DocumentCode
133704
Title
Ship detection for automating navigational watch
Author
Matsumoto, Yohei
Author_Institution
Dept. of Marine Eng., Tokyo Univ. of Marine Sci. & Technol., Tokyo, Japan
fYear
2014
fDate
3-7 Aug. 2014
Firstpage
103
Lastpage
108
Abstract
Automatic navigational watch system is currently under development. Because the system should detect ships over a miles away from own ship, the depth of the field is extremely deeper than that of ordinal computer vision applications. This paper shows the results applying typical HOG (Histograms of Oriented Gradients)-SVM (Support Vector Machine) based sliding window object detector to the navigational images, and discusses the difficulties and their possible remedies.
Keywords
computer vision; computerised navigation; marine navigation; object detection; ships; support vector machines; HOG-SVM; automatic navigational watch system; computer vision application; histograms of oriented gradient; ship detection; sliding window object detector; support vector machine; Manuals; Support vector machine classification; Training; HOG; Navigation; SVM; Ship;
fLanguage
English
Publisher
ieee
Conference_Titel
World Automation Congress (WAC), 2014
Conference_Location
Waikoloa, HI
Type
conf
DOI
10.1109/WAC.2014.6935702
Filename
6935702
Link To Document