UAV first view landmark localization with active reinforcement learning

Xinran Wang, Chao Li, Leijian Yu, Lirong Han, Xiaogang Deng, Erfu Yang, Peng Ren

Research output: Contribution to journalArticle

Abstract

We present an active reinforcement learning framework for unmanned aerial vehicle (UAV) first view landmark localization. We formulate the problem of landmark localization as that of a Markov decision process and introduce an active landmark-localization network (ALLNet) to address it. The aim of the ALLNet is to locate a bounding box that surrounds the landmark in a first view image sequence. To this end, it is trained in a reinforcement learning fashion. Specifically, it employs support vector machine (SVM) scores on the bounding box patches as rewards and learns the bounding box transformations as actions. Furthermore, each SVM score indicates whether or not the landmark is detected by the bounding box such that it enables the ALLNet to have the capability of judging whether the landmark leaves or re-enters a first view image. Therefore, the operation of the ALLNet is not only dominated by the reinforcement learning process but also supplemented by an active learning motivated manner. Once the landmark is considered to leave the first view image, the ALLNet stops operating until the SVM detects its re-entry to the view. The active reinforcement learning model enables training a robust ALLNet for landmark localization. The experimental results validate the effectiveness of the proposed model for UAV first view landmark localization.
LanguageEnglish
Pages549-555
Number of pages7
JournalPattern Recognition Letters
Volume125
Early online date18 Mar 2019
DOIs
Publication statusPublished - 1 Jul 2019

Fingerprint

Reinforcement learning
Unmanned aerial vehicles (UAV)
Support vector machines
Reentry

Keywords

  • reinforcement learning
  • first view landmark localization
  • unmanned aerial vehicle

Cite this

Wang, Xinran ; Li, Chao ; Yu, Leijian ; Han, Lirong ; Deng, Xiaogang ; Yang, Erfu ; Ren, Peng. / UAV first view landmark localization with active reinforcement learning. In: Pattern Recognition Letters. 2019 ; Vol. 125. pp. 549-555.
@article{5ae90d522d7a488581d248ae66835a98,
title = "UAV first view landmark localization with active reinforcement learning",
abstract = "We present an active reinforcement learning framework for unmanned aerial vehicle (UAV) first view landmark localization. We formulate the problem of landmark localization as that of a Markov decision process and introduce an active landmark-localization network (ALLNet) to address it. The aim of the ALLNet is to locate a bounding box that surrounds the landmark in a first view image sequence. To this end, it is trained in a reinforcement learning fashion. Specifically, it employs support vector machine (SVM) scores on the bounding box patches as rewards and learns the bounding box transformations as actions. Furthermore, each SVM score indicates whether or not the landmark is detected by the bounding box such that it enables the ALLNet to have the capability of judging whether the landmark leaves or re-enters a first view image. Therefore, the operation of the ALLNet is not only dominated by the reinforcement learning process but also supplemented by an active learning motivated manner. Once the landmark is considered to leave the first view image, the ALLNet stops operating until the SVM detects its re-entry to the view. The active reinforcement learning model enables training a robust ALLNet for landmark localization. The experimental results validate the effectiveness of the proposed model for UAV first view landmark localization.",
keywords = "reinforcement learning, first view landmark localization, unmanned aerial vehicle",
author = "Xinran Wang and Chao Li and Leijian Yu and Lirong Han and Xiaogang Deng and Erfu Yang and Peng Ren",
year = "2019",
month = "7",
day = "1",
doi = "10.1016/j.patrec.2019.03.011",
language = "English",
volume = "125",
pages = "549--555",
journal = "Pattern Recognition Letters",
issn = "0167-8655",

}

UAV first view landmark localization with active reinforcement learning. / Wang, Xinran; Li, Chao; Yu, Leijian; Han, Lirong; Deng, Xiaogang; Yang, Erfu; Ren, Peng.

In: Pattern Recognition Letters, Vol. 125, 01.07.2019, p. 549-555.

Research output: Contribution to journalArticle

TY - JOUR

T1 - UAV first view landmark localization with active reinforcement learning

AU - Wang, Xinran

AU - Li, Chao

AU - Yu, Leijian

AU - Han, Lirong

AU - Deng, Xiaogang

AU - Yang, Erfu

AU - Ren, Peng

PY - 2019/7/1

Y1 - 2019/7/1

N2 - We present an active reinforcement learning framework for unmanned aerial vehicle (UAV) first view landmark localization. We formulate the problem of landmark localization as that of a Markov decision process and introduce an active landmark-localization network (ALLNet) to address it. The aim of the ALLNet is to locate a bounding box that surrounds the landmark in a first view image sequence. To this end, it is trained in a reinforcement learning fashion. Specifically, it employs support vector machine (SVM) scores on the bounding box patches as rewards and learns the bounding box transformations as actions. Furthermore, each SVM score indicates whether or not the landmark is detected by the bounding box such that it enables the ALLNet to have the capability of judging whether the landmark leaves or re-enters a first view image. Therefore, the operation of the ALLNet is not only dominated by the reinforcement learning process but also supplemented by an active learning motivated manner. Once the landmark is considered to leave the first view image, the ALLNet stops operating until the SVM detects its re-entry to the view. The active reinforcement learning model enables training a robust ALLNet for landmark localization. The experimental results validate the effectiveness of the proposed model for UAV first view landmark localization.

AB - We present an active reinforcement learning framework for unmanned aerial vehicle (UAV) first view landmark localization. We formulate the problem of landmark localization as that of a Markov decision process and introduce an active landmark-localization network (ALLNet) to address it. The aim of the ALLNet is to locate a bounding box that surrounds the landmark in a first view image sequence. To this end, it is trained in a reinforcement learning fashion. Specifically, it employs support vector machine (SVM) scores on the bounding box patches as rewards and learns the bounding box transformations as actions. Furthermore, each SVM score indicates whether or not the landmark is detected by the bounding box such that it enables the ALLNet to have the capability of judging whether the landmark leaves or re-enters a first view image. Therefore, the operation of the ALLNet is not only dominated by the reinforcement learning process but also supplemented by an active learning motivated manner. Once the landmark is considered to leave the first view image, the ALLNet stops operating until the SVM detects its re-entry to the view. The active reinforcement learning model enables training a robust ALLNet for landmark localization. The experimental results validate the effectiveness of the proposed model for UAV first view landmark localization.

KW - reinforcement learning

KW - first view landmark localization

KW - unmanned aerial vehicle

U2 - 10.1016/j.patrec.2019.03.011

DO - 10.1016/j.patrec.2019.03.011

M3 - Article

VL - 125

SP - 549

EP - 555

JO - Pattern Recognition Letters

T2 - Pattern Recognition Letters

JF - Pattern Recognition Letters

SN - 0167-8655

ER -