A 2D shape structure for decomposition and part similarity

Leonard, Kathryn; Morin, Géraldine; Hahmann, Stéfanie; Carlier, Aurélie

doi:10.1109/icpr.2016.7900130

Cited by 18 publications

(15 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We then order the reduced set of branches according to branch importance. For both steps, we use a salience measure, the Weighted Extended Distance Function (WEDF) function on the skeleton [20], to determine the relative importance of branches. The WEDF function has been shown to measure relative importance of shape parts in a way that matches human perception [8].…”

Section: Parametric Skeletonsmentioning

confidence: 99%

SkelNetOn 2019: Dataset and Challenge on Deep Learning for Geometric Shape Understanding

Demir¹,

Hahn

Leonard

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

Self Cite

View full text Add to dashboard Cite

Figure 1: SkelNetOn Challenges: Example shapes and corresponding skeletons are demonstrated for the three challenge tracks in pixel (left), point (middle), and parametric domain (right). AbstractWe present SkelNetOn 2019 Challenge and Deep Learning for Geometric Shape Understanding workshop to utilize existing and develop novel deep learning architectures for shape understanding. We observed that unlike traditional segmentation and detection tasks, geometry understanding is still a new area for deep learning techniques. SkelNetOn aims to bring together researchers from different domains to foster learning methods on global shape understanding tasks. We aim to improve and evaluate the state-of-theart shape understanding approaches, and to serve as reference benchmarks for future research. Similar to other challenges in computer vision [6,22], SkelNetOn proposes three datasets and corresponding evaluation methodologies; all coherently bundled in three competitions with a dedicated workshop co-located with CVPR 2019 conference. In this paper, we describe and analyze characteristics of datasets, define the evaluation criteria of the public competitions, and provide baselines for each task.Computer vision approaches have shown tremendous progress toward understanding shapes from various data formats, especially since entering the deep learning era. Although detection, recognition, and segmentation approaches achieve highly accurate results, there has been rel-arXiv:1903.09233v3 [cs.CV]

show abstract

Section: Parametric Skeletonsmentioning

confidence: 99%

SkelNetOn 2019: Dataset and Challenge on Deep Learning for Geometric Shape Understanding

Demir¹,

Hahn

Leonard

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

Self Cite

View full text Add to dashboard Cite

show abstract

“…While most of the 3D methods described above do not make use of skeletons, in part because of their perceived instability and difficult to compute, recent methods for 2D shape analysis, decomposition and clustering [5,6,31] successfully exploit the geometry of the Blum medial axis for various shape tasks. These methods have shown that the use of the skeletons can be complemented with properly chosen functions that provide robustness and stability.…”

Section: Preliminariesmentioning

confidence: 99%

“…Below, we introduce such a measure, called Weighted Extended Distance Function (WEDF), which is defined on a curve skeleton centered within a 3D shape. WEDF has been introduced recently for 2D shape contours [5] and successfully used for defining a salience measure and decomposition of 2D curves [6]. Generalizing to the 3D setting takes two steps: first defining WEDF on a 3D skeleton and deducing a partition, second, mapping the measure on the surface mesh.…”

mentioning

confidence: 99%

“…The second step assesses salience of surface mesh parts: it establishes an injective mapping between surface mesh vertices and the skeleton points while ensuring simultaneously that the inverse mapping produces coherent shape parts. In 2D, a straightforward association of boundary points to the WEDF values of the corresponding skeletal points generates a good partition of a shape [6], but the 3D setting is more challenging: a direct extension of the 2D approach to 3D generates a poor partition (see Fig. 10, where the legs run into the body of the ant).…”

mentioning

confidence: 99%

See 1 more Smart Citation

A salience measure for 3D shape decomposition and sub-parts classification

et al. 2018

Self Cite

View full text Add to dashboard Cite

Open Archive Toulouse Archive OuverteOATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible a A B S T R A C T This paper introduces a measure of significance on a curve skeleton of a 3D piecewise linear shape mesh, allowing the computation of both the shape's parts and their saliency. We begin by reformulating three existing pruning measures into a non-linear PCA along the skeleton. From this PCA, we then derive a volume-based salience measure, the 3D WEDF, that determines the relative importance to the global shape of the shape part associated to a point of the skeleton. First, we provide robust algorithms for computing the 3D WEDF on a curve skeleton, independent on the number of skeleton branches. Then, we cluster the WEDF values to partition the curve skeleton, and coherently map the decomposition to the associated surface mesh. Thus, we develop an unsupervised hierarchical decomposition of the mesh faces into visually meaningful shape regions that are or-dered according to their degree of perceptual salience. The shape analysis tools introduced in this paper are important for many applications including shape comparison, editing, and compression.

show abstract

“…In this challenge, a pre-segmented image dataset with the corresponding skeleton representations in three tracks is provided [25]. The first track has posed the challenge of extracting the skeleton pixels from the given pre-segmented images [25] [16] [19] [24]. We have approached this challenge as an edge detection problem and introduced a version of HED architecture in the decoder part of our proposed architecture.…”

Section: Introductionmentioning

confidence: 99%

SkeletonNet: Shape Pixel to Skeleton Pixel

Nathan¹,

Kansal²

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

View full text Add to dashboard Cite

Deep Learning for Geometric Shape Understating has organized a challenge for extracting different kinds of skeletons from the images of different objects. This competition is organized in association with CVPR 2019. There are three different tracks of this competition. The present manuscript describes the method used to train the model for the dataset provided in the first track. The first track aims to extract skeleton pixels from the shape pixels of 89 different objects. For the purpose of extracting the skeleton, a U-net model which is comprised of an encoderdecoder structure has been used. In our proposed architecture, unlike the plain decoder in the traditional U net, we have designed the decoder in the format of HED architecture, wherein we have introduced 4 side layers and fused them to one dilation convolutional layer to connect the broken links of the skeleton. Our proposed architecture achieved the F1 score of 0.77 on test data.

show abstract

A 2D shape structure for decomposition and part similarity

Cited by 18 publications

References 32 publications

SkelNetOn 2019: Dataset and Challenge on Deep Learning for Geometric Shape Understanding

SkelNetOn 2019: Dataset and Challenge on Deep Learning for Geometric Shape Understanding

A salience measure for 3D shape decomposition and sub-parts classification

SkeletonNet: Shape Pixel to Skeleton Pixel

Contact Info

Product

Resources

About