Deep Learning Approach to Update Road Network using VGI Data

Prajowal Manandhar, Prashanth Marpu, Zeyar Aung

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

In our earlier work, we worked on extraction of the total width of road by agents traversing in the direction guided by Volunteered Geographic Information (VGI). The only downfall of VGI approach is its inability to update the new road developments. In this paper, we introduce deep learning approach to update the road network. We make use of the output of our previous work which forms as an input to train the Convolutional Neural Network (CNN). Then, further post processing is performed to remove non-road segments (such as buildings, vegetation, etc) on the output of CNN and finally, obtain the updated road map.

Original languageBritish English
Title of host publication2018 International Conference on Signal Processing and Information Security, ICSPIS 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728102573
DOIs
StatePublished - 14 Feb 2019
Event2018 International Conference on Signal Processing and Information Security, ICSPIS 2018 - Dubai, United Arab Emirates
Duration: 7 Nov 20188 Nov 2018

Publication series

Name2018 International Conference on Signal Processing and Information Security, ICSPIS 2018

Conference

Conference2018 International Conference on Signal Processing and Information Security, ICSPIS 2018
Country/TerritoryUnited Arab Emirates
CityDubai
Period7/11/188/11/18

Fingerprint

Dive into the research topics of 'Deep Learning Approach to Update Road Network using VGI Data'. Together they form a unique fingerprint.

Cite this