Lightweight in-harbor ship detection based on UAV aerial images

Chunlei Zhang; Jing Nie

doi:10.1117/12.2682664

1 June 2023 Lightweight in-harbor ship detection based on UAV aerial images

Chunlei Zhang, Jing Nie

Proceedings Volume 12710, International Conference on Remote Sensing, Surveying, and Mapping (RSSM 2023); 127100U (2023) https://doi.org/10.1117/12.2682664
Event: International Conference on Remote Sensing, Surveying, and Mapping (RSSM 2023), 2023, Changsha, China

Abstract

Using unmanned aerial vehicles (UAVs) to take aerial images and detect ships in harbors provides a new way of harbor monitoring. However, due to the limitation of computing resources and storage resources, ship detection on UAVs is very challenging, which puts a higher demand on model lightweighting. In this paper, we propose an efficient model lightweighting scheme based on knowledge distillation. We use two advanced large-scale models YOLOv7 and PP-YOLO as teacher models, and transfer the excellent detection ability of these two models to small-scale student models YOLOv7-tiny through knowledge distillation. This scheme not only greatly reduces the parameter scale and computation, but also retains the ship detection performance equivalent to that of large models.

Citation Download Citation

Chunlei Zhang and Jing Nie "Lightweight in-harbor ship detection based on UAV aerial images", Proc. SPIE 12710, International Conference on Remote Sensing, Surveying, and Mapping (RSSM 2023), 127100U (1 June 2023); https://doi.org/10.1117/12.2682664

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