Sea transportation is the cheapest way to ship goods all over the world. This type of transportation involves legal and illegal transportation. As an example of illegal activities, we can exemplify drug transportation, immigrants transportation etc. Compared to road transportation where all the routes are connected and monitored by cameras installed in different places, in sea transportation this cannot be possible. Other methods are used for ship monitoring such as satellites. Due to fact that satellite imagery requires lots of resources and high costs, alternative methods can be used. An alternative method in areas close to coastal sites are RGB cameras placed in specific zones suitable for anchoring. In this paper, we proposed to identify different type of images with boats taken in the water. To this goal, we trained a convolutional neural network (CNN) on a dataset comprising images with different types of boats such as bowrider, deck, dinghy, ponton, runabout, vela. Due to fact that at the time of image acquisition noise can be inserted in images, mathematical morphology is used to remove this noise. We obtained 98.27% accuracy using our proposed method for boat classification.
Boats Imagery Classification Using Deep Learning
Lect. Notes in Networks, Syst.
World Conference on Information Systems and Technologies ; 2024 ; Lodz, Poland March 26, 2024 - March 28, 2024
Good Practices and New Perspectives in Information Systems and Technologies ; Kapitel : 43 ; 456-465
2024-05-13
10 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Tema Archiv | 1992
|Deep Learning Waterline Detection for Low-Cost Autonomous Boats
Springer Verlag | 2018
|Classification of warships and torpedo boats
Engineering Index Backfile | 1898
|Online Contents | 1996
|Online Contents | 1995
|