scholarly journals Uncertainty Estimation in Deep Neural Networks for Point Cloud Segmentation in Factory Planning

Modelling ◽  
2021 ◽  
Vol 2 (1) ◽  
pp. 1-17
Author(s):  
Christina Petschnigg ◽  
Jürgen Pilz
Keyword(s):  
Point Cloud ◽  
Point Clouds ◽  
Scientific Data ◽  
Production Plant ◽  
Data Set ◽  
Factory Planning ◽  
3D Data ◽  
Future Production ◽  
Point Cloud Segmentation ◽  
Approximate Bayesian

The digital factory provides undoubtedly great potential for future production systems in terms of efficiency and effectivity. A key aspect on the way to realize the digital copy of a real factory is the understanding of complex indoor environments on the basis of three-dimensional (3D) data. In order to generate an accurate factory model including the major components, i.e., building parts, product assets, and process details, the 3D data that are collected during digitalization can be processed with advanced methods of deep learning. For instance, the semantic segmentation of a point cloud enables the identification of relevant objects within the environment. In this work, we propose a fully Bayesian and an approximate Bayesian neural network for point cloud segmentation. Both of the networks are used within a workflow in order to generate an environment model on the basis of raw point clouds. The Bayesian and approximate Bayesian networks allow us to analyse how different ways of estimating uncertainty in these networks improve segmentation results on raw point clouds. We achieve superior model performance for both, the Bayesian and the approximate Bayesian model compared to the frequentist one. This performance difference becomes even more striking when incorporating the networks’ uncertainty in their predictions. For evaluation, we use the scientific data set S3DIS as well as a data set, which was collected by the authors at a German automotive production plant. The methods proposed in this work lead to more accurate segmentation results and the incorporation of uncertainty information also makes this approach especially applicable to safety critical applications aside from our factory planning use case.

scholarly journals Panicle-3D: Efficient Phenotyping Tool for Precise Semantic Segmentation of Rice Panicle Point Cloud

2021 ◽  
Vol 2021 ◽  
pp. 1-9
Author(s):  
Liang Gong ◽  
Xiaofeng Du ◽  
Kai Zhu ◽  
Ke Lin ◽  
Qiaojun Lou ◽  
...  
Keyword(s):  
Point Cloud ◽  
Network Models ◽  
Semantic Segmentation ◽  
Spatial Location ◽  
Point Clouds ◽  
Utilization Efficiency ◽  
Data Set ◽  
Cloud Data ◽  
Point Cloud Segmentation ◽  
Segmentation Accuracy

The automated measurement of crop phenotypic parameters is of great significance to the quantitative study of crop growth. The segmentation and classification of crop point cloud help to realize the automation of crop phenotypic parameter measurement. At present, crop spike-shaped point cloud segmentation has problems such as fewer samples, uneven distribution of point clouds, occlusion of stem and spike, disorderly arrangement of point clouds, and lack of targeted network models. The traditional clustering method can realize the segmentation of the plant organ point cloud with relatively independent spatial location, but the accuracy is not acceptable. This paper first builds a desktop-level point cloud scanning apparatus based on a structured-light projection module to facilitate the point cloud acquisition process. Then, the rice ear point cloud was collected, and the rice ear point cloud data set was made. In addition, data argumentation is used to improve sample utilization efficiency and training accuracy. Finally, a 3D point cloud convolutional neural network model called Panicle-3D was designed to achieve better segmentation accuracy. Specifically, the design of Panicle-3D is aimed at the multiscale characteristics of plant organs, combined with the structure of PointConv and long and short jumps, which accelerates the convergence speed of the network and reduces the loss of features in the process of point cloud downsampling. After comparison experiments, the segmentation accuracy of Panicle-3D reaches 93.4%, which is higher than PointNet. Panicle-3D is suitable for other similar crop point cloud segmentation tasks.

scholarly journals LatticeNet: fast spatio-temporal point cloud segmentation using permutohedral lattices

2021 ◽  
Author(s):  
Radu Alexandru Rosu ◽  
Peer Schütt ◽  
Jan Quenzel ◽  
Sven Behnke
Keyword(s):  
Point Cloud ◽  
Semantic Segmentation ◽  
Point Clouds ◽  
Local Geometry ◽  
Deep Convolutional Neural Networks ◽  
3D Data ◽  
Novel Approach ◽  
Point Cloud Segmentation ◽  
Spatio Temporal ◽  
Fast Convolutions

AbstractDeep convolutional neural networks have shown outstanding performance in the task of semantically segmenting images. Applying the same methods on 3D data still poses challenges due to the heavy memory requirements and the lack of structured data. Here, we propose LatticeNet, a novel approach for 3D semantic segmentation, which takes raw point clouds as input. A PointNet describes the local geometry which we embed into a sparse permutohedral lattice. The lattice allows for fast convolutions while keeping a low memory footprint. Further, we introduce DeformSlice, a novel learned data-dependent interpolation for projecting lattice features back onto the point cloud. We present results of 3D segmentation on multiple datasets where our method achieves state-of-the-art performance. We also extend and evaluate our network for instance and dynamic object segmentation.

scholarly journals ON A KNOWLEDGE-BASED APPROACH TO THE CLASSIFICATION OF MOBILE LASER SCANNING POINT CLOUDS

2018 ◽  
Vol XLII-4 ◽  
pp. 343-349 ◽  
Author(s):  
M. Lemmens
Keyword(s):  
Point Cloud ◽  
Laser Scanning ◽  
Decision Rules ◽  
Point Clouds ◽  
Bench Mark ◽  
Automated Classification ◽  
Initial Experiment ◽  
Data Set ◽  
Knowledge Based

<p><strong>Abstract.</strong> A knowledge-based system exploits the knowledge, which a human expert uses for completing a complex task, through a database containing decision rules, and an inference engine. Already in the early nineties knowledge-based systems have been proposed for automated image classification. Lack of success faded out initial interest and enthusiasm, the same fate neural networks struck at that time. Today the latter enjoy a steady revival. This paper aims at demonstrating that a knowledge-based approach to automated classification of mobile laser scanning point clouds has promising prospects. An initial experiment exploiting only two features, height and reflectance value, resulted in an overall accuracy of 79<span class="thinspace"></span>% for the Paris-rue-Madame point cloud bench mark data set.</p>

scholarly journals An Efficient Encoding Voxel-Based Segmentation (EVBS) Algorithm Based on Fast Adjacent Voxel Search for Point Cloud Plane Segmentation

2019 ◽  
Vol 11 (23) ◽  
pp. 2727 ◽  
Author(s):  
Ming Huang ◽  
Pengcheng Wei ◽  
Xianglei Liu
Keyword(s):  
Point Cloud ◽  
Initial Point ◽  
Computation Time ◽  
Region Growing ◽  
Point Clouds ◽  
Recall Rate ◽  
Seed Selection ◽  
Cloud Processing ◽  
Point Cloud Segmentation ◽  
Noise Data

Plane segmentation is a basic yet important process in light detection and ranging (LiDAR) point cloud processing. The traditional point cloud plane segmentation algorithm is typically affected by the number of point clouds and the noise data, which results in slow segmentation efficiency and poor segmentation effect. Hence, an efficient encoding voxel-based segmentation (EVBS) algorithm based on a fast adjacent voxel search is proposed in this study. First, a binary octree algorithm is proposed to construct the voxel as the segmentation object and code the voxel, which can compute voxel features quickly and accurately. Second, a voxel-based region growing algorithm is proposed to cluster the corresponding voxel to perform the initial point cloud segmentation, which can improve the rationality of seed selection. Finally, a refining point method is proposed to solve the problem of under-segmentation in unlabeled voxels by judging the relationship between the points and the segmented plane. Experimental results demonstrate that the proposed algorithm is better than the traditional algorithm in terms of computation time, extraction accuracy, and recall rate.

scholarly journals A New Framework For Interactive Segmentation of Point Clouds

2014 ◽  
Vol XL-5 ◽  
pp. 357-362 ◽  
Author(s):  
K. Liu ◽  
J. Boehm
Keyword(s):  
Point Cloud ◽  
Fundamental Problem ◽  
Weighted Graph ◽  
Point Clouds ◽  
Target Object ◽  
General Point ◽  
Interactive Segmentation ◽  
Mobile Mapping ◽  
Point Cloud Segmentation ◽  
New Framework

Point cloud segmentation is a fundamental problem in point processing. Segmenting a point cloud fully automatically is very challenging due to the property of point cloud as well as different requirements of distinct users. In this paper, an interactive segmentation method for point clouds is proposed. Only two strokes need to be drawn intuitively to indicate the target object and the background respectively. The draw strokes are sparse and don't necessarily cover the whole object. Given the strokes, a weighted graph is built and the segmentation is formulated as a minimization problem. The problem is solved efficiently by using the Max Flow Min Cut algorithm. In the experiments, the mobile mapping data of a city area is utilized. The resulting segmentations demonstrate the efficiency of the method that can be potentially applied for general point clouds.

scholarly journals THE CONTRIBUTION OF GEOMATIC TECHNOLOGIES TO BIM

2019 ◽  
Vol XLII-5/W3 ◽  
pp. 85-89
Author(s):  
K. Pavelka ◽  
E. Matoušková ◽  
K. Pavelka jr.
Keyword(s):  
Point Cloud ◽  
Laser Scanning ◽  
Modern Technology ◽  
Data Conversion ◽  
Digital Information ◽  
Historical Building ◽  
Data Set ◽  
Common Definition ◽  
3D Data ◽  
Building Information

Abstract. There are many definitions of the commonly used abbreviation BIM, but one can say that each user or data supplier has different idea about it. There can be an economic view, or other aspects like surveying, material, engineering, maintenance, etc. The common definition says that Building Information Modelling or Building Information Management (BIM) is a digital model representing a physical and functional object with its characteristics. The model serves as a database of object information for its design, construction and operation over its life cycle, i.e. from the initial concept to the removal of the building. BIM is a collection of interconnected digital information in both protected and open formats, recording graphical and non-graphical data on model elements. There are two facets: a) BIM created simultaneously with the project, or project designed directly in BIM (it is typical of new objects designed in CAD systems - for example in the Revit software) or b) BIM for old or historical objects. The former is a modern technology, which is nowadays used worldwide. From the engineer’s perspective, the issue is the creation of BIM for older objects. In this case, it is crucial to obtain a precise 3D data set - complex 3D documentation of an object is needed and it is created using various surveying techniques. The most popular technique is laser scanning or digital automatic photogrammetry, from which a point cloud is derived. But this is not the main result. While classical geodesy gives selective localized information, the above-mentioned technologies give unselected information and provide huge datasets. Fully automatic technologies that would select important information from the point cloud are still under development. This seems to be a task for the coming years. Large amounts of data can be acquired automatically and quickly, but getting the expected information is another matter. These problems will be analysed in this paper. Data conversion to BIM, especially for older objects, will be shown on several case studies. The first is an older technical building complex transferred to BIM, the second one is a historical building, and the third one will be a historic medieval bridge (Charles Bridge in Prague). The last part of this paper will refer to aspects and benefits of using Virtual Reality in BIM.

scholarly journals A Novel Recursive Non-Parametric DBSCAN Algorithm for 3D Data Analysis with an Application in Rockfall Detection

2021 ◽  
Vol 16 (4) ◽  
pp. 579-587
Author(s):  
Pitisit Dillon ◽  
Pakinee Aimmanee ◽  
Akihiko Wakai ◽  
Go Sato ◽  
Hoang Viet Hung ◽  
...  
Keyword(s):  
Point Cloud ◽  
Crack Detection ◽  
Spatial Clustering ◽  
Point Clouds ◽  
Divide And Conquer ◽  
Dbscan Algorithm ◽  
3D Data ◽  
Data Points ◽  
Clustering Data ◽  
Better Than

The density-based spatial clustering of applications with noise (DBSCAN) algorithm is a well-known algorithm for spatial-clustering data point clouds. It can be applied to many applications, such as crack detection, rockfall detection, and glacier movement detection. Traditional DBSCAN requires two predefined parameters. Suitable values of these parameters depend upon the distribution of the input point cloud. Therefore, estimating these parameters is challenging. This paper proposed a new version of DBSCAN that can automatically customize the parameters. The proposed method consists of two processes: initial parameter estimation based on grid analysis and DBSCAN based on the divide-and-conquer (DC-DBSCAN) approach, which repeatedly performs DBSCAN on each cluster separately and recursively. To verify the proposed method, we applied it to a 3D point cloud dataset that was used to analyze rockfall events at the Puiggcercos cliff, Spain. The total number of data points used in this study was 15,567. The experimental results show that the proposed method is better than the traditional DBSCAN in terms of purity and NMI scores. The purity scores of the proposed method and the traditional DBSCAN method were 96.22% and 91.09%, respectively. The NMI scores of the proposed method and the traditional DBSCAN method are 0.78 and 0.49, respectively. Also, it can detect events that traditional DBSCAN cannot detect.

scholarly journals Clustering-Based Plane Refitting of Non-planar Patches for Voxel-Based 3D Point Cloud Segmentation Using K-Means Clustering

2020 ◽  
Vol 37 (6) ◽  
pp. 1019-1027
Author(s):  
Ali Saglam ◽  
Hasan B. Makineci ◽  
Ömer K. Baykan ◽  
Nurdan Akhan Baykan
Keyword(s):  
Point Cloud ◽  
Point Group ◽  
Three Dimensional ◽  
Point Clouds ◽  
Segmentation Method ◽  
Train Station ◽  
Cloud Processing ◽  
Point Cloud Processing ◽  
Point Cloud Segmentation ◽  
Geometric Cues

Point cloud processing is a struggled field because the points in the clouds are three-dimensional and irregular distributed signals. For this reason, the points in the point clouds are mostly sampled into regularly distributed voxels in the literature. Voxelization as a pretreatment significantly accelerates the process of segmenting surfaces. The geometric cues such as plane directions (normals) in the voxels are mostly used to segment the local surfaces. However, the sampling process may include a non-planar point group (patch), which is mostly on the edges and corners, in a voxel. These voxels can cause misleading the segmentation process. In this paper, we separate the non-planar patches into planar sub-patches using k-means clustering. The largest one among the planar sub-patches replaces the normal and barycenter properties of the voxel with those of itself. We have tested this process in a successful point cloud segmentation method and measure the effects of the proposed method on two point cloud segmentation datasets (Mosque and Train Station). The method increases the accuracy success of the Mosque dataset from 83.84% to 87.86% and that of the Train Station dataset from 85.36% to 87.07%.

scholarly journals COMBINATION OF TLS POINT CLOUDS AND 3D DATA FROM KINECT V2 SENSOR TO COMPLETE INDOOR MODELS

2016 ◽  
Vol XLI-B5 ◽  
pp. 659-666 ◽  
Author(s):  
E. Lachat ◽  
T. Landes ◽  
P. Grussenmeyer
Keyword(s):  
Point Cloud ◽  
Low Cost ◽  
Measurement Techniques ◽  
Laser Scanner ◽  
Point Clouds ◽  
Final Model ◽  
Kinect V2 ◽  
Multiple Sensors ◽  
3D Data ◽  
Final Data

The combination of data coming from multiple sensors is more and more applied for remote sensing issues (multi-sensor imagery) but also in cultural heritage or robotics, since it often results in increased robustness and accuracy of the final data. In this paper, the reconstruction of building elements such as window frames or door jambs scanned thanks to a low cost 3D sensor (Kinect v2) is presented. Their combination within a global point cloud of an indoor scene acquired with a terrestrial laser scanner (TLS) is considered. If the added elements acquired with the Kinect sensor enable to reach a better level of detail of the final model, an adapted acquisition protocol may also provide several benefits as for example time gain. The paper aims at analyzing whether the two measurement techniques can be complementary in this context. The limitations encountered during the acquisition and reconstruction steps are also investigated.

scholarly journals SEGMENTATION OF LARGE UNSTRUCTURED POINT CLOUDS USING OCTREE-BASED REGION GROWING AND CONDITIONAL RANDOM FIELDS

2017 ◽  
Vol XLII-2/W8 ◽  
pp. 25-30 ◽  
Author(s):  
M. Bassier ◽  
M. Bonduel ◽  
B. Van Genechten ◽  
M. Vergauwen
Keyword(s):  
Random Fields ◽  
Point Cloud ◽  
Conditional Random Fields ◽  
Conditional Random Field ◽  
Region Growing ◽  
Point Clouds ◽  
Single Type ◽  
Point Cloud Data ◽  
Cloud Data ◽  
Point Cloud Segmentation

Point cloud segmentation is a crucial step in scene understanding and interpretation. The goal is to decompose the initial data into sets of workable clusters with similar properties. Additionally, it is a key aspect in the automated procedure from point cloud data to BIM. Current approaches typically only segment a single type of primitive such as planes or cylinders. Also, current algorithms suffer from oversegmenting the data and are often sensor or scene dependent.<br><br> In this work, a method is presented to automatically segment large unstructured point clouds of buildings. More specifically, the segmentation is formulated as a graph optimisation problem. First, the data is oversegmented with a greedy octree-based region growing method. The growing is conditioned on the segmentation of planes as well as smooth surfaces. Next, the candidate clusters are represented by a Conditional Random Field after which the most likely configuration of candidate clusters is computed given a set of local and contextual features. The experiments prove that the used method is a fast and reliable framework for unstructured point cloud segmentation. Processing speeds up to 40,000 points per second are recorded for the region growing. Additionally, the recall and precision of the graph clustering is approximately 80%. Overall, nearly 22% of oversegmentation is reduced by clustering the data. These clusters will be classified and used as a basis for the reconstruction of BIM models.

Sign in / Sign up

Export Citation Format

Share Document