Improved real time object detection method for remote sensing image based on YOLOv4

Luxuan Bian; Jue Wang; Bo Li; Ying Shi

doi:10.1117/12.2661016

16 December 2022 Improved real time object detection method for remote sensing image based on YOLOv4

Luxuan Bian, Jue Wang, Bo Li, Ying Shi

Proceedings Volume 12500, Fifth International Conference on Mechatronics and Computer Technology Engineering (MCTE 2022); 125003S (2022) https://doi.org/10.1117/12.2661016
Event: 5th International Conference on Mechatronics and Computer Technology Engineering (MCTE 2022), 2022, Chongqing, China

Abstract

The rapid development of computer vision has brought great convenience to remote sensing image processing technology, significantly improved the accuracy of target detection in remote sensing images and saved human effort. However, in the real-time remote sensing detection work, the complex and changeable weather conditions bring great difficulties to the detection. The existing models have shortcomings such as missed detection, slow detection speed, bloated and complex, and large number of parameters. In this paper, we propose a lightweight rapid deployment network framework based on ghostnet improved yolov4. First, ghostnet real-time lightweight extraction is introduced to speed up the search speed and save training costs; secondly, improve the Panet fusion network and add a self-attention mechanism at the up sampling point; use the K++ algorithm to optimize the size of the anchor of the remote sensing image detection target, and finally improve the loss function and add hyperparameters to suppress the imbalance between the target and background categories during training. Experiments show that compared with the original version of YOLOV4mAP in the data set, the mAP is increased by 2.2%, the calculation amount is reduced by 35%, and the FPS is increased by 78%. Therefore, G-YOLOv4 does have better performance in mAP, model size and FPS compared to YOLOV4's mAP, model size and FPS.

Citation Download Citation

Luxuan Bian, Jue Wang, Bo Li, and Ying Shi "Improved real time object detection method for remote sensing image based on YOLOv4", Proc. SPIE 12500, Fifth International Conference on Mechatronics and Computer Technology Engineering (MCTE 2022), 125003S (16 December 2022); https://doi.org/10.1117/12.2661016

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

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
6 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Remote sensing

Target detection

Convolution

Detection and tracking algorithms

Feature extraction

Image fusion

Image processing

Show All Keywords

Keywords/Phrases

Search In:

Publication Years