2022
DOI: 10.3390/rs14225761
|View full text |Cite
|
Sign up to set email alerts
|

BoxPaste: An Effective Data Augmentation Method for SAR Ship Detection

Abstract: Data augmentation is a crucial technique for convolutional neural network (CNN)-based object detection. Thus, this work proposes BoxPaste, a simple but powerful data augmentation method appropriate for ship detection in Synthetic Aperture Radar (SAR) imagery. BoxPaste crops ship objects from one SAR image using bounding box annotations and pastes them on another SAR image to artificially increase the object density in each training image. Furthermore, we dive deep into the characteristics of the SAR ship detec… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
12
0

Year Published

2023
2023
2024
2024

Publication Types

Select...
7
2

Relationship

0
9

Authors

Journals

citations
Cited by 16 publications
(12 citation statements)
references
References 48 publications
0
12
0
Order By: Relevance
“…Accordingly, target velocity SAR phase refocusing presented in section 2 was inversely utilized in order to generate defocused SAR images. In contrast to conventional data augmentation methods in object detection performed to image chips [28], [29], inversion of refocusing function generates the SAR image with entire SLC dimension. Training data was augmented 20 times in quantity, defocused from m/s to m/s with 1 m/s spacing.…”
Section: A Quantitative Enhancement Of Training Data and Vessel Detec...mentioning
confidence: 99%
“…Accordingly, target velocity SAR phase refocusing presented in section 2 was inversely utilized in order to generate defocused SAR images. In contrast to conventional data augmentation methods in object detection performed to image chips [28], [29], inversion of refocusing function generates the SAR image with entire SLC dimension. Training data was augmented 20 times in quantity, defocused from m/s to m/s with 1 m/s spacing.…”
Section: A Quantitative Enhancement Of Training Data and Vessel Detec...mentioning
confidence: 99%
“…proposed a method that uses a discrete optimization strategy to model the data augmentation process. Suo et al 34 . proposed a simple but effective data augmentation strategy named BoxPaste, which crops the target from one image and pastes it onto another image.…”
Section: Related Workmentioning
confidence: 99%
“…Zoph et al 33 proposed a method that uses a discrete optimization strategy to model the data augmentation process. Suo et al 34 proposed a simple but effective data augmentation strategy named BoxPaste, which crops the target from one image and pastes it onto another image. Data-augmentation techniques have been employed to augment the number of small object instances, enhancing the detection performance of small objects to a certain extent.…”
Section: Small Object Detectionmentioning
confidence: 99%
“…Copy-paste [5] is an easy but effective data augmentation method, which achieved good performance in the instance segmentation task of natural images and can solve the problem of unbalance between classes to a certain extent. BoxPaste [6] has introduced the Copy-paste method into the SAR image data augmentation, which uses the bounding box of ships to extract ship chips, followed by pasting chips to other images. However, BoxPaste did not take into account the ship characteristics in remote sensing images and was only applied to increase the number of ships to improve ship detection accuracy.…”
Section: Introductionmentioning
confidence: 99%