Primitive-Based Shape Abstraction via Nonparametric Bayesian Inference

Wu, Yuwei; Liu, Weixiao; Ruan, Sipu; Chirikjian, Gregory S.

doi:10.1007/978-3-031-19812-0_28

Cited by 10 publications

(12 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More recently, Liu et al [24] formulate the problem in a probabilistic fashion and propose a geometric strategy to avoid local optimum, bringing a significant improvement in robustness to outlier and fitting accuracy. Wu et al [47] extend and recast the work as a nonparametric Bayesian inference problem so as to improve the applicability on complex shapes. To the best of the authors' knowledge, the existing computational methods are all based on range images or point clouds, which suffer from geometric ambiguities [48].…”

Section: Related Workmentioning

confidence: 99%

“…Baselines: We compare our method with both the stateof-the-art learning-based method [31] and the computational method [47], which infer superquadric abstractions of the input objects. For convenience, we refer to [31] as SQs, and [47] as NB. We use the official codes and follow the implementation details as stated in these papers, respectively.…”

Section: Evaluation On Datasetsmentioning

confidence: 99%

“…Therefore, researchers turn to exploring the possibility of interpreting complex objects and scenes with basic geometric primitives. Taking advantage of the primitive-based representation, many higher-level tasks, such as segmentation [14,16,21,30], scene understanding [29,31,41,47], grasping [33,44,45] and motion planning [35,36], are able to be solved efficiently.…”

Section: Introductionmentioning

confidence: 99%

“…Later in [7,22], their method is extended to represent more complex objects with multiple primitives. More recently, the authors of [24,47] reformulate the task as a probabilistic inference problem with enhanced accuracy and robustness to noise and outliers. At the same time, with the surge of data-driven techniques, researchers attempt to train neural networks to infer cuboids [27,38,41,48,50] and superquadrics [29,31] representations in an end-to-end fashion.…”

Section: Introductionmentioning

confidence: 99%

“…For exam-ple, the algorithms tend to fill empty spaces of a non-convex object with primitives by mistake, due to the inside/outside ambiguity of a surface depicted by a set of points [24,48]. The main drawback of the learning approaches lies in the lack of generalizability beyond the object category on which the model is trained [24,31,47,48]. Also, the shape abstraction accuracy is inferior to the computational methods.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

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%