Pseudo-EV: Enhancing 3D Visual Grounding With Pseudo Embodied Viewpoint

3D Visual Grounding based on natural language is a fundamental task in Embodied AI. One of the fundamental challenges in localizing objects in 3D scenes through natural language descriptions arises from the variable perception of spatial relationships among objects when viewed from different perspectives. To address this issue, we introduce a model named Pseudo-EV, which decomposes the problem of 3D visual grounding into two stages: <xref ref-type="disp-formula" rid="deqn1" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">(1)</xref> predicting an embodied viewpoint and <xref ref-type="disp-formula" rid="deqn2" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">(2)</xref> determining the target object within that viewpoint, thereby eliminating viewpoint ambiguity. Given the scarcity of annotations for embodied viewpoint prediction, we employ a large language model (LLM) to generate pseudo-labels for existing datasets as intermediate training targets. However, directly predicting viewpoints in continuous Euclidean space proves inefficient, leading to weaker alignment with textual queries and scene semantics, as well as higher training overhead. To overcome these limitations, we introduce two streamlined strategies: an Embodied Viewpoint with Semantic Structure and a Decoupled Target Prediction Strategy. Extensive experiments demonstrate that predicting intermediate embodied viewpoints substantially boosts the performance of 3D visual grounding, achieving state-of-the-art results on both ScanRefer and Nr3D/Sr3D. Moreover, our framework significantly reduces computational cost compared to other viewpoint-aware approaches.