Performance Evaluation of 3D Keypoint Detectors

By Samuele Salti, Federico Tombari, Luigi Di Stefano. 2012

472 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Geology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

In Geology

Citations in the corpus

Histogram of Oriented Principal Components for Cross-View Action Recognition (2014, methodology): “y not capture sufﬁcient information about an action if it is performed slowly. 7.1 Spatial Scale Selection Several automatic spatial scale selection methods have been proposed for 3D object retrieval [57]. We adapt the method proposed by Mian et al. [58] in object retrieval for action recognition in”
HOPC: Histogram of Oriented Principal Components of 3D Pointclouds for Action Recognition (2014, background): “Several automatic spatial scale detection methods [28] have been proposed for 3D object recognition.”
D3Feat: Joint Learning of Dense Detection and Description of 3D Local Features (2020, background): “A comprehensive review of such methods can be found in [34].”
TEASER: Fast and Certifiable Point Cloud Registration (2020, methodology): “Large outlier rates are typical of 3-D keypoint detection and matching [16].”
Recognizing Objects in 3D Point Clouds with Multi-Scale Local Features (2014, methodology): “For example, the scale repeatability of MeshHoG, KPQ-AS and SP algorithms is, respectively, 41%, 51% and 43%, when tested on the University of Western Australia (UWA) laser scanner dataset [26].”
SHOT: Unique signatures of histograms for surface and texture description (2014, unknown): “A performance evaluation of 3D keypoint detection algorithms has been recently proposed in [12].”
Fast Global Registration (2016, background): “Some pipelines use point-to-point matches based on local geometric descriptors [16, 40], others define correspondences on pairs or tuples of points [1, 8, 26, 29].”
3D-Div: A novel local surface descriptor for feature matching and pairwise range image registration (2013, background): “In addition, they are not computationally efficient, making them not geared towards real-time applications [10].”
A Comprehensive Performance Evaluation of 3D Local Feature Descriptors (2015, methodology): “In the synthetic datasets (e.g., Random views), the ground-truth transformation is obtained during the process of simulation (Tombari et al. 2013).”
Learning to Detect Good 3D Keypoints (2017, methodology): “As discussed in Tombari et al. (2013), the main benefit brought in by an adaptive-scale 3D detector concerns the ability to gather more features than a fixed-scale approach, as more scales are probed by the former and some surface patches might show up as salient at certain scales rather than…”
Deep Learning Advances in Computer Vision with 3D Data (2017, methodology): “…are Local Surface Patches (LSPs) [Chen and Bhanu 2007], Intrinsic Shape Signatures (ISSs) [Zhong 2009], and the 3D detector proposed in Mian et al. [2010] termed as “KeyPoint Quality” (KPQ) in Tombari et al. [2013], while a popular adaptive-scale detector is MeshDoG [Zaharescu et al. 2009].”
Intelligent multi-camera video surveillance: A review (2013, background): “Comparisons of various keypoint detectors and local descriptors can be found in (Salti et al., 2011; Mikolajczyk and Schmid, 2005).”
Image Matching from Handcrafted to Deep Features: A Survey (2020, background): “Readers are referred to Tombari et al. (2013) for further discussion on other adaptive-scale detectors.”
3D Object Recognition in Cluttered Scenes with Local Surface Features: A Survey (2014, methodology): “number of keypoints in the two images [33], [39], [40], [41], [42].”
Rotational Projection Statistics for 3D Local Surface Description and Object Recognition (2013, background): “As reported in (Tombari et al., 2013), the scale repeatability of the adaptive-scale detector in (Mian et al., 2010) were less than 85% and 60% on the Retrieval dataset and Random Views dataset, respectively.”