Toggle Main Menu Toggle Search

Open Access padlockePrints

A Comparison of Deep Neural Network Architectures in Aircraft Detection from SAR Imagery

Lookup NU author(s): Dr Jin XingORCiD, Dr Shuo LiORCiD, Professor Phil BlytheORCiD, Dr Yanghanzi ZhangORCiD, Simon Edwards


Full text for this publication is not currently held within this repository. Alternative links are provided below where available.


© 2022, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.Although numerous deep neural networks have been explored for aircraft detection using synthetic aperture radar (SAR) imagery, limited work has been conducted with their performance comparison, since different neural networks are designed and tested using different datasets and measured with different metrics. In this book chapter, we compare the performance of six popular deep neural networks for aircraft detection from SAR imagery, to verify their performance in tackling the scale heterogeneity, the background interference and the speckle noise challenges in the SAR-based aircraft detection. We choose SAR images acquired from three major airports in China as the testing datasets, due to the lack of ubiquitously agreed SAR benchmark dataset in aircraft detection. This comparison work does not only confirm the value of deep learning in aircraft detection but also highlights the advantages and disadvantages of these techniques, which paves the path for the design and development of workflow guidance in SAR-based aircraft detection using deep neural networks. It also serves as a baseline for future deep learning comparison in remote sensing data analytics, so as to facilitate the domain knowledge integration and design of innovative aircraft detection deep learning techniques.

Publication metadata

Author(s): Xing J, Luo R, Chen L, Wang J, Cai X, Li S, Blythe P, Zhang Y, Edwards S

Editor(s): Rysz M; Tsokas A; Dipple KM; Fair KL; Pardalos PM

Publication type: Book Chapter

Publication status: Published

Book Title: Synthetic Aperture Radar (SAR) Data Applications

Year: 2023

Volume: 199

Pages: 91-111

Print publication date: 19/01/2023

Online publication date: 19/01/2023

Acceptance date: 02/04/2022

Series Title: Springer Optimization and Its Applications

Publisher: Springer

Place Published: Cham


DOI: 10.1007/978-3-031-21225-3_5

Library holdings: Search Newcastle University Library for this item

ISBN: 9783031212246