Object detection algorithm based on improved Yolov5

Hua Wang; Jiang Yin; Shuang Zhang; Daishuang Hou

doi:10.1117/12.2672682

28 April 2023 Object detection algorithm based on improved Yolov5

Hua Wang, Jiang Yin, Shuang Zhang, Daishuang Hou

Proceedings Volume 12610, Third International Conference on Artificial Intelligence and Computer Engineering (ICAICE 2022); 126104X (2023) https://doi.org/10.1117/12.2672682
Event: Third International Conference on Artificial Intelligence and Computer Engineering (ICAICE 2022), 2022, Wuhan, China

Abstract

A more accurate target detection model is proposed in this research based on Yolov5 target detection algorithm, aiming at its low regression accuracy to the target boundary box. Firstly, coordinate attention mechanism is added to the backbone network to improve the position information of the perceived target in the underlying feature information. Secondly, GIOU is replaced with EIOU to improve the convergence speed. Finally, the feature extraction network is replaced with BiFPN to more efficiently fuse different feature information. Using PASCAL VOC 2007 and 2012 datasets and redividing the training set and verification set, this algorithm is better than the original algorithm mAP@0.5 increased by 2.9%, mAP@0.5:0.95 increased by 1.4%.

Citation Download Citation

Hua Wang, Jiang Yin, Shuang Zhang, and Daishuang Hou "Object detection algorithm based on improved Yolov5", Proc. SPIE 12610, Third International Conference on Artificial Intelligence and Computer Engineering (ICAICE 2022), 126104X (28 April 2023); https://doi.org/10.1117/12.2672682

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