Re-identification of vessels with convolutional neural networks

Research output: Contribution to conferencePaperAcademic

Abstract

In order to perform a reliable vessel behavior analysis for maritime surveillance, re-identification of previously detected vessels, passing through new camera locations, is of vital importance. However, challenging outdoor conditions of the maritime environment heavily restrict the application of conventional methods. Additionally, vessels are large objects and capturing a vessel from different viewpoints may provide entirely different visual appearances. To address these challenges, this paper proposes an Identity Oriented Re-identification network (IORnet) for the re-identification of vessels. This CNN-based approach incorporates the triplet loss method combined with a new loss function, which leads to improved vessel reidentification. Experimental results on our real-world evaluation dataset reveal that the proposed method achieves 81.5% and 91.2% on mAP and Rank1 scores, respectively. As an additional contribution, we also provide our annotated vessel reidentification dataset to the open public access.

Conference

Conference5th International Conference on Computer and Technology Applications, (ICCTA2019)
Abbreviated titleICCTA2019
CountryTurkey
CityIstanbul
Period16/04/1917/04/19
Internet address

Fingerprint

Neural networks
Cameras

Keywords

  • CNNs
  • Maritime dataset
  • Maritime surveillance
  • Vessel re-identification

Cite this

Ghahremani, A., Kong, Y., Bondarev, E., & de With, P. H. N. (2019). Re-identification of vessels with convolutional neural networks. 93-97. Paper presented at 5th International Conference on Computer and Technology Applications, (ICCTA2019), Istanbul, Turkey.DOI: 10.1145/3323933.3324075
Ghahremani, Amir ; Kong, Yitian ; Bondarev, Egor ; de With, Peter H.N./ Re-identification of vessels with convolutional neural networks. Paper presented at 5th International Conference on Computer and Technology Applications, (ICCTA2019), Istanbul, Turkey.5 p.
@conference{7cd45f15d4a94a08982447acc4d609c8,
title = "Re-identification of vessels with convolutional neural networks",
abstract = "In order to perform a reliable vessel behavior analysis for maritime surveillance, re-identification of previously detected vessels, passing through new camera locations, is of vital importance. However, challenging outdoor conditions of the maritime environment heavily restrict the application of conventional methods. Additionally, vessels are large objects and capturing a vessel from different viewpoints may provide entirely different visual appearances. To address these challenges, this paper proposes an Identity Oriented Re-identification network (IORnet) for the re-identification of vessels. This CNN-based approach incorporates the triplet loss method combined with a new loss function, which leads to improved vessel reidentification. Experimental results on our real-world evaluation dataset reveal that the proposed method achieves 81.5{\%} and 91.2{\%} on mAP and Rank1 scores, respectively. As an additional contribution, we also provide our annotated vessel reidentification dataset to the open public access.",
keywords = "CNNs, Maritime dataset, Maritime surveillance, Vessel re-identification",
author = "Amir Ghahremani and Yitian Kong and Egor Bondarev and {de With}, {Peter H.N.}",
year = "2019",
month = "1",
day = "1",
doi = "10.1145/3323933.3324075",
language = "English",
pages = "93--97",
note = "5th International Conference on Computer and Technology Applications, (ICCTA2019), ICCTA2019 ; Conference date: 16-04-2019 Through 17-04-2019",
url = "http://www.wikicfp.com/cfp/servlet/event.showcfp?eventid=79331",

}

Ghahremani, A, Kong, Y, Bondarev, E & de With, PHN 2019, 'Re-identification of vessels with convolutional neural networks' Paper presented at 5th International Conference on Computer and Technology Applications, (ICCTA2019), Istanbul, Turkey, 16/04/19 - 17/04/19, pp. 93-97. DOI: 10.1145/3323933.3324075

Re-identification of vessels with convolutional neural networks. / Ghahremani, Amir; Kong, Yitian; Bondarev, Egor; de With, Peter H.N.

2019. 93-97 Paper presented at 5th International Conference on Computer and Technology Applications, (ICCTA2019), Istanbul, Turkey.

Research output: Contribution to conferencePaperAcademic

TY - CONF

T1 - Re-identification of vessels with convolutional neural networks

AU - Ghahremani,Amir

AU - Kong,Yitian

AU - Bondarev,Egor

AU - de With,Peter H.N.

PY - 2019/1/1

Y1 - 2019/1/1

N2 - In order to perform a reliable vessel behavior analysis for maritime surveillance, re-identification of previously detected vessels, passing through new camera locations, is of vital importance. However, challenging outdoor conditions of the maritime environment heavily restrict the application of conventional methods. Additionally, vessels are large objects and capturing a vessel from different viewpoints may provide entirely different visual appearances. To address these challenges, this paper proposes an Identity Oriented Re-identification network (IORnet) for the re-identification of vessels. This CNN-based approach incorporates the triplet loss method combined with a new loss function, which leads to improved vessel reidentification. Experimental results on our real-world evaluation dataset reveal that the proposed method achieves 81.5% and 91.2% on mAP and Rank1 scores, respectively. As an additional contribution, we also provide our annotated vessel reidentification dataset to the open public access.

AB - In order to perform a reliable vessel behavior analysis for maritime surveillance, re-identification of previously detected vessels, passing through new camera locations, is of vital importance. However, challenging outdoor conditions of the maritime environment heavily restrict the application of conventional methods. Additionally, vessels are large objects and capturing a vessel from different viewpoints may provide entirely different visual appearances. To address these challenges, this paper proposes an Identity Oriented Re-identification network (IORnet) for the re-identification of vessels. This CNN-based approach incorporates the triplet loss method combined with a new loss function, which leads to improved vessel reidentification. Experimental results on our real-world evaluation dataset reveal that the proposed method achieves 81.5% and 91.2% on mAP and Rank1 scores, respectively. As an additional contribution, we also provide our annotated vessel reidentification dataset to the open public access.

KW - CNNs

KW - Maritime dataset

KW - Maritime surveillance

KW - Vessel re-identification

UR - http://www.scopus.com/inward/record.url?scp=85066813627&partnerID=8YFLogxK

U2 - 10.1145/3323933.3324075

DO - 10.1145/3323933.3324075

M3 - Paper

SP - 93

EP - 97

ER -

Ghahremani A, Kong Y, Bondarev E, de With PHN. Re-identification of vessels with convolutional neural networks. 2019. Paper presented at 5th International Conference on Computer and Technology Applications, (ICCTA2019), Istanbul, Turkey. Available from, DOI: 10.1145/3323933.3324075