Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches

Unmanned aircraft systems or drones enable us to record or capture many scenes from the bird’s-eye view and they have been fast deployed to a wide range of practical domains, i.e., agriculture, aerial photography, fast delivery and surveillance. Object detection task is one of the core steps in unde...

Full description

Bibliographic Details
Main Authors: Khang Nguyen, Nhut T. Huynh, Phat C. Nguyen, Khanh-Duy Nguyen, Nguyen D. Vo, Tam V. Nguyen
Format: Article
Language:English
Published: MDPI AG 2020-03-01
Series:Electronics
Subjects:
SSD
Online Access:https://www.mdpi.com/2079-9292/9/4/583
id doaj-59e88af9fb5b40a59211bd3984003407
record_format Article
spelling doaj-59e88af9fb5b40a59211bd39840034072020-11-25T03:08:39ZengMDPI AGElectronics2079-92922020-03-01958358310.3390/electronics9040583Detecting Objects from Space: An Evaluation of Deep-Learning Modern ApproachesKhang Nguyen0Nhut T. Huynh1Phat C. Nguyen2Khanh-Duy Nguyen3Nguyen D. Vo4Tam V. Nguyen5Multimedia Communications Laboratory, University of Information Technology, Ho Chi Minh City 700000, VietnamDepartment of Software Engineering, University of Information Technology, Ho Chi Minh City 700000, VietnamDepartment of Software Engineering, University of Information Technology, Ho Chi Minh City 700000, VietnamMultimedia Communications Laboratory, University of Information Technology, Ho Chi Minh City 700000, VietnamMultimedia Communications Laboratory, University of Information Technology, Ho Chi Minh City 700000, VietnamDepartment of Computer Science, University of Dayton, Dayton, OH 45469, USAUnmanned aircraft systems or drones enable us to record or capture many scenes from the bird’s-eye view and they have been fast deployed to a wide range of practical domains, i.e., agriculture, aerial photography, fast delivery and surveillance. Object detection task is one of the core steps in understanding videos collected from the drones. However, this task is very challenging due to the unconstrained viewpoints and low resolution of captured videos. While deep-learning modern object detectors have recently achieved great success in general benchmarks, i.e., PASCAL-VOC and MS-COCO, the robustness of these detectors on aerial images captured by drones is not well studied. In this paper, we present an evaluation of state-of-the-art deep-learning detectors including Faster R-CNN (Faster Regional CNN), RFCN (Region-based Fully Convolutional Networks), SNIPER (Scale Normalization for Image Pyramids with Efficient Resampling), Single-Shot Detector (SSD), YOLO (You Only Look Once), RetinaNet, and CenterNet for the object detection in videos captured by drones. We conduct experiments on VisDrone2019 dataset which contains 96 videos with 39,988 annotated frames and provide insights into efficient object detectors for aerial images.https://www.mdpi.com/2079-9292/9/4/583object detectionVisDrone2019aerial imageryFaster R-CNNSSDRFCN
collection DOAJ
language English
format Article
sources DOAJ
author Khang Nguyen
Nhut T. Huynh
Phat C. Nguyen
Khanh-Duy Nguyen
Nguyen D. Vo
Tam V. Nguyen
spellingShingle Khang Nguyen
Nhut T. Huynh
Phat C. Nguyen
Khanh-Duy Nguyen
Nguyen D. Vo
Tam V. Nguyen
Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches
Electronics
object detection
VisDrone2019
aerial imagery
Faster R-CNN
SSD
RFCN
author_facet Khang Nguyen
Nhut T. Huynh
Phat C. Nguyen
Khanh-Duy Nguyen
Nguyen D. Vo
Tam V. Nguyen
author_sort Khang Nguyen
title Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches
title_short Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches
title_full Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches
title_fullStr Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches
title_full_unstemmed Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches
title_sort detecting objects from space: an evaluation of deep-learning modern approaches
publisher MDPI AG
series Electronics
issn 2079-9292
publishDate 2020-03-01
description Unmanned aircraft systems or drones enable us to record or capture many scenes from the bird’s-eye view and they have been fast deployed to a wide range of practical domains, i.e., agriculture, aerial photography, fast delivery and surveillance. Object detection task is one of the core steps in understanding videos collected from the drones. However, this task is very challenging due to the unconstrained viewpoints and low resolution of captured videos. While deep-learning modern object detectors have recently achieved great success in general benchmarks, i.e., PASCAL-VOC and MS-COCO, the robustness of these detectors on aerial images captured by drones is not well studied. In this paper, we present an evaluation of state-of-the-art deep-learning detectors including Faster R-CNN (Faster Regional CNN), RFCN (Region-based Fully Convolutional Networks), SNIPER (Scale Normalization for Image Pyramids with Efficient Resampling), Single-Shot Detector (SSD), YOLO (You Only Look Once), RetinaNet, and CenterNet for the object detection in videos captured by drones. We conduct experiments on VisDrone2019 dataset which contains 96 videos with 39,988 annotated frames and provide insights into efficient object detectors for aerial images.
topic object detection
VisDrone2019
aerial imagery
Faster R-CNN
SSD
RFCN
url https://www.mdpi.com/2079-9292/9/4/583
work_keys_str_mv AT khangnguyen detectingobjectsfromspaceanevaluationofdeeplearningmodernapproaches
AT nhutthuynh detectingobjectsfromspaceanevaluationofdeeplearningmodernapproaches
AT phatcnguyen detectingobjectsfromspaceanevaluationofdeeplearningmodernapproaches
AT khanhduynguyen detectingobjectsfromspaceanevaluationofdeeplearningmodernapproaches
AT nguyendvo detectingobjectsfromspaceanevaluationofdeeplearningmodernapproaches
AT tamvnguyen detectingobjectsfromspaceanevaluationofdeeplearningmodernapproaches
_version_ 1724665031147126784