ScanQA

ScanQA is a dataset that contains question-answer pairs related to 3D scenes, used to evaluate the performance of models in embodied question answering tasks.

Papers using ScanQA (8)

SpatialPrompting: Keyframe-driven Zero-Shot Spatial Reasoning with Off-the-Shelf Multimodal Large Language Models2025 · 1 cites3D Question Answering via only 2D Vision-Language Models2025 · 1 citesNot All Modalities Are Equal: Instruction-Aware Gating for Multimodal Videos2026Chat-Scene++: Exploiting Context-Rich Object Identification for 3D LLM2026Chain of Questions: Guiding Multimodal Curiosity in Language Models20253D-MoRe: Unified Modal-Contextual Reasoning for Embodied Question Answering2025Bridging the Gap between 2D and 3D Visual Question Answering: A Fusion Approach for 3D VQA2024 · 19 citesEvaluating Zero-Shot GPT-4V Performance on 3D Visual Question Answering Benchmarks2024 · 1 cites