Object detection algorithm for unmanned surface vehicle using faster R-CNN

Heesu Kim, Evangelos Boulougouris, Sang-Hyun Kim

Research output: Contribution to conferencePaper

9 Downloads (Pure)

Abstract

The purpose of this research is development of vision-based object detection algorithm that recognizes a marine object, localizes the object on captured frames, and estimates the distance to the object. Faster R-CNN and stereo vision based depth estimation are combined for real-time marine object detection. The performance of this algorithm is verified by model ship detection test in towing tank. The test results showed that this algorithm is potentially applicable to real USV.
Original languageEnglish
Number of pages12
Publication statusPublished - 4 Dec 2018
EventWorld Maritime Technology Conference 2018 - Renaissance Shanghai Zhongshan, Part Hotel Shanghai, Shanghai, China
Duration: 4 Dec 20187 Dec 2018
Conference number: 6

Conference

ConferenceWorld Maritime Technology Conference 2018
Abbreviated titleWMTC2018
CountryChina
CityShanghai
Period4/12/187/12/18

    Fingerprint

Keywords

  • unmanned surface vehicle
  • vision-based object detection
  • aster region with convolutional neural network
  • depth estimation

Cite this

Kim, H., Boulougouris, E., & Kim, S-H. (2018). Object detection algorithm for unmanned surface vehicle using faster R-CNN. Paper presented at World Maritime Technology Conference 2018, Shanghai, China.