A Deep Learning Streaming Methodology for Trajectory Classification
Due to the vast amount of available tracking sensors in recent years, high-frequency and high-volume streams of data are generated every day. The maritime domain is no different as all larger vessels are obliged to be equipped with a vessel tracking system that transmits their location periodically....
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-04-01
|
Series: | ISPRS International Journal of Geo-Information |
Subjects: | |
Online Access: | https://www.mdpi.com/2220-9964/10/4/250 |
_version_ | 1797538338427633664 |
---|---|
author | Ioannis Kontopoulos Antonios Makris Konstantinos Tserpes |
author_facet | Ioannis Kontopoulos Antonios Makris Konstantinos Tserpes |
author_sort | Ioannis Kontopoulos |
collection | DOAJ |
description | Due to the vast amount of available tracking sensors in recent years, high-frequency and high-volume streams of data are generated every day. The maritime domain is no different as all larger vessels are obliged to be equipped with a vessel tracking system that transmits their location periodically. Consequently, automated methodologies able to extract meaningful information from high-frequency, large volumes of vessel tracking data need to be developed. The automatic identification of vessel mobility patterns from such data in real time is of utmost importance since it can reveal abnormal or illegal vessel activities in due time. Therefore, in this work, we present a novel approach that transforms streaming vessel trajectory patterns into images and employs deep learning algorithms to accurately classify vessel activities in near real time tackling the Big Data challenges of volume and velocity. Two real-world data sets collected from terrestrial, vessel-tracking receivers were used to evaluate the proposed methodology in terms of both classification and streaming execution performance. Experimental results demonstrated that the vessel activity classification performance can reach an accuracy of over <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>96</mn><mo>%</mo></mrow></semantics></math></inline-formula> while achieving sub-second latencies in streaming execution performance. |
first_indexed | 2024-03-10T12:30:04Z |
format | Article |
id | doaj.art-90986eba6fe44fd1a59e534ac2a1173b |
institution | Directory Open Access Journal |
issn | 2220-9964 |
language | English |
last_indexed | 2024-03-10T12:30:04Z |
publishDate | 2021-04-01 |
publisher | MDPI AG |
record_format | Article |
series | ISPRS International Journal of Geo-Information |
spelling | doaj.art-90986eba6fe44fd1a59e534ac2a1173b2023-11-21T14:45:04ZengMDPI AGISPRS International Journal of Geo-Information2220-99642021-04-0110425010.3390/ijgi10040250A Deep Learning Streaming Methodology for Trajectory ClassificationIoannis Kontopoulos0Antonios Makris1Konstantinos Tserpes2Department of Informatics and Telematics, Harokopio University of Athens, 9 Omirou Str., 17778 Athens, GreeceDepartment of Informatics and Telematics, Harokopio University of Athens, 9 Omirou Str., 17778 Athens, GreeceDepartment of Informatics and Telematics, Harokopio University of Athens, 9 Omirou Str., 17778 Athens, GreeceDue to the vast amount of available tracking sensors in recent years, high-frequency and high-volume streams of data are generated every day. The maritime domain is no different as all larger vessels are obliged to be equipped with a vessel tracking system that transmits their location periodically. Consequently, automated methodologies able to extract meaningful information from high-frequency, large volumes of vessel tracking data need to be developed. The automatic identification of vessel mobility patterns from such data in real time is of utmost importance since it can reveal abnormal or illegal vessel activities in due time. Therefore, in this work, we present a novel approach that transforms streaming vessel trajectory patterns into images and employs deep learning algorithms to accurately classify vessel activities in near real time tackling the Big Data challenges of volume and velocity. Two real-world data sets collected from terrestrial, vessel-tracking receivers were used to evaluate the proposed methodology in terms of both classification and streaming execution performance. Experimental results demonstrated that the vessel activity classification performance can reach an accuracy of over <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>96</mn><mo>%</mo></mrow></semantics></math></inline-formula> while achieving sub-second latencies in streaming execution performance.https://www.mdpi.com/2220-9964/10/4/250trajectory classificationdeep learningneural networkscomputer visiondistributed processingstream processing |
spellingShingle | Ioannis Kontopoulos Antonios Makris Konstantinos Tserpes A Deep Learning Streaming Methodology for Trajectory Classification ISPRS International Journal of Geo-Information trajectory classification deep learning neural networks computer vision distributed processing stream processing |
title | A Deep Learning Streaming Methodology for Trajectory Classification |
title_full | A Deep Learning Streaming Methodology for Trajectory Classification |
title_fullStr | A Deep Learning Streaming Methodology for Trajectory Classification |
title_full_unstemmed | A Deep Learning Streaming Methodology for Trajectory Classification |
title_short | A Deep Learning Streaming Methodology for Trajectory Classification |
title_sort | deep learning streaming methodology for trajectory classification |
topic | trajectory classification deep learning neural networks computer vision distributed processing stream processing |
url | https://www.mdpi.com/2220-9964/10/4/250 |
work_keys_str_mv | AT ioanniskontopoulos adeeplearningstreamingmethodologyfortrajectoryclassification AT antoniosmakris adeeplearningstreamingmethodologyfortrajectoryclassification AT konstantinostserpes adeeplearningstreamingmethodologyfortrajectoryclassification AT ioanniskontopoulos deeplearningstreamingmethodologyfortrajectoryclassification AT antoniosmakris deeplearningstreamingmethodologyfortrajectoryclassification AT konstantinostserpes deeplearningstreamingmethodologyfortrajectoryclassification |