CBAM-Yolov5: improved Yolov5 based on attention model for infrared ship detection

Lize Miao; Ning Li; Minglong Zhou; Huiyu Zhou

doi:10.1117/12.2631130

18 March 2022 CBAM-Yolov5: improved Yolov5 based on attention model for infrared ship detection

Lize Miao, Ning Li, Minglong Zhou, Huiyu Zhou

Proceedings Volume 12168, International Conference on Computer Graphics, Artificial Intelligence, and Data Processing (ICCAID 2021); 121682F (2022) https://doi.org/10.1117/12.2631130
Event: International Conference on Computer Graphics, Artificial Intelligence, and Data Processing (ICCAID 2021), 2021, Harbin, China

Abstract

Infrared image has the low contrast and resolution, and there are few available features for infrared small targets, so the quality of bounding box obtained by the target detection model is substandard. To solve the problems, a model named CBAM-Yolov5 is proposed. Firstly, the convolutional block attention module is added to enhance the feature extraction ability of the backbone network. Then a scale bias factor is designed to improve the regression effect on the bounding box of small targets, which will increase the weight of the small targets on the loss function. Through experimental verification, the mAP and recall rate of our model can be 93.3% and 90.4% on the infrared ship dataset, and compared with Yolov5, it has increased by 1.1% and 2.0%, respectively.

Citation Download Citation

Lize Miao, Ning Li, Minglong Zhou, and Huiyu Zhou "CBAM-Yolov5: improved Yolov5 based on attention model for infrared ship detection", Proc. SPIE 12168, International Conference on Computer Graphics, Artificial Intelligence, and Data Processing (ICCAID 2021), 121682F (18 March 2022); https://doi.org/10.1117/12.2631130

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