also be improved for detecting boats during low opti-
cal lighting conditions by training them with data cap-
tured in low optical lighting. The position estimator
could be improved by creating a better model for cor-
rection the panoramic distortion along with a better
calibration of the setup. A sensor constantly monitor-
ing the water level could also be implemented to more
precisely determine the cameras height above the wa-
ter. The position estimator should ideally be tested
using more accurate ground truth data. A tracking al-
gorithm could also be implemented for the purpose of
tracking the detected boats in the images. This would
ease the needed computations since the object detec-
tor would not need to be run for each frame. This
tracker could also provide more information such as
the path of certain boats and their velocity. An ad-
ditional advantage of tracking would be the ability to
automatically pan and tilt the camera to follow a spe-
cific boat. Classifying detected boats would be ben-
eficial in order to gain further statistical data about
the boats entering and leaving ports. This could be
done, provided enough data, by retraining both detec-
tion models to detect particular types of boats such as
sailboats, motorboats, and tankers.
REFERENCES
Agarwal, A., Jawahar, C. V., and Narayanan, P. J. (2005). A
survey of planar homography estimation techniques.
Technical report.
Akiyama, T., Kobayashi, Y., Kishigami, J., and Muto, K.
(2018). Cnn-based boat detection model for alert
system using surveillance video camera. In 2018
IEEE 7th Global Conference on Consumer Electron-
ics (GCCE), pages 669–670.
Branch, A. E. (1986). Port traffic control, pages 124–143.
Springer Netherlands, Dordrecht.
Chen, K., Pang, J., Wang, J., Xiong, Y., Li, X., Sun, S.,
Feng, W., Liu, Z., Shi, J., Ouyang, W., Loy, C. C.,
and Lin, D. (2019). Hybrid task cascade for instance
segmentation.
Council of the European Union (2008). Maritime surveil-
lance - overview of ongoing activities. Technical re-
port, European Union, Brussels.
Dalal, N. and Triggs, B. (2005). Histograms of oriented gra-
dients for human detection. In 2005 IEEE Computer
Society Conference on Computer Vision and Pattern
Recognition (CVPR’05), volume 1, pages 886–893
vol. 1.
Drawil, N., Amar, H., and Basir, O. (2013). Gps localization
accuracy classification: A context-based approach. In-
telligent Transportation Systems, IEEE Transactions
on, 14:262–273.
Gade, R. and Moeslund, T. (2014). Thermal cameras and
applications: A survey. Machine Vision and Applica-
tions, 25:245–262.
Girshick, R., Donahue, J., Darrell, T., and Malik, J. (2013).
Rich feature hierarchies for accurate object detection
and semantic segmentation.
Goshtasby, A. (1986). Piecewise linear mapping functions
for image registration. Pattern Recognition, 19(6):459
– 466.
Goshtasby, A. (1988). Image registration by local ap-
proximation methods. Image and Vision Computing,
6(4):255 – 261.
Hikvision (2019). Ds-2td4166-25(50). https://
www.hikvision.com/en/Products/Thermal-Camera/
Network-Speed-Dome/640512-Series/DS-2TD4166-
25(50).
Huang, J., Rathod, V., Chen, S., Zhu, M., Korattikara, A.,
Fathi, A., Fischer, I., Wojna, Z., Yang, S., Guadar-
rama, S., and Murphy, K. (2017). Speed/accuracy
trade-offs for modern convolutional object detectors.
Hughes, J. F., van Dam, A., McGuire, M., Sklar, D. F.,
Foley, J. D., Feiner, S. K., and Akeley, K. (2013).
Computer graphics: principles and practice (3rd ed.).
Addison-Wesley Professional, Boston, MA, USA.
IMO (2001). Guidelines for the onboard operational use
of shipborne automatic identification systems (ais).
page 6.
John, V., Tsuchizawa, S., Liu, Z., and Mita, S. (2016). Fu-
sion of thermal and visible cameras for the application
of pedestrian detection. Signal Image and Video Pro-
cessing.
Lin, T., Maire, M., Belongie, S. J., Bourdev, L. D., Girshick,
R. B., Hays, J., Perona, P., Ramanan, D., Doll
´
ar, P.,
and Zitnick, C. L. (2014). Microsoft COCO: common
objects in context. CoRR, abs/1405.0312.
Liu, L., Ouyang, W., Wang, X., Fieguth, P., Chen, J., Liu,
X., and Pietik
¨
ainen, M. (2018). Deep Learning for
Generic Object Detection: A Survey. arXiv e-prints,
page arXiv:1809.02165.
Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S.,
Fu, C.-Y., and Berg, A. C. (2016). Ssd: Single shot
multibox detector. Lecture Notes in Computer Sci-
ence, page 21–37.
Luhmann, T. (2008). A historical review on panorama
photogrammetry. International Archives of the Pho-
togrammetry, Remote Sensing and Spatial Informa-
tion Sciences, 34.
MarineTraffic (2019). Marinetraffic: Global ship track-
ing intelligence — ais marine traffic. https://www.
marinetraffic.com/.
Mohan, D. and Ram, D. A. R. (2015). A review on depth es-
timation for computer vision applications. volume 4.
Redmon, J., Divvala, S., Girshick, R., and Farhadi, A.
(2016). You only look once: Unified, real-time ob-
ject detection. pages 779–788.
Redmon, J. and Farhadi, A. (2017). Yolo9000: Better,
faster, stronger. 2017 IEEE Conference on Computer
Vision and Pattern Recognition (CVPR), pages 6517–
6525.
Redmon, J. and Farhadi, A. (2018). Yolov3: An incremental
improvement. CoRR, abs/1804.02767.