Split-and-Fit: Learning B-Reps via Structure-Aware Voronoi Partitioning
2406.05261
0
0
Abstract
We introduce a novel method for acquiring boundary representations (B-Reps) of 3D CAD models which involves a two-step process: it first applies a spatial partitioning, referred to as the split, followed by a fit operation to derive a single primitive within each partition. Specifically, our partitioning aims to produce the classical Voronoi diagram of the set of ground-truth (GT) B-Rep primitives. In contrast to prior B-Rep constructions which were bottom-up, either via direct primitive fitting or point clustering, our Split-and-Fit approach is top-down and structure-aware, since a Voronoi partition explicitly reveals both the number of and the connections between the primitives. We design a neural network to predict the Voronoi diagram from an input point cloud or distance field via a binary classification. We show that our network, coined NVD-Net for neural Voronoi diagrams, can effectively learn Voronoi partitions for CAD models from training data and exhibits superior generalization capabilities. Extensive experiments and evaluation demonstrate that the resulting B-Reps, consisting of parametric surfaces, curves, and vertices, are more plausible than those obtained by existing alternatives, with significant improvements in reconstruction quality. Code will be released on https://github.com/yilinliu77/NVDNet.
Create account to get full access
Overview
- This paper introduces a new method called "Split-and-fit" for acquiring boundary representations (B-Reps) of 3D CAD shapes from point cloud data.
- The key idea is to first partition the 3D space using Voronoi diagrams, and then fit a single primitive shape (e.g., a plane, sphere, cylinder) to each partition.
- This approach offers a different perspective on the problem of 3D shape reconstruction compared to traditional methods.
Plain English Explanation
The paper presents a novel technique called "Split-and-fit" for reconstructing 3D shapes from point cloud data, which is often the kind of data captured by 3D sensors. The core insight is to first divide up the 3D space into smaller regions using a mathematical tool called a Voronoi diagram, and then find a simple geometric shape (like a plane, sphere, or cylinder) that best fits each of those regions.
This is different from other 3D reconstruction methods, which might try to fit a single, complex 3D model to the entire point cloud. The "Split-and-fit" approach breaks the problem down into simpler subproblems, which can lead to more accurate and efficient reconstructions, especially for complex 3D shapes. By representing the shape as a collection of simple primitives, it also provides a more compact and interpretable boundary representation (B-Rep) of the 3D object.
Technical Explanation
The "Split-and-fit" method first uses Voronoi diagrams to partition the 3D space occupied by the input point cloud. Voronoi diagrams divide space based on proximity to a set of seed points, creating a tessellation of convex polyhedra.
The key insight is that each of these Voronoi cells can then be approximated by a single primitive shape, such as a plane, sphere, cylinder, or cone. The paper explores several techniques for robustly fitting these primitives to the points within each Voronoi cell.
By combining the set of fitted primitives, the method constructs a boundary representation (B-Rep) of the original 3D shape. This B-Rep can be more compact and semantically meaningful than other reconstruction techniques, which may fit a single, complex surface model to the entire point cloud.
The paper demonstrates the "Split-and-fit" approach on a variety of 3D shapes, including those from the PartNet and ShapeNet datasets. The results show that it can produce accurate B-Rep reconstructions, even for challenging shapes with complex geometries.
Critical Analysis
The "Split-and-fit" method offers a fresh perspective on 3D shape reconstruction, but it also has some limitations:
- The reliance on Voronoi diagrams means the method may struggle with very irregular or non-convex shapes, as the Voronoi cells may not align well with the underlying geometry.
- The primitive fitting step relies on classic optimization techniques, which can be sensitive to initialization and may not always converge to the globally optimal solution.
- The method currently requires the number of primitives to be specified in advance, which may not be known for arbitrary 3D shapes.
Further research could explore ways to address these limitations, such as using more advanced spatial partitioning techniques or learned primitive fitting models, as explored in papers like BRep-Gen and Part-Aware Shape Generation. Additionally, the method could be extended to handle noisy or incomplete input data, which is common in real-world 3D sensing applications.
Conclusion
The "Split-and-fit" method presented in this paper offers a new way to approach the problem of 3D shape reconstruction from point cloud data. By first partitioning the 3D space and then fitting simple primitive shapes to each partition, it can produce compact and semantically meaningful boundary representations of complex 3D objects.
While the method has some limitations, it represents an interesting step forward in the field of 3D shape modeling and could inspire further research into efficient and interpretable techniques for 3D reconstruction. As 3D sensing technologies continue to advance, methods like "Split-and-fit" may become increasingly valuable for a wide range of applications, from computer-aided design to autonomous robotics.
This summary was produced with help from an AI and may contain inaccuracies - check out the links to read the original source documents!
Related Papers
📈
BrepGen: A B-rep Generative Diffusion Model with Structured Latent Geometry
Xiang Xu, Joseph G. Lambourne, Pradeep Kumar Jayaraman, Zhengqing Wang, Karl D. D. Willis, Yasutaka Furukawa
0
0
This paper presents BrepGen, a diffusion-based generative approach that directly outputs a Boundary representation (B-rep) Computer-Aided Design (CAD) model. BrepGen represents a B-rep model as a novel structured latent geometry in a hierarchical tree. With the root node representing a whole CAD solid, each element of a B-rep model (i.e., a face, an edge, or a vertex) progressively turns into a child-node from top to bottom. B-rep geometry information goes into the nodes as the global bounding box of each primitive along with a latent code describing the local geometric shape. The B-rep topology information is implicitly represented by node duplication. When two faces share an edge, the edge curve will appear twice in the tree, and a T-junction vertex with three incident edges appears six times in the tree with identical node features. Starting from the root and progressing to the leaf, BrepGen employs Transformer-based diffusion models to sequentially denoise node features while duplicated nodes are detected and merged, recovering the B-Rep topology information. Extensive experiments show that BrepGen advances the task of CAD B-rep generation, surpassing existing methods on various benchmarks. Results on our newly collected furniture dataset further showcase its exceptional capability in generating complicated geometry. While previous methods were limited to generating simple prismatic shapes, BrepGen incorporates free-form and doubly-curved surfaces for the first time. Additional applications of BrepGen include CAD autocomplete and design interpolation. The code, pretrained models, and dataset are available at https://github.com/samxuxiang/BrepGen.
5/17/2024
Part123: Part-aware 3D Reconstruction from a Single-view Image
Anran Liu, Cheng Lin, Yuan Liu, Xiaoxiao Long, Zhiyang Dou, Hao-Xiang Guo, Ping Luo, Wenping Wang
0
0
Recently, the emergence of diffusion models has opened up new opportunities for single-view reconstruction. However, all the existing methods represent the target object as a closed mesh devoid of any structural information, thus neglecting the part-based structure, which is crucial for many downstream applications, of the reconstructed shape. Moreover, the generated meshes usually suffer from large noises, unsmooth surfaces, and blurry textures, making it challenging to obtain satisfactory part segments using 3D segmentation techniques. In this paper, we present Part123, a novel framework for part-aware 3D reconstruction from a single-view image. We first use diffusion models to generate multiview-consistent images from a given image, and then leverage Segment Anything Model (SAM), which demonstrates powerful generalization ability on arbitrary objects, to generate multiview segmentation masks. To effectively incorporate 2D part-based information into 3D reconstruction and handle inconsistency, we introduce contrastive learning into a neural rendering framework to learn a part-aware feature space based on the multiview segmentation masks. A clustering-based algorithm is also developed to automatically derive 3D part segmentation results from the reconstructed models. Experiments show that our method can generate 3D models with high-quality segmented parts on various objects. Compared to existing unstructured reconstruction methods, the part-aware 3D models from our method benefit some important applications, including feature-preserving reconstruction, primitive fitting, and 3D shape editing.
5/28/2024
🛸
Part-aware Shape Generation with Latent 3D Diffusion of Neural Voxel Fields
Yuhang Huang, SHilong Zou, Xinwang Liu, Kai Xu
0
0
This paper presents a novel latent 3D diffusion model for the generation of neural voxel fields, aiming to achieve accurate part-aware structures. Compared to existing methods, there are two key designs to ensure high-quality and accurate part-aware generation. On one hand, we introduce a latent 3D diffusion process for neural voxel fields, enabling generation at significantly higher resolutions that can accurately capture rich textural and geometric details. On the other hand, a part-aware shape decoder is introduced to integrate the part codes into the neural voxel fields, guiding the accurate part decomposition and producing high-quality rendering results. Through extensive experimentation and comparisons with state-of-the-art methods, we evaluate our approach across four different classes of data. The results demonstrate the superior generative capabilities of our proposed method in part-aware shape generation, outperforming existing state-of-the-art methods.
6/24/2024
Neural Bounding
Stephanie Wenxin Liu, Michael Fischer, Paul D. Yoo, Tobias Ritschel
0
0
Bounding volumes are an established concept in computer graphics and vision tasks but have seen little change since their early inception. In this work, we study the use of neural networks as bounding volumes. Our key observation is that bounding, which so far has primarily been considered a problem of computational geometry, can be redefined as a problem of learning to classify space into free or occupied. This learning-based approach is particularly advantageous in high-dimensional spaces, such as animated scenes with complex queries, where neural networks are known to excel. However, unlocking neural bounding requires a twist: allowing -- but also limiting -- false positives, while ensuring that the number of false negatives is strictly zero. We enable such tight and conservative results using a dynamically-weighted asymmetric loss function. Our results show that our neural bounding produces up to an order of magnitude fewer false positives than traditional methods. In addition, we propose an extension of our bounding method using early exits that accelerates query speeds by 25%. We also demonstrate that our approach is applicable to non-deep learning models that train within seconds. Our project page is at: https://wenxin-liu.github.io/neural_bounding/.
5/27/2024