ASOD-YOLO：基于YOLOv8n改进的航空小目标检测模型

计算机工程与科学 ›› 2026, Vol. 48 ›› Issue (1): 133-145.

ASOD-YOLO：基于YOLOv8n改进的航空小目标检测模型

曹利，徐慧英，谢刚，李毅，黄晓，陈昊，朱信忠

(1.贵州师范大学大数据与计算机科学学院，贵州贵阳 550025；2.浙江师范大学人工智能技术与应用研究所，浙江金华 321004;
3.浙江师范大学计算机科学与技术学院（人工智能学院），浙江金华 321004;
4.浙江师范大学教育学院（教师教育学院），浙江金华 321004;5.北京极智嘉科技股份有限公司,北京101318)

收稿日期:2024-03-18 修回日期:2024-08-20 出版日期:2026-01-25 发布日期:2026-01-25
基金资助:
国家自然科学基金（62376252）；浙江省自然科学基金(LZ22F030003)

ASOD-YOLO: An improved aerial small object detection model based on YOLOv8n

CAO Li1,2,XU Huiying2,3,XIE Gang1,LI Yi2,3,HUANG Xiao4,CHEN Hao2,3,ZHU Xinzhong2,3,5

(1.School of Big Data and Computer Science,Guizhou Normal University,Guiyang 550025;
2.Institute of Artificial Intelligence Technology and Application,Zhejiang Normal University,Jinhua 321004;
3.School of Computer Science and Technology(School of Artificial Intelligence),Zhejiang Normal University,Jinhua 321004;
4.College of Education(College of Teacher Education),Zhejiang Normal University,Jinhua 321004;
5.Beijing Geekplus Technology Co.,Ltd.,Beijing 101318,China)

Received:2024-03-18 Revised:2024-08-20 Online:2026-01-25 Published:2026-01-25

摘要/Abstract

摘要： 针对无人机航拍图像中目标分布不集中、尺寸变化大和特征不明显等特性造成的漏检、误检等问题，提出了一种基于YOLOv8n改进的航空小目标检测模型ASOD-YOLO。首先，重新设计了特征融合网络，将原有的特征金字塔结构的自上而下部分替换为低层的信息分发结构（Low-GD），在减少特征损失的同时增强不同尺度间信息融合的能力。其次，将原有的20×20的大目标检测头替换为160×160的小目标检测头，以增强对小目标的检测能力，并且改进了多尺度跨层连接，为检测头提供更加丰富的语义信息。同时，引入了快速傅里叶卷积（FFC）模块，以减少下采样后小目标信息丢失率，并增强全局上下文信息的提取能力。在航空小目标数据集VisDrone上的实验结果表明，ASOD-YOLO模型相较于基线YOLOv8n模型，在mAP@50指标上提升了4.1个百分点，在mAP@50:95指标上提升了2.3个百分点，单幅图像的处理时间仅有6.8 ms，表明了提出的ASOD-YOLO模型能有效完成航空小目标检测的任务。

关键词: 深度学习, 航空小目标检测；快速傅里叶卷积；特征融合

Abstract: To address issues such as missed detections and false detections caused by characteristics like scattered object distributions, significant size variations, and indistinct features in UAV images, this paper proposes an improved aerial small object detection model based on YOLOv8n, named ASOD-YOLO. Firstly, the feature fusion network is redesigned: the top-down part of the original feature pyramid structure is replaced with a low-level information distribution (Low-GD) structure. This modification reduces feature loss while enhancing the information fusion ability across different scales. Secondly, the original 20×20 large-object detection head is replaced with a 160×160 small-object detection head to improve the detection capability for small objects. Additionally, the multi-scale cross-layer connections are optimized to provide the detection head with richer semantic information. Meanwhile, a fast fourier convolution module(FFCBlock) is introduced to reduce the loss of small-object information after downsampling and enhance the ability to extract global contextual information. Experimental results on the VisDrone aerial small-object dataset show that, compared with the baseline YOLOv8n model, the ASOD-YOLO model achieves a 4.1 percentage points improvement in the mAP@50 metric and 2.3 percentage points improvement in the mAP@50:95 metric, with a single-image processing time of only 6.8 ms. These results demonstrate that the proposed ASOD-YOLO model can effectively accomplish the task of aerial small-object detection.

Key words: deep learning, aerial small object detection, fast Fourier convolution, feature fusion

曹利, 徐慧英, 谢刚, 李毅, 黄晓, 陈昊, 朱信忠. ASOD-YOLO：基于YOLOv8n改进的航空小目标检测模型[J]. 计算机工程与科学, 2026, 48(1): 133-145.

CAO Li1, 2, XU Huiying2, 3, XIE Gang1, LI Yi2, 3, HUANG Xiao4, CHEN Hao2, 3, ZHU Xinzhong2, 3, 5. ASOD-YOLO: An improved aerial small object detection model based on YOLOv8n[J]. Computer Engineering & Science, 2026, 48(1): 133-145.