MM-Conv: A Multimodal Dataset and Benchmark for Context-Aware Grounding in 3D Dialogue
This work provides a new benchmark and method for context-aware grounding in 3D dialogue, addressing a key limitation of current VLMs in dynamic, multi-turn settings.
The authors introduce a benchmark for referential communication in dynamic 3D environments, built from 6.7 hours of egocentric VR interaction, and a two-stage grounding pipeline that improves grounding performance by 11-22 percentage points, with a pure detector reaching 56.7% on pronominals after rewriting.
Grounding language in the physical world requires AI systems to interpret references that emerge dynamically during conversation. While current vision-language models (VLMs) excel at static image tasks, they struggle to resolve ambiguous expressions in spontaneous, multi-turn dialogue. We address this gap by introducing (1) a benchmark for referential communication in dynamic 3D environments, built from 6.7 hours of egocentric VR interaction with synchronized speech, motion, gaze, and 3D scene geometry, and (2) a two-stage grounding pipeline that explicitly resolves conversational ambiguity before visual localization. The benchmark includes over 4,200 manually verified referring expressions spanning full, partitive, and pronominal types. Our contextual rewriting approach improves grounding performance by 11-22 percentage points on average, with a pure detector (GroundingDINO) reaching 56.7% on pronominals after rewriting, nearly double the best end-to-end baseline. Results demonstrate that decoupling linguistic reasoning from visual perception is more effective than end-to-end approaches for conversational grounding.