpenEQA is a benchmark for evaluating embodied question answering in 3D environments, focusing on the ability of vision-language models to gather context from multiple viewpoints for complex spatial reasoning.

Papers using OpenEQA (6)

Cov: Chain-of-view Prompting For Spatial Reasoning2026Upstreamqa: A Modular Framework For Explicit Reasoning On Video Question Answering Tasks2026Prune-Then-Plan: Step-Level Calibration for Stable Frontier Exploration in Embodied Question Answering2025EQA-RM: A Generative Embodied Reward Model With Test-time Scaling2025Vl-kng: Persistent Spatiotemporal Knowledge Graphs From Egocentric Video For Embodied Scene Understanding2025Bridgeeqa: Virtual Embodied Agents For Real Bridge Inspections2025