Segmentation of 3D lidar data in non-flat urban environments using a local convexity criterion

By Frank Moosmann, Oliver Pink, Christoph Stiller. 2009

332 citations recorded by the source corpus.

Subfields: Automotive Engineering, Environmental Engineering, Instrumentation

How researchers see this paper

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

In Automotive Engineering

In Environmental Engineering

In Instrumentation

Citations in the corpus

3D fully convolutional network for vehicle detection in point cloud (2016, background): “Candidates can be proposed by delicate segmentation algorithms [2], [5], [11], [14], [21], [22], [28], [29], [31], sliding window [30], random sampling [13], or the recently popular Region Proposal Network (RPN) [26].”
Vehicle Detection from 3D Lidar Using Fully Convolutional Network (2016, background): “In some early works, rule-based segmentation is suggested for speciﬁc scene [10, 20, 5].”
SqueezeSegV2: Improved Model Structure and Unsupervised Domain Adaptation for Road-Object Segmentation from a LiDAR Point Cloud (2018, methodology): “Non-deep-learning methods [1], [7], [8] usually involve several stages such as ground removal, clustering, and classification.”
Object discovery in 3D scenes via shape analysis (2013, background): “A natural choice is to consider the dot product between two normals ni, n j, but inspired by some prior work on segmentation in 3D scenes [26], we found that significantly more pleasing results can be obtained using a local curvature-aware metric.”
On the segmentation of 3D LIDAR point clouds (2011, background): “The contribution of this work is multi-fold: a set of segmentation algorithms are proposed for various data densities, which neither assume the ground to be flat nor require a prior knowledge of the location of the objects to be segmented.”
Fast segmentation of 3D point clouds: A paradigm on LiDAR data for autonomous vehicle applications (2017, background): “[4] are creating an undirected graph and compare local changes of plane normals in order to characterize changes in the slope.”
MaskFusion: Real-Time Recognition, Tracking and Reconstruction of Multiple Moving Objects (2018, background): “[29] successfully segment 3D laser data based on this assumption.”
Laser-based segment classification using a mixture of bag-of-words (2013, background): “In contrast to these approaches, other solutions [13], [14] explicitly exploit the sensing method to attain real-time capable solutions with graph-based methods.”
SqueezeSeg: Convolutional Neural Nets with Recurrent CRF for Real-Time Road-Object Segmentation from 3D LiDAR Point Cloud (2017, background): “This paradigm, despite its popularity [2], [3], [4], [5], has several disadvantages: a) Ground segmentation in the above pipeline usually relies on hand-crafted features or decision rules – some approaches rely on a scalar threshold [6] and others require more complicated features such as surface no”
Perception, Planning, Control, and Coordination for Autonomous Vehicles (2017, methodology): “Poorly selected points may result in inadequate and inefficient segmentations, and different choices of seed points usually lead to different segmentations [25].”
Opening up Open World Tracking (2021, methodology): “Early methods in vision-based tracking [27,59,85] and robotic perception [54,76] utilized class-agnostic, bottom-up segmentation as a tracking cue, e.g ., based on LiDAR point cloud clustering [53, 79] or background modeling and foreground grouping [32,73,85].”
Performance of global descriptors for velodyne-based urban object recognition (2014, methodology): “For every scan (the set of points collected during the Velodyne LIDAR spins once), we assume that the ground segmentation and object cluster steps ( [1], [2], [3], [4], [5], [6]) have already been performed, and the outputs of the two steps are different objects.”
Fast segmentation of 3D point clouds for ground vehicles (2010, background): “Performance of [6] is reported to be close to, but not up to real-time, whereas in [7] real-time performance 2010 IEEE Intelligent Vehicles Symposium University of California, San Diego, CA, USA June 21-24, 2010 WeB1.”