Research on automatic detection method of infrared marine buoy based on CNN

Shousheng Liu; Zhigang Gai; Lin Cao; Hui Li; Ding Hu; Fengxiang Guo; Huimin Qiu

doi:10.1117/12.2553051

31 January 2020 Research on automatic detection method of infrared marine buoy based on CNN

Shousheng Liu, Zhigang Gai, Lin Cao, Hui Li, Ding Hu, Fengxiang Guo, Huimin Qiu

Proceedings Volume 11427, Second Target Recognition and Artificial Intelligence Summit Forum; 114273N (2020) https://doi.org/10.1117/12.2553051
Event: Second Target Recognition and Artificial Intelligence Summit Forum, 2019, Changchun, China

Abstract

Most of objects on marine surface are ships and buoys. There are a lot of researches on object recognition and localization for ship, but few studies are focus on marine buoy detection. If marine buoys could be detected in time, most of the damage caused by collision with ships or unmanned vessels will be avoided. We propose a method for automatic detection of marine buoy: by using infrared camera we obtain video stream data in which we select and annotate infrared image samples with buoys, by training those samples we fine-tune a CNN detection model based on MobleNet-SSD which is pre-trained on the Pascal VOC dataset. Any marine buoy could be recognized and localized by our detection method, whose average precision on our test set is up to 90%. We get real-time performance with FPS up to 30 on portable embedded devices like NVIDIA Jetson AGX Xavier and Intel Neural Compute Stick 2.

Citation Download Citation

Shousheng Liu, Zhigang Gai, Lin Cao, Hui Li, Ding Hu, Fengxiang Guo, and Huimin Qiu "Research on automatic detection method of infrared marine buoy based on CNN", Proc. SPIE 11427, Second Target Recognition and Artificial Intelligence Summit Forum, 114273N (31 January 2020); https://doi.org/10.1117/12.2553051

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available