ScanRefer

ScanRefer is a dataset that contains text annotations for 3D scenes, used to evaluate the performance of models in generating object descriptions and answering questions in complex 3D environments.

Papers using ScanRefer (8)

PruneGround: Plug-and-play Spatial Pruning for 3D Visual Grounding2026Think While You Map: Asynchronous Vision-Language Agents for Incremental 3D Scene Graphs20263D CoCa v2: Contrastive Learners with Test-Time Search for Generalizable Spatial Intelligence2026Hierarchical Collaborative Fusion for 3D Instance-aware Referring Expression Segmentation2026Chat-Scene++: Exploiting Context-Rich Object Identification for 3D LLM2026Direction-aware 3D Large Multimodal Models20263D-MoRe: Unified Modal-Contextual Reasoning for Embodied Question Answering2025Zero-shot 3D Visual Grounding From Vision-language Models2025