Point Cloud Semantic Segmentation Network Based on Multi-Scale Feature Fusion

Du, Juan; Jiang, Zuning; Huang, Shangfeng; Wang, Zongyue; Su, Jinhe; Su, Songjian; Wu, Yundong; Cai, Guorong

doi:10.3390/s21051625

Cited by 13 publications

(9 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The most common application of image fusion in LiDAR remote sensing is the fusion of 3D point clouds and RGB images to train a deep learning model for classification and segmentation tasks [ 20 , 21 , 22 ]. The features extracted from both types of data are used to enhance the performance of each class in the application of each class.…”

Section: Point Cloud Computingmentioning

confidence: 99%

Deep Learning for LiDAR Point Cloud Classification in Remote Sensing

Diab

Kashef

Shaker

2022

Sensors

View full text Add to dashboard Cite

Point clouds are one of the most widely used data formats produced by depth sensors. There is a lot of research into feature extraction from unordered and irregular point cloud data. Deep learning in computer vision achieves great performance for data classification and segmentation of 3D data points as point clouds. Various research has been conducted on point clouds and remote sensing tasks using deep learning (DL) methods. However, there is a research gap in providing a road map of existing work, including limitations and challenges. This paper focuses on introducing the state-of-the-art DL models, categorized by the structure of the data they consume. The models’ performance is collected, and results are provided for benchmarking on the most used datasets. Additionally, we summarize the current benchmark 3D datasets publicly available for DL training and testing. In our comparative study, we can conclude that convolutional neural networks (CNNs) achieve the best performance in various remote-sensing applications while being light-weighted models, namely Dynamic Graph CNN (DGCNN) and ConvPoint.

show abstract

Section: Point Cloud Computingmentioning

confidence: 99%

Deep Learning for LiDAR Point Cloud Classification in Remote Sensing

Diab

Kashef

Shaker

2022

Sensors

View full text Add to dashboard Cite

show abstract

“…For instance, PointNet++ [14] exploited Furthest Point Sampling (FPS) to hierarchically downsample point clouds and iteratively extract features from PointNet in each sampling layer. In order to describe large-scale scenes from multi-resolutions, MSSCN [30] concatenated point features with different densities, PointSIFT [31] paid attention to encoding both multi-orientations and multi-scales for local details, and PointCNN [32] employed a fully convolutional point network with a series of abstraction layers, feature learners at different scales, and a merging layer. When the random sampling strategy in large-scale scenes is time-consuming since it works on the original points, the use of supervoxels [33], which were inspired by superpixels in 2D image processing, greatly cuts down the number of points and provides a more natural and compact representation for local operations.…”

Section: Deep Learning Network For Semantic Segmentationmentioning

confidence: 99%

IAGC: Interactive Attention Graph Convolution Network for Semantic Segmentation of Point Clouds in Building Indoor Environment

Zhai

Zou

et al. 2022

IJGI

View full text Add to dashboard Cite

Point-based networks have been widely used in the semantic segmentation of point clouds owing to the powerful 3D convolution neural network (CNN) baseline. Most of the current methods resort to intermediate regular representations for reorganizing the structure of point clouds for 3D CNN networks, but they may neglect the inherent contextual information. In our work, we focus on capturing discriminative features with the interactive attention mechanism and propose a novel method consisting of the regional simplified dual attention network and global graph convolution network. Firstly, we cluster homogeneous points into superpoints and construct a superpoint graph to effectively reduce the computation complexity and greatly maintain spatial topological relations among superpoints. Secondly, we integrate cross-position attention and cross-channel attention into a single head attention module and design a novel interactive attention gating (IAG)-based multilayer perceptron (MLP) network (IAG–MLP), which is utilized for the expansion of the receptive field and augmentation of discriminative features in local embeddings. Afterwards, the combination of stacked IAG–MLP blocks and the global graph convolution network, called IAGC, is proposed to learn high-dimensional local features in superpoints and progressively update these local embeddings with the recurrent neural network (RNN) network. Our proposed framework is evaluated on three indoor open benchmarks, and the 6-fold cross-validation results of the S3DIS dataset show that the local IAG–MLP network brings about 1% and 6.1% improvement in overall accuracy (OA) and mean class intersection-over-union (mIoU), respectively, compared with the PointNet local network. Furthermore, our IAGC network outperforms other CNN-based approaches in the ScanNet V2 dataset by at least 7.9% in mIoU. The experimental results indicate that the proposed method can better capture contextual information and achieve competitive overall performance in the semantic segmentation task.

show abstract

“…Camera data is fused with LiDAR data in order to detect better objects [ 26 ]. In some works, the detection of objects is approached by performing semantic segmentation on LiDAR data [ 29 , 30 ] or camera-LiDAR fused data [ 31 ].…”

Section: Related Workmentioning

confidence: 99%

Obstacle Detection Using a Facet-Based Representation from 3-D LiDAR Measurements

Dulău

Oniga

2021

Sensors

View full text Add to dashboard Cite

In this paper, we propose an obstacle detection approach that uses a facet-based obstacle representation. The approach has three main steps: ground point detection, clustering of obstacle points, and facet extraction. Measurements from a 64-layer LiDAR are used as input. First, ground points are detected and eliminated in order to select obstacle points and create object instances. To determine the objects, obstacle points are grouped using a channel-based clustering approach. For each object instance, its contour is extracted and, using an RANSAC-based approach, the obstacle facets are selected. For each processing stage, optimizations are proposed in order to obtain a better runtime. For the evaluation, we compare our proposed approach with an existing approach, using the KITTI benchmark dataset. The proposed approach has similar or better results for some obstacle categories but a lower computational complexity.

show abstract

Point Cloud Semantic Segmentation Network Based on Multi-Scale Feature Fusion

Cited by 13 publications

References 49 publications

Deep Learning for LiDAR Point Cloud Classification in Remote Sensing

Deep Learning for LiDAR Point Cloud Classification in Remote Sensing

IAGC: Interactive Attention Graph Convolution Network for Semantic Segmentation of Point Clouds in Building Indoor Environment

Obstacle Detection Using a Facet-Based Representation from 3-D LiDAR Measurements

Contact Info

Product

Resources

About