Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces StructureGS, a novel framework for reconstructing articulated objects by integrating structure-aware guidance into 3D Gaussian Splatting. By leveraging oriented bounding boxes to enforce spatial coherence and structural connectivity, the method effectively disentangles geometry, appearance, and motion parameters, overcoming the limitations of existing photometric supervision techniques. Extensive experiments show that StructureGS achieves state-of-the-art performance, yielding high-quality reconstructions with clearly defined part geometries and improved boundary clarity.
StructureGS achieves unprecedented clarity in articulated object reconstruction by enforcing spatial coherence and structural connectivity, setting a new standard in the field.
Reconstructing articulated objects with multiple movable parts is essential for understanding object structure and enabling physical interaction. However, this reconstruction task poses significant challenges due to the entanglement of geometry, appearance, and motion parameters during optimization. Existing methods rely primarily on photometric supervision, which commonly fails to disentangle these interdependent components, resulting in poor part decomposition with blurred boundaries and geometric artifacts. To address this limitation, we introduce StructureGS, a reconstruction framework for articulated objects that integrates structure-aware guidance into 3D Gaussian Splatting. Our approach leverages oriented bounding boxes of object parts to enforce two key structural properties: spatial coherence, which constrains each part's geometry to remain compact and spatially coherent within its designated region, and structural connectivity, which enforces physically plausible contact relationships between adjacent parts. These properties are realized through structure-aware losses that inject explicit structural constraints into the optimization process. Extensive experiments demonstrate that our method achieves state-of-the-art performance in articulated object reconstruction, producing high-quality results with well-defined part geometries.