SHREC’20: Shape correspondence with non-isometric deformations

Dyke, Roberto M.; Lai, Yu‐Kun; Rosin, Paul L.; Zappalá, Stefano; Dykes, Seana; Guo, Daoliang; Li, Kun; Marin, Riccardo; Melzi, Simone; Yang, Jingyu

doi:10.1016/j.cag.2020.08.008

Cited by 38 publications

(35 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Through hierarchical segmentation, this information could be translated into semantic part organization (i.e., by identifying overlapping image regions corresponding to 'bird' , Scientific Reports | (2020) 10:22141 | https://doi.org/10.1038/s41598-020-79072-w www.nature.com/scientificreports/ 'leg' and 'wing'). This information could be used to predict point-to-point correspondences on a large scale 45 ; and also to morph objects into each other in a fashion in accordance with human perception. Figuring out perceptually sensible morphs without detailed user input (i.e., manual definition of anchor points in both objects), has been an endeavor in computer graphics and image processing for a long time (e.g., 46,47 ).…”

Section: Relative Importance Of Semantics and Geometrical Featuresmentioning

confidence: 99%

The role of semantics in the perceptual organization of shape

Schmidt

Kleis

Morgenstern

et al. 2020

Sci Rep

View full text Add to dashboard Cite

Establishing correspondence between objects is fundamental for object constancy, similarity perception and identifying transformations. Previous studies measured point-to-point correspondence between objects before and after rigid and non-rigid shape transformations. However, we can also identify ‘similar parts’ on extremely different objects, such as butterflies and owls or lizards and whales. We measured point-to-point correspondence between such object pairs. In each trial, a dot was placed on the contour of one object, and participants had to place a dot on ‘the corresponding location’ of the other object. Responses show correspondence is established based on similarities between semantic parts (such as head, wings, or legs). We then measured correspondence between ambiguous objects with different labels (e.g., between ‘duck’ and ‘rabbit’ interpretations of the classic ambiguous figure). Despite identical geometries, correspondences were different across the interpretations, based on semantics (e.g., matching ‘Head’ to ‘Head’, ‘Tail’ to ‘Tail’). We present a zero-parameter model based on labeled semantic part data (obtained from a different group of participants) that well explains our data and outperforms an alternative model based on contour curvature. This demonstrates how we establish correspondence between very different objects by evaluating similarity between semantic parts, combining perceptual organization and cognitive processes.

show abstract

Section: Relative Importance Of Semantics and Geometrical Featuresmentioning

confidence: 99%

The role of semantics in the perceptual organization of shape

Schmidt

Kleis

Morgenstern

et al. 2020

Sci Rep

View full text Add to dashboard Cite

show abstract

“…To perform the evaluations, we use two datasets consisting of triangular meshes: the TOSCA dataset [BBK08] and the SHREC 2019 Shape Correspondence with Isometric and Non‐Isometric Deformations benchmark dataset [DSL*19]. The former consists of a set of collections of synthetic humanoid and animal figures in different, near‐isometric poses; the latter is made up of 3D scans of real‐world objects exhibiting a wide variety of deformations.…”

Section: Discussionmentioning

confidence: 99%

“…However, most of these methods learn local descriptors from existing handcrafted techniques [KZK17, DBI18], rather than input data [SSS19]. More generally, many state‐of‐the art pipelines for shape registration and correspondence directly incorporate ”deterministic” descriptors such as SHOT in some capacity [VLB*17, LYLG18, DSL*19] and outperform learned approaches with the proper settings [DSL*19, SSDS19].…”

Section: Related Workmentioning

confidence: 99%

“…We use the SHREC 2019 Shape Correspondence Benchmark dataset to evaluate the quality of each descriptor under rigid articulations, near‐isometric and non‐isometric deformations, and topological and geometric changes in a sparse shape correspondence regime. The dataset consists of fifty meshes constructed from 3D scans of real‐world objects; as a byproduct, the real‐world scans contain noise, varying trianglulations, occluded geometry and various other sources of interference [DSL*19]. The dataset contains 76 pre‐defined pairs of meshes partitioned into four increasingly challenging test sets: (1) articulations and rigid deformations, (2) near‐isometric deformations, (3) non‐isometric deformations and (4) topological and geometric changes.…”

Section: Sparse Correspondencesmentioning

confidence: 99%

See 1 more Smart Citation

ECHO: Extended Convolution Histogram of Orientations for Local Surface Description

Mitchel

Rusinkiewicz

Chirikjian

et al. 2020

Computer Graphics Forum

View full text Add to dashboard Cite

This paper presents a novel, highly distinctive and robust local surface feature descriptor. Our descriptor is predicated on a simple observation: instead of describing the points in the vicinity of a feature point relative to a reference frame at the feature point, all points in the region describe the feature point relative to their own frames. Isometry invariance is a byproduct of this construction. Our descriptor is derived relative to the extended convolution – a generalization of the standard convolution that allows the filter to adaptively transform as it passes over the domain. As such, we name our descriptor the Extended Convolution Histogram of Orientations (ECHO). It exhibits superior performance compared to popular surface descriptors in both feature matching and shape correspondence experiments. In particular, the ECHO descriptor is highly stable under near‐isometric deformations and remains distinctive under significant levels of noise, tessellation, complex deformations and the kinds of interference commonly found in real data.

show abstract

“…Methods for the retrieval of objects based on their surface geometric relief patterns were evaluated in [8] ; such methods are useful, e.g., in the classification of cultural heritage fragments and the organizers provided a dataset of 220 surfaces with 11 different reliefs; 7 groups participated in the evaluation with at least 1 method variation each. Non-isometric deformations for shape correspondence continues to be a challenging problem and the authors of this track [9] used a library of corresponding animal shapes with highly non-isometric deformations; ground-truth correspondences have been labelled by experts and 10 methods were submitted for evaluation.…”

mentioning

confidence: 99%