HybridSDF: Combining Deep Implicit Shapes and Geometric Primitives for 3D Shape Representation and Manipulation

Vasu, Subeesh; Talabot, Nicolas; Lukoianov, Artem; Baqué, Pierre; Donier, Jonathan; Fua, Pascal

doi:10.1109/3dv57658.2022.00072

Cited by 8 publications

(2 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, it becomes popular to encode shapes as a continuous SDF with neural networks [28]. Vasu et al [43] further improve the shape encoding quality by enforcing local regularities with geometric primitives. In [8], the authors adopt a twostage meta-learning approach to further extend the generalization capabilities of neural SDF.…”

Section: Related Workmentioning

confidence: 99%

“…It is the basic framework for many classic 3D reconstruction algorithms such as TSDF volume reconstruction [10,15], KinectFusion [19], and Dynamic-Fusion [26]. Recently, the SDF representation is adapted to the deep learning frameworks, and exhibits boosted potentials in shape encoding [8,18,28,43], surface reconstruction [23,46], and shape completion [11,12,34]. Usually, triangular mesh surfaces are extracted from the SDF representation with the marching cubes algorithm [25].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Marching-Primitives: Shape Abstraction from Signed Distance Function

Liu¹,

Wu²,

Ruan³

et al. 2023

Preprint

View full text Add to dashboard Cite

Representing complex objects with basic geometric primitives has long been a topic in computer vision. Primitive-based representations have the merits of compactness and computational efficiency in higher-level tasks such as physics simulation, collision checking, and robotic manipulation. Unlike previous works which extract polygonal meshes from a signed distance function (SDF), in this paper, we present a novel method, named Marching-Primitives, to obtain a primitive-based abstraction directly from an SDF. Our method grows geometric primitives (such as superquadrics) iteratively by analyzing the connectivity of voxels while marching at different levels of signed distance. For each valid connected volume of interest, we march on the scope of voxels from which a primitive is able to be extracted in a probabilistic sense and simultaneously solve for the parameters of the primitive to capture the underlying local geometry. We evaluate the performance of our method on both synthetic and real-world datasets. The results show that the proposed method outperforms the state-of-the-art in terms of accuracy, and is directly generalizable among different categories and scales.The code is open-sourced at https://github.com/ ChirikjianLab/Marching-Primitives.git.

show abstract

Section: Related Workmentioning

confidence: 99%