Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Sep 26;23(19):8093.
doi: 10.3390/s23198093.

Image-Based Ship Detection Using Deep Variational Information Bottleneck

Duc-Dat Ngo  1 , Van-Linh Vo  1 , Tri Nguyen  2 , Manh-Hung Nguyen  1 , My-Ha Le  1
Affiliations

Image-Based Ship Detection Using Deep Variational Information Bottleneck

Duc-Dat Ngo et al. Sensors (Basel). .

Abstract

Image-based ship detection is a critical function in maritime security. However, lacking high-quality training datasets makes it challenging to train a robust supervision deep learning model. Conventional methods use data augmentation to increase training samples. This approach is not robust because the data augmentation may not present a complex background or occlusion well. This paper proposes to use an information bottleneck and a reparameterization trick to address the challenge. The information bottleneck learns features that focus only on the object and eliminate all backgrounds. It helps to avoid background variance. In addition, the reparameterization introduces uncertainty during the training phase. It helps to learn more robust detectors. Comprehensive experiments show that the proposed method outperforms conventional methods on Seaship datasets, especially when the number of training samples is small. In addition, this paper discusses how to integrate the information bottleneck and the reparameterization into well-known object detection frameworks efficiently.

Keywords: information bottleneck; maritime security; ship detection.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
The contribution of the proposed method. (a) Performance on different scale datasets (b) A heat map by the proposed method (c) A heat map by a baseline method.
Figure 2
Figure 2
Network structure. (a) The overview of proposed VIB-based object detection. (b) The proposed VIB-based classification head.
Figure 3
Figure 3
Darknet structure.
Figure 4
Figure 4
PAFPN structure.
Figure 5
Figure 5
The mean average precision (mAP) on different αVIB. x-axis means the αKL parameter, and y-axis is the mean average precision (mAP) over all classes.
Figure 6
Figure 6
Heatmaps on the classifier head and the neck.
Figure 7
Figure 7
VIB on the mid of classification head.
Figure 8
Figure 8
VIB at the beginning of the decouple head.
Figure 9
Figure 9
The loss over a training process. The bold line is smooth values over iterations. The light line is the actual value in an iteration.
See this image and copyright information in PMC

References

    1. Szeto A., Pelot R. The use of long range identification and tracking (LRIT) for modelling the risk of ship-based oil spills; Proceedings of the AMOP Technical Seminar on Environmental Contamination and Response 2011; Banff, AB, Canada. 4–6 October 2011.
    1. Mao S., Tu E., Zhang G., Rachmawati L., Rajabally E., Huang G. An Automatic Identification System (AIS) Database for Maritime Trajectory Prediction and Data Mining. arXiv. 20161607.03306
    1. Paterniani G., Sgreccia D., Davoli A., Guerzoni G., Di Viesti P., Valenti A.C., Vitolo M., Vitetta G.M., Boriani G. Radar-Based Monitoring of Vital Signs: A Tutorial Overview. Proc. IEEE. 2023;111:277–317. doi: 10.1109/JPROC.2023.3244362. - DOI
    1. Zhou X., Gong W., Fu W., Du F. Application of deep learning in object detection; Proceedings of the 2017 IEEE/ACIS 16th International Conference on Computer and Information Science (ICIS); Wuhan, China. 24–26 May 2017; pp. 631–634. - DOI
    1. Girshick R., Donahue J., Darrell T., Malik J. Rich feature hierarchies for accurate object detection and semantic segmentation. arXiv. 20131311.2524

LinkOut - more resources