Vehicle Detection in Aerial Images

Michael Ying Yang, Wentong Liao, Xinbo Li, Yanpeng Cao, Bodo Rosenhahn

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

The detection of vehicles in aerial images is widely applied in many applications. Comparing with object detection in the ground view images, vehicle detection in aerial images remains a challenging problem because of small vehicle size and the complex background. In this paper, we propose a novel double focal loss convolutional neural network (DFL-CNN) framework. In the proposed framework, the skip connection is used in the CNN structure to enhance the feature learning. Also, the focal loss function is used to substitute for conventional cross entropy loss function in both of the region proposal network (RPN) and the final classifier. We further introduce the first large-scale vehicle detection dataset ITCVD with ground truth annotations for all the vehicles in the scene. We demonstrate the performance of our model on the existing benchmark German Aerospace Center (DLR) 3K dataset as well as the ITCVD dataset. The experimental results show that our DFL-CNN outperforms the baselines on vehicle detection.
Original languageEnglish
Pages (from-to)297-304
Number of pages8
JournalPhotogrammetric engineering and remote sensing : PE&RS
Volume85
Issue number4
DOIs
Publication statusPublished - 1 Apr 2019

Fingerprint

Antennas
Neural networks
Antenna grounds
entropy
detection
vehicle
Classifiers
Entropy
learning
loss

Keywords

  • ITC-ISI-JOURNAL-ARTICLE

Cite this

Yang, Michael Ying ; Liao, Wentong ; Li, Xinbo ; Cao, Yanpeng ; Rosenhahn, Bodo. / Vehicle Detection in Aerial Images. In: Photogrammetric engineering and remote sensing : PE&RS. 2019 ; Vol. 85, No. 4. pp. 297-304.
@article{6f0986baa126444db8a7b34dd824ce09,
title = "Vehicle Detection in Aerial Images",
abstract = "The detection of vehicles in aerial images is widely applied in many applications. Comparing with object detection in the ground view images, vehicle detection in aerial images remains a challenging problem because of small vehicle size and the complex background. In this paper, we propose a novel double focal loss convolutional neural network (DFL-CNN) framework. In the proposed framework, the skip connection is used in the CNN structure to enhance the feature learning. Also, the focal loss function is used to substitute for conventional cross entropy loss function in both of the region proposal network (RPN) and the final classifier. We further introduce the first large-scale vehicle detection dataset ITCVD with ground truth annotations for all the vehicles in the scene. We demonstrate the performance of our model on the existing benchmark German Aerospace Center (DLR) 3K dataset as well as the ITCVD dataset. The experimental results show that our DFL-CNN outperforms the baselines on vehicle detection.",
keywords = "ITC-ISI-JOURNAL-ARTICLE",
author = "Yang, {Michael Ying} and Wentong Liao and Xinbo Li and Yanpeng Cao and Bodo Rosenhahn",
year = "2019",
month = "4",
day = "1",
doi = "10.14358/PERS.85.4.297",
language = "English",
volume = "85",
pages = "297--304",
journal = "Photogrammetric engineering and remote sensing : PE&RS",
issn = "0099-1112",
publisher = "American Society for Photogrammetry and Remote Sensing",
number = "4",

}

Vehicle Detection in Aerial Images. / Yang, Michael Ying; Liao, Wentong; Li, Xinbo; Cao, Yanpeng; Rosenhahn, Bodo.

In: Photogrammetric engineering and remote sensing : PE&RS, Vol. 85, No. 4, 01.04.2019, p. 297-304.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Vehicle Detection in Aerial Images

AU - Yang, Michael Ying

AU - Liao, Wentong

AU - Li, Xinbo

AU - Cao, Yanpeng

AU - Rosenhahn, Bodo

PY - 2019/4/1

Y1 - 2019/4/1

N2 - The detection of vehicles in aerial images is widely applied in many applications. Comparing with object detection in the ground view images, vehicle detection in aerial images remains a challenging problem because of small vehicle size and the complex background. In this paper, we propose a novel double focal loss convolutional neural network (DFL-CNN) framework. In the proposed framework, the skip connection is used in the CNN structure to enhance the feature learning. Also, the focal loss function is used to substitute for conventional cross entropy loss function in both of the region proposal network (RPN) and the final classifier. We further introduce the first large-scale vehicle detection dataset ITCVD with ground truth annotations for all the vehicles in the scene. We demonstrate the performance of our model on the existing benchmark German Aerospace Center (DLR) 3K dataset as well as the ITCVD dataset. The experimental results show that our DFL-CNN outperforms the baselines on vehicle detection.

AB - The detection of vehicles in aerial images is widely applied in many applications. Comparing with object detection in the ground view images, vehicle detection in aerial images remains a challenging problem because of small vehicle size and the complex background. In this paper, we propose a novel double focal loss convolutional neural network (DFL-CNN) framework. In the proposed framework, the skip connection is used in the CNN structure to enhance the feature learning. Also, the focal loss function is used to substitute for conventional cross entropy loss function in both of the region proposal network (RPN) and the final classifier. We further introduce the first large-scale vehicle detection dataset ITCVD with ground truth annotations for all the vehicles in the scene. We demonstrate the performance of our model on the existing benchmark German Aerospace Center (DLR) 3K dataset as well as the ITCVD dataset. The experimental results show that our DFL-CNN outperforms the baselines on vehicle detection.

KW - ITC-ISI-JOURNAL-ARTICLE

UR - https://ezproxy2.utwente.nl/login?url=https://library.itc.utwente.nl/login/2019/isi/yang_veh.pdf

UR - https://ezproxy2.utwente.nl/login?url=https://doi.org/10.14358/PERS.85.4.297

U2 - 10.14358/PERS.85.4.297

DO - 10.14358/PERS.85.4.297

M3 - Article

VL - 85

SP - 297

EP - 304

JO - Photogrammetric engineering and remote sensing : PE&RS

JF - Photogrammetric engineering and remote sensing : PE&RS

SN - 0099-1112

IS - 4

ER -