Forward-Looking Sonar Patch Matching: Modern CNNs, Ensembling, and Uncertainty

被引:0
|
作者
Mallick, Arka [1 ]
Ploger, Paul [1 ]
Valdenegro-Toro, Matias [2 ]
机构
[1] Hsch Bonn Rhein Sieg, Dept Comp Sci, D-53757 St Augustin, Germany
[2] German Res Ctr Artificial Intelligence, D-28359 Bremen, Germany
关键词
IMAGE;
D O I
暂无
中图分类号
U6 [水路运输]; P75 [海洋工程];
学科分类号
0814 ; 081505 ; 0824 ; 082401 ;
摘要
Application of underwater robots are on the rise, most of them are dependent on sonar for underwater vision, but the lack of strong perception capabilities limits them in this task. An important issue in sonar perception is matching image patches, which can enable other techniques like localization, change detection, and mapping. There is a rich literature for this problem in color images, but for acoustic images, it is lacking, due to the physics that produce these images. In this paper we improve on our previous results for this problem (Valdenegro-Toro et al, 2017), instead of modeling features manually, a Convolutional Neural Network (CNN) learns a similarity function and predicts if two input sonar images are similar or not. With the objective of improving the sonar image matching problem further, three state of the art CNN architectures are evaluated on the Marine Debris dataset, namely DenseNet, and VGG, with a siamese or two-channel architecture, and contrastive loss. To ensure a fair evaluation of each network, thorough hyper-parameter optimization is executed. We find that the best performing models are DenseNet Two-Channel network with 0.955 AUC, VGG-Siamese with contrastive loss at 0.949 AUC and DenseNet Siamese with 0.921 AUC. By ensembling the top performing DenseNet two-channel and DenseNet-Siamese models overall highest prediction accuracy obtained is 0.978 AUC, showing a large improvement over the 0.91 AUC in the state of the art.
引用
收藏
页数:7
相关论文
共 50 条
  • [1] A comparison of forward-looking sonar design alternatives
    Loggins, CD
    [J]. OCEANS 2001 MTS/IEEE: AN OCEAN ODYSSEY, VOLS 1-4, CONFERENCE PROCEEDINGS, 2001, : 1536 - 1545
  • [2] Forward-Looking Sonar Image Stitching Based on Midline Template Matching in Polar Image
    Liu, Hong
    Ye, Xiufen
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2024, 62 : 1 - 10
  • [3] Sensors for a forward-looking high resolution AUV sonar
    Nussbaum, F
    Stevens, GT
    Kelly, JG
    [J]. PROCEEDINGS OF THE 1996 SYMPOSIUM ON AUTONOMOUS UNDERWATER VEHICLE TECHNOLOGY, 1996, : 141 - 145
  • [4] Forward-looking Sonar Image Mosaicking by Feature Tracking
    Song, Sanming
    Herrmann, J. Michael
    Liu, Kaizhou
    Li, Shuo
    Feng, Xisheng
    [J]. 2016 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND BIOMIMETICS (ROBIO), 2016, : 1613 - 1618
  • [5] A Combinatorial Registration Method for Forward-Looking Sonar Image
    Li, Bufang
    Yan, Weisheng
    Li, Huiping
    [J]. IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, 2024, 20 (02) : 2682 - 2691
  • [6] Advanced Forward-Looking Sonar and Imagery Data Processing
    Berdnikova, J.
    Kozevnikov, V.
    Zamarajev, J.
    Raja, A.
    [J]. ELEKTRONIKA IR ELEKTROTECHNIKA, 2010, (04) : 55 - 58
  • [7] Location and Mapping for AUV Using a Forward-Looking Sonar
    Zhang, Tiedong
    Pang, Yongjie
    Jiang, Dapeng
    Zeng, Wenjing
    [J]. 2008 7TH WORLD CONGRESS ON INTELLIGENT CONTROL AND AUTOMATION, VOLS 1-23, 2008, : 9083 - 9088
  • [8] Pricing and Matching with Forward-Looking Buyers and Sellers
    Chen, Yiwei
    Hu, Ming
    [J]. M&SOM-MANUFACTURING & SERVICE OPERATIONS MANAGEMENT, 2020, 22 (04) : 717 - 734
  • [9] Extreme Uncertainty and Forward-looking Disclosure Properties
    Krause, Julia
    Sellhorn, Thorsten
    Ahmed, Kamran
    [J]. ABACUS-A JOURNAL OF ACCOUNTING FINANCE AND BUSINESS STUDIES, 2017, 53 (02): : 240 - 272
  • [10] The forward-looking competitive firm under uncertainty
    Lence, SH
    Hayes, DJ
    [J]. AMERICAN JOURNAL OF AGRICULTURAL ECONOMICS, 1998, 80 (02) : 303 - 312