Classification method of LiDAR point cloud based on threedimensional convolutional neural network

Zhao, Zhongyang; Cheng, Yinglei; Xiaosong, Shi; Qin, Xianxiang

doi:10.1088/1742-6596/1168/6/062013

Cited by 5 publications

(2 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In most cases, the input data requires to be preprocessed before feeding it into an ANN to filter irrelevant data, as well as transform the data into a more explicit format or to execute a first feature extraction [2]- [4]. The selection of the preprocessing technique depends on the input data type as well as the purpose of the application.…”

Section: Introductionmentioning

confidence: 99%

Investigation for the Need of Traditional Data-Preprocessing when Applying Artificial Neural Networks to FMCW-Radar Data

Valtl

Mèndez

Mauro

et al. 2022

2022 29th International Conference on Systems, Signals and Image Processing (IWSSIP)

View full text Add to dashboard Cite

Robust functionality of autonomous driving vehicles relies on their ability to detect obstables and various scenarios on the road. This can be only achieved by applying robust, fast and efficient AI-based signal processing to radar data. In this work we present an empirical investigation on the question, whether one can apply artificial neural networks (ANNs) directly to frequency modulated continuous wave (FMCW) radar raw data. We show that preproceessing is not necessary if one has enough raw data.In our experiment we have data of 153 648 frames collected with a 60 GHz FMCW radar. We compare systematically the options of preprocessing the data using variational autoencoder, applying traditional preprocessing or omit data-preprocessing and apply ANN directly to raw data. We show that the last option results in 28% faster signal processing and highest accuracy. This is a promising result, since it enables edge computing and direct signal processing at the sensor level.

show abstract

Section: Introductionmentioning

confidence: 99%

Investigation for the Need of Traditional Data-Preprocessing when Applying Artificial Neural Networks to FMCW-Radar Data

Valtl

Mèndez

Mauro

et al. 2022

2022 29th International Conference on Systems, Signals and Image Processing (IWSSIP)

View full text Add to dashboard Cite

show abstract

“…The basis of LiDAR data processing is an unclassified point cloud. In order to extract classes of buildings from this cloud, it is necessary to filter the data using classification methods and techniques such as the support vector machine (SVM) [31], methods using local neighborhood statistics [32], and methods based on neural networks [33] with improvements [34].…”

Section: Introductionmentioning

confidence: 99%

Identification of Roof Surfaces from LiDAR Cloud Points by GIS Tools: A Case Study of Lučenec, Slovakia

et al. 2020

View full text Add to dashboard Cite

The identification of roof surfaces is characterized by a sequence of several processing steps. The boundary detection of different types of roof is realized from light detection and ranging (LiDAR) cloud points and can confirm the real boundary of the roof. In the process of processing LiDAR data, shortcomings have been found regarding the inappropriate classification of points (class 6 “buildings”) concerning the roofs (the points of the building facade were marked as outliers and reclassified). In cases of insufficient point density, there is a problem with not being able to capture either the roof boundary or small roof objects, along with the possible occurrence of gaps inside the roof areas. This study proposes a processing procedure in a geographic information system (GIS) environment that advocates the identification of roof surfaces based on the LiDAR point cloud. We created the contours of a roof surface boundary with a simplified regular shape. From 824 roofs in the studied area, six different types of roof were selected, which this study presents in detail. The expected result of the study is the generation of segments inside the roof boundary. The study also includes the visualization of the outcomes of the spatial analyses of the identified roof surfaces, which forms the basis for determining the potential of solar systems with respect to green roofs for the development of smart city buildings.

show abstract

Fully Automatic Point Cloud Analysis for Powerline Corridor Mapping

Nardinocchi

Balsi

Esposito

2020

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

Classification method of LiDAR point cloud based on threedimensional convolutional neural network

Cited by 5 publications

References 3 publications

Investigation for the Need of Traditional Data-Preprocessing when Applying Artificial Neural Networks to FMCW-Radar Data

Investigation for the Need of Traditional Data-Preprocessing when Applying Artificial Neural Networks to FMCW-Radar Data

Identification of Roof Surfaces from LiDAR Cloud Points by GIS Tools: A Case Study of Lučenec, Slovakia

Fully Automatic Point Cloud Analysis for Powerline Corridor Mapping

Contact Info

Product

Resources

About