Lightweight UAV Small Target Detection Algorithm Based on Improved YOLOv8.

Saved in:

Bibliographic Details
Title:	Lightweight UAV Small Target Detection Algorithm Based on Improved YOLOv8.
Authors:	Zhang, Bo¹ 2365223458@qq.com, Tao, Ye² taibeijack@163.com, Cui, Wenhua³
Source:	IAENG International Journal of Computer Science. Jul2026, Vol. 53 Issue 7, p2778-2789. 12p.
Subjects:	Object recognition (Computer vision), Loss functions (Statistics)
Abstract:	Aiming at the problems of low proportion of small targets, complex background interference in UAV aerial photography scenes, and the large number of parameters of the traditional YOLOv8 algorithm which makes it difficult to adapt to UAV embedded hardware, a lightweight improved UAV small target detection algorithm named SDSE-YOLOv8 is proposed. Firstly, the Star_C2f module is introduced into the backbone network of YOLOv8, which adopts fully connected layers and incorporates Anchor-guided Axial Gated Attention (AGA) to accurately locate small targets. Secondly, the CSDM (Convolution-SGE-Depthwise Separable Conv Module) is inserted into the shallow layer to efficiently extract spatial finegrained information of small targets. Thirdly, the YOLOv8 network architecture is optimized by removing low-resolution feature layers and retaining high-resolution feature maps to improve gradient propagation efficiency. Finally, the EIoU loss function is used to replace the original CIoU, which improves the bounding box regression accuracy and accelerates model convergence. Experimental results on the VisDrone2019 dataset show that the improved SDSE-YOLOv8 algorithm reduces the number of parameters to 2.42M, increases mAP@0.5 to 41.6%, mAP@50:95 to 25.3%, and achieves an inference speed of 87FPS. It balances lightweight performance and detection accuracy, and can meet the real-time detection requirements of UAV embedded platforms. [ABSTRACT FROM AUTHOR]
	Copyright of IAENG International Journal of Computer Science is the property of International Association of Engineers (IAENG) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	Aiming at the problems of low proportion of small targets, complex background interference in UAV aerial photography scenes, and the large number of parameters of the traditional YOLOv8 algorithm which makes it difficult to adapt to UAV embedded hardware, a lightweight improved UAV small target detection algorithm named SDSE-YOLOv8 is proposed. Firstly, the Star_C2f module is introduced into the backbone network of YOLOv8, which adopts fully connected layers and incorporates Anchor-guided Axial Gated Attention (AGA) to accurately locate small targets. Secondly, the CSDM (Convolution-SGE-Depthwise Separable Conv Module) is inserted into the shallow layer to efficiently extract spatial finegrained information of small targets. Thirdly, the YOLOv8 network architecture is optimized by removing low-resolution feature layers and retaining high-resolution feature maps to improve gradient propagation efficiency. Finally, the EIoU loss function is used to replace the original CIoU, which improves the bounding box regression accuracy and accelerates model convergence. Experimental results on the VisDrone2019 dataset show that the improved SDSE-YOLOv8 algorithm reduces the number of parameters to 2.42M, increases mAP@0.5 to 41.6%, mAP@50:95 to 25.3%, and achieves an inference speed of 87FPS. It balances lightweight performance and detection accuracy, and can meet the real-time detection requirements of UAV embedded platforms. [ABSTRACT FROM AUTHOR]
ISSN:	1819656X