Segmentation-Driven Initialization for Sparse-view 3D Gaussian Splatting

Dental3R is a new method for intraoral 3D reconstruction that addresses the limitations of traditional intraoral scanning, which is often inaccessible for remote tele-orthodontics. Conventional techniques struggle with sparse smartphone imagery due to large view baselines, inconsistent lighting, and reflective surfaces, leading to challenges in pose and geometry estimation. Dental3R utilizes a pose-free, graph-guided approach to achieve robust, high-fidelity reconstructions from sparse intraoral photographs, overcoming issues like frequency bias that can result in loss of diagnostic details.

RoboTidy is a new benchmark designed for language-guided household tidying, addressing limitations in current benchmarks that fail to model user preferences and mobility. It features 500 photorealistic 3D Gaussian Splatting household scenes, an 'Action (Object, Container)' list for tidying tasks, and provides extensive manipulation and navigation trajectories for training. This initiative aims to enhance the assessment of integrated language-to-action capabilities in robotic systems.

The paper presents a novel approach to modeling hand-object interactions in dynamic scenes without relying on object priors. It introduces interaction-aware hand-object Gaussians with optimizable parameters to improve structural representation. The method incorporates hand shape into the object deformation field to model flexible motions and employs a progressive optimization strategy to address challenges in dynamic regions and static backgrounds.

Image-Based Gaussian Splatting (IBGS) is a novel approach to 3D Gaussian Splatting (3DGS) that enhances novel view synthesis (NVS) by utilizing high-resolution source images. This method addresses limitations in capturing spatially varying colors and view-dependent effects, such as specular highlights, which are often hindered by low-degree spherical harmonics. By modeling pixel colors as a combination of base colors and learned residuals from neighboring images, IBGS promotes accurate surface alignment and enables the rendering of high-frequency details.