TY - GEN
T1 - Target localization in synthetic aperture sonar imagery using convolutional neural networks
AU - Berthomier, Thibaud
AU - Williams, David P.
AU - Dugelay, Samantha
N1 - Funding Information:
ACKNOWLEDGMENT This work was supported by the NATO STO Centre for Maritime Research and Experimentation (CMRE) and the French Government Defence. Sonar data used for the experiments were collected by the CMRE with funding provided by the NATO Allied Command Transformation (ACT).
Publisher Copyright:
© 2019 Marine Technology Society.
PY - 2019/10
Y1 - 2019/10
N2 - Automatic target recognition (ATR) in synthetic aperture sonar (SAS) is usually performed in two stages: object detection and target classification. The detector aims to localize all the potential targets whereas the classifier distinguishes between real targets and false alarms. The probability of detection at this first stage must be the highest as possible to ensure that targets are not missed. Unfortunately, this generally implies a significant false alarm rate. Therefore, the challenge of the second stage, classification, is to drastically reduce the number of false alarms while keeping the detected targets. Using a large database of SAS images, efficient CNN classifiers have been demonstrated for underwater target classification tasks. In this paper, we suggest applying a pretrained classification CNN for localizing targets in SAS images. In so doing, we show the feasibility of target detection and classification in one-step using CNNs.
AB - Automatic target recognition (ATR) in synthetic aperture sonar (SAS) is usually performed in two stages: object detection and target classification. The detector aims to localize all the potential targets whereas the classifier distinguishes between real targets and false alarms. The probability of detection at this first stage must be the highest as possible to ensure that targets are not missed. Unfortunately, this generally implies a significant false alarm rate. Therefore, the challenge of the second stage, classification, is to drastically reduce the number of false alarms while keeping the detected targets. Using a large database of SAS images, efficient CNN classifiers have been demonstrated for underwater target classification tasks. In this paper, we suggest applying a pretrained classification CNN for localizing targets in SAS images. In so doing, we show the feasibility of target detection and classification in one-step using CNNs.
UR - http://www.scopus.com/inward/record.url?scp=85079055248&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85079055248&partnerID=8YFLogxK
U2 - 10.23919/OCEANS40490.2019.8962774
DO - 10.23919/OCEANS40490.2019.8962774
M3 - Conference contribution
AN - SCOPUS:85079055248
T3 - OCEANS 2019 MTS/IEEE Seattle, OCEANS 2019
BT - OCEANS 2019 MTS/IEEE Seattle, OCEANS 2019
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2019 OCEANS MTS/IEEE Seattle, OCEANS 2019
Y2 - 27 October 2019 through 31 October 2019
ER -