3D referring expression instance segmentation aims to identify and segment target objects in complex scenes according to natural language descriptions. This task is particularly challenging due to spatial ambiguity and unstable query-instance associations caused by weak linguistic supervision. Previous methods relying on implicit spatial modeling often struggle with confused localization and ambiguous instance aggregation. To address these limitations, we propose SpaDis, a new end-to-end framework that jointly achieves Spatial Semantic Enhancement (SSE) and Discriminative Query Refinement (DQR). Specifically, the SSE module explicitly encodes spatial relationships via geometric cues to strengthen query-level semantic representations, while the DQR module performs a discriminative refinement on queries to establish robust associations, enforcing intra-instance consistency and inter-instance separation before decoding. Extensive experiments on multiple benchmarks demonstrate that SpaDis achieves consistent performance gains, validating the effectiveness of explicit spatial modeling and discriminative feature optimization in complex 3D scenes.

Vision-based 3D sensing provides an effective approach for constructing human-centric digital twins in applications such as ergonomic assessment and human–robot collaboration. However, reconstructing reliable dynamic human digital twins from visual sensor observations remains challenging due to large articulated deformation, which often causes surface discontinuities in existing 3D Gaussian Splatting (3DGS)-based human avatars. Although current methods achieve high image-level rendering quality, their Gaussian primitives are not explicitly adapted to pose-induced surface deformation, resulting in cracks and holes during motion. In this work, we propose ElasticGS, a geometry-consistent human digital twin reconstruction framework for vision-based 3D sensing. ElasticGS introduces a training-free Pose-Aware Dynamic Covariance Adaptation strategy, which dynamically adjusts the rotation and anisotropic scale of Gaussian primitives according to local human surface deformation. Specifically, localized Principal Component Analysis (PCA) is performed on deformed SMPL-X mesh neighborhoods to estimate surface stretching directions and magnitudes, enabling Gaussian representations to maintain continuous surface coverage under complex poses. Furthermore, we propose the Avatar Surface Integrity Score (ASIS), a reference-free metric that evaluates structural completeness and surface compactness of reconstructed human avatars without requiring pixel-aligned ground truth. Experiments on the X-Humans, AvatarReX, and People-Snapshot datasets demonstrate that ElasticGS significantly improves surface integrity while maintaining competitive image-level sensing consistency. Compared with ExAvatar, ElasticGS maintains satisfactory rendering quality while improving ASIS from 0.741 to 0.961 on X-Humans, from 0.871 to 0.949 on AvatarReX, and from 0.874 to 0.952 on People-Snapshot. These results demonstrate the effectiveness of pose-aware covariance adaptation for robust human digital twin reconstruction in vision-based 3D sensing systems.

Vision-based 3D sensing provides an effective approach for constructing human-centric digital twins in applications such as ergonomic assessment and human–robot collaboration. However, reconstructing reliable dynamic human digital twins from visual sensor observations remains challenging due to large articulated deformation, which often causes surface discontinuities in existing 3D Gaussian Splatting (3DGS)-based human avatars. Although current methods achieve high image-level rendering quality, their Gaussian primitives are not explicitly adapted to pose-induced surface deformation, resulting in cracks and holes during motion. In this work, we propose ElasticGS, a geometry-consistent human digital twin reconstruction framework for vision-based 3D sensing. ElasticGS introduces a training-free Pose-Aware Dynamic Covariance Adaptation strategy, which dynamically adjusts the rotation and anisotropic scale of Gaussian primitives according to local human surface deformation. Specifically, localized Principal Component Analysis (PCA) is performed on deformed SMPL-X mesh neighborhoods to estimate surface stretching directions and magnitudes, enabling Gaussian representations to maintain continuous surface coverage under complex poses. Furthermore, we propose the Avatar Surface Integrity Score (ASIS), a reference-free metric that evaluates structural completeness and surface compactness of reconstructed human avatars without requiring pixel-aligned ground truth. Experiments on the X-Humans and AvatarReX datasets demonstrate that ElasticGS significantly improves surface integrity while maintaining competitive image-level sensing consistency. Compared with ExAvatar, ElasticGS improves ASIS from 0.642 to 0.846 on X-Humans and from 0.603 to 0.688 on AvatarReX. These results demonstrate the effectiveness of pose-aware covariance adaptation for robust human digital twin reconstruction in vision-based 3D sensing systems.