A lightweight method for small object detection in aerial imagery on UAV edge devices

XU He; REN Zhiyang; JI Yimu; LI Peng

doi:null

您当前的位置：

首页 >

文章列表页 >

A lightweight method for small object detection in aerial imagery on UAV edge devices

更新时间：2026-06-15

- A lightweight method for small object detection in aerial imagery on UAV edge devices
- Chinese Journal on Internet of Things (2026)
- 作者机构：
  
  1.南京邮电大学计算机学院，江苏南京 210023
  2.南京邮电大学江苏省高性能计算与智能处理工程研究中心，江苏南京 210023
- 作者简介：
- 基金信息：
  
  Postgraduate Research & Practice Innovation Program of Jiangsu Province(SJCX26)
- DOI：
  CLC： TP391.4
- Received：17 May 2026，
  
  Revised：2026-06-15，
  
  Accepted：15 June 2026，
- 稿件说明：
移动端阅览
XU He, REN Zhiyang, JI Yimu, et al. A lightweight method for small object detection in aerial imagery on UAV edge devices[J/OL]. Chinese Journal on Internet of Things, 2026.
DOI：

XU He, REN Zhiyang, JI Yimu, et al. A lightweight method for small object detection in aerial imagery on UAV edge devices[J/OL]. Chinese Journal on Internet of Things, 2026. DOI：

摘要

针对无人机航拍图像中小目标占比高、密集遮挡严重、浅层细节易随下采样衰减，以及机载边缘设备资源受限等问题，提出一种面向无人机边缘端的轻量化航拍小目标检测方法。该方法以 YOLO11s 为基线，将原始 P3/P4/P5 检测结构重构为 P2/P3/P4 结构，使高分辨率浅层特征直接参与微小目标预测；同时移除对无人机小目标场景收益有限的 P5 分支，并将高层语义建模模块前移至 P4 层，以降低冗余参数开销。进一步设计选择性引导模块、P2 引导的跨层再注入机制和自适应融合模块，以增强浅层细节表达并缓解多源特征融合冲突。实验结果表明，在 VisDrone 验证集上，该方法较 YOLO11s 基线模型的平均精度均值（mAP

mean average precision）@0.5和mAP@0.5:0.95分别提升 4.40 和 3.22 个百分点，参数量由 9.46M 降至 3.82M；在Jetson Orin NX 16 GB上采用TensorRT 32位浮点精度部署时，推理速度达到 52.63帧/s。结果表明，该方法在提升无人机小目标检测精度的同时，兼顾了模型规模和边缘端实时推理能力。

Abstract

To address the problems of high small-object proportion

severe dense occlusion

shallow-detail degradation during downsampling

and limited onboard edge resources in aerial images captured by unmanned aerial vehicles (UAVs)

a lightweight small-object detection method for UAV edge devices was proposed. YOLO11s was adopted as the baseline. The original P3/P4/P5 detection structure was reconstructed into a P2/P3/P4 structure

so that high-resolution shallow features could directly participate in tiny-object prediction. Meanwhile

the P5 branch with limited benefit for UAV small-object scenes was removed

and the high-level semantic modeling module was moved to the P4 layer to reduce redundant parameter overhead. A selective guidance block

a P2-guided cross-layer re-injection mechanism

and an adaptive fusion module were further designed to enhance shallow details and alleviate multi-source feature fusion conflicts. Experimental results on the VisDrone validation set showed that mAP (mean average precision)@0.5 and mAP@0.5:0.95 were improved by 4.40 and 3.22 percentage points

respectively

compared with the YOLO11s baseline

while the number of parameters was reduced from 9.46M to 3.82M. When deployed with TensorRT using 32-bit floating-point precision (FP32) on a Jetson Orin NX 16 GB platform

an inference speed of 52.63 frames per second (FPS) was achieved. The results indicate that the proposed method improves detection accuracy while maintaining a compact model size and real-time edge inference capability.

关键词

Keywords

references

杨永刚，姜文韬，高志云 . 低空无人机实时目标检测算法 [J ] . 航空学报， 2025 ， 46 ( 16 ): 331619 .

YANG Y G , JIANG W T , GAO Z Y . Real-time object detection algorithm for low-altitude UAVs [J ] . Acta Aeronautica et Astronautica Sinica , 2025 , 46 ( 16 ): 331619 .

吴一全，童康 . 基于深度学习的无人机航拍图像小目标检测研究进展 [J ] . 航空学报， 2025 ， 46 ( 3 ): 30848 .

WU Y Q , TONG K . Research progress on small object detection in UAV aerial images based on deep learning [J ] . Acta Aeronautica et Astronautica Sinica , 2025 , 46 ( 3 ): 30848 .

高广帅，尚云琦，董燕 . 光学遥感图像小目标检测深度学习算法综述 [J ] . 中国图象图形学报， 2025 ， 30 ( 11 ): 3479 - 3505 .

GAO G S , SHANG Y Q , DONG Y . Review of deep learning algorithms for small object detection in optical remote sensing images [J ] . Journal of Image and Graphics , 2025 , 30 ( 11 ): 3479 - 3505 .

张琴，郭为安 . 深度学习小目标检测算法综述 [J ] . 计算机应用研究， 2025 ， 42 ( 10 ): 2893 - 2904 .

ZHANG Q , GUO W A . Survey of deep learning-based small object detection algorithms [J ] . Application Research of Computers , 2025 , 42 ( 10 ): 2893 - 2904 .

徐鹤，张恩俊，谭萍 . AGRF-Net：基于边界感知与自适应门控残差融合的RGB-D语义分割网络 [J/OL ] . 南京邮电大学学报（自然科学版）， 2026 : 1 - 11 .

XU H , ZHANG E J , TAN P . AGRF-Net: An RGB-D semantic segmentation network based on boundary-aware and adaptive gated residual fusion [J/OL ] . Journal of Nanjing University of Posts and Telecommunications (Natural Science Edition) , 2026 : 1 - 11 .

刘尚东，吴健，徐鹤，等 . 基于改进CycleGAN的低光域图像增强模型 [J/OL ] . 数据采集与处理， 2026 : 1 - 16 .

LIU S D , WU J , XU H , et al . A low-light domain image enhancement model based on improved CycleGAN [J/OL ] . Journal of Data Acquisition and Processing , 2026 : 1 - 16 .

ZHANG L , XIONG N , PAN X , et al . Improved object detection method utilizing YOLOv7-Tiny for unmanned aerial vehicle photographic imagery [J ] . Algorithms , 2023 , 16 ( 11 ): 520 .

NISA U , MOHD POZI M S , SAIP M A . A decade of research in small object detection: A comprehensive bibliometric analysis [J ] . International Journal of Data Science and Analytics , 2025 , 20 ( 8 ): 7331 - 7355 .

NIKOUEI M , BAROUTIAN B , NABAVI S , et al . Small object detection: A comprehensive survey on challenges, techniques and real-world applications [J ] . Intelligent Systems with Applications , 2025 , 27 : 200561 .

ZHAO Y , SUN H , WANG S . Small object detection in medium-low-resolution remote sensing images based on degradation reconstruction [J ] . Remote Sensing , 2024 , 16 ( 14 ): 2645 .

张杨，程正东，朱斌 . 基于改进Faster R-CNN的无人机小目标检测算法 [J ] . 量子电子学报， 2022 ， 39 ( 3 ): 354 - 363 .

ZHANG Y , CHENG Z D , ZHU B . UAV small object detection algorithm based on improved Faster R-CNN [J ] . Chinese Journal of Quantum Electronics , 2022 , 39 ( 3 ): 354 - 363 .

LIU M , WANG X , ZHOU A , et al . UAV-YOLO: Small object detection on unmanned aerial vehicle perspective [J ] . Sensors , 2020 , 20 ( 8 ): 2238 .

LI Y , LI Q , PAN J , et al . SOD-YOLO: Small-object-detection algorithm based on improved YOLOv8 for UAV images [J ] . Remote Sensing , 2024 , 16 ( 16 ): 3057 .

ZHAO X , ZHANG H , ZHANG W , et al . MSUD-YOLO: a novel multiscale small object detection model for UAV aerial images [J ] . Drones , 2025 , 9 ( 6 ): 429 .

LIU S , SHAO F , CHU W , et al . LCM-YOLO: a small object detection method for UAV imagery based on YOLOv5 [J ] . IET Image Processing , 2025 , 19 : e70051 .

胡惠娟，秦一锋，徐鹤，等 . 面向无人机航拍图像的YOLOv8目标检测改进算法 [J ] . 计算机科学， 2025 ， 52 ( 4 ): 202 - 211 .

HU H J , QIN Y F , XU H , et al . An improved YOLOv8 object detection algorithm for UAV aerial images [J ] . Computer Science , 2025 , 52 ( 4 ): 202 - 211 .

王胜科，任鹏飞，吕昕，等 . 基于中心点和双重注意力机制的无人机高分辨率图像小目标检测算法 [J ] . 应用科学学报， 2021 ， 39 ( 4 ): 650 - 659 .

WANG S K , REN P F , LV X , et al . Small object detection algorithm for UAV high-resolution images based on center points and dual attention mechanism [J ] . Journal of Applied Sciences , 2021 , 39 ( 4 ): 650 - 659 .

张伟，庄幸涛，王雪力，等 . DS-YOLO：一种部署在无人机终端上的小目标实时检测算法 [J ] . 南京邮电大学学报（自然科学版）， 2021 ， 41 ( 1 ): 86 - 98 .

ZHANG W , ZHUANG X T , WANG X L , et al . DS-YOLO: A real-time small object detection algorithm deployed on UAV terminals [J ] . Journal of Nanjing University of Posts and Telecommunications (Natural Science Edition) , 2021 , 41 ( 1 ): 86 - 98 .

张寅，朱桂熠，施天俊，等 . 基于特征融合与注意力的遥感图像小目标检测 [J ] . 光学学报， 2022 ， 42 ( 24 ): 2415001 .

ZHANG Y , ZHU G Y , SHI T J , et al . Small object detection in remote sensing images based on feature fusion and attention [J ] . Acta Optica Sinica , 2022 , 42 ( 24 ): 2415001 .

HUA W , CHEN Q . A survey of small object detection based on deep learning in aerial images [J ] . Artificial Intelligence Review , 2025 , 58 : 162 .

RABBI J , RAY N , SCHUBERT M , et al . Small-object detection in remote sensing images with end-to-end edge-enhanced GAN and object detector network [J ] . Remote Sensing , 2020 , 12 ( 9 ): 1432 .

MAHAUR B , MISHRA K K . Small-object detection based on YOLOv5 in autonomous driving systems [J ] . Pattern Recognition Letters , 2023 , 168 : 115 - 122 .

YUAN Z , GONG J , GUO B , et al . Small object detection in UAV remote sensing images based on intra-group multi-scale fusion attention and adaptive weighted feature fusion mechanism [J ] . Remote Sensing , 2024 , 16 ( 22 ): 4265 .

LEE Y W , KIM B G . Attention-based scale sequence network for small object detection [J ] . Heliyon , 2024 , 10 ( 12 ): e32931 .

郭磊，王邱龙，薛伟，等 . 基于改进 YOLOv5 的小目标检测算法 [J ] . 电子科技大学学报， 2022 ， 51 ( 2 ): 251 - 258 .

GUO L , WANG Q L , XUE W , et al . Small object detection algorithm based on improved YOLOv5 [J ] . Journal of University of Electronic Science and Technology of China , 2022 , 51 ( 2 ): 251 - 258 .

杨海舟，李丹 . 基于改进SPPNet的YOLOv4目标检测 [J ] . 电子制作， 2021 ( 22 ): 52 - 54 .

YANG H Z , LI D . YOLOv4 object detection based on improved SPPNet [J ] . Practical Electronics , 2021 ( 22 ): 52 - 54 .

常青，韩文，王清华，等 . 改进YOLO轻量化网络的行人检测算法 [J ] . 光学技术， 2022 ， 48 ( 1 ): 80 - 85 .

CHANG Q , HAN W , WANG Q H , et al . Pedestrian detection algorithm based on improved lightweight YOLO network [J ] . Optical Technique , 2022 , 48 ( 1 ): 80 - 85 .

MA Z , ZHOU L , WU D , et al . A small object detection method with context information for high altitude images [J ] . Pattern Recognition Letters , 2025 , 188 : 22 - 28 .

MA Y C , MA X , HAO T R , et al . Knowledge distillation via hierarchical matching for small object detection [J ] . Journal of Computer Science and Technology , 2024 , 39 ( 4 ): 798 - 810 .

WANG A , CHEN H , LIU L , et al . YOLOv10: real-time end-to-end object detection [C ] // Advances in Neural Information Processing Systems . Red Hook : Curran Associates, Inc. , 2024 , 37 : 107984 - 108011 .

TIAN Y , YE Q , DOERMANN D . YOLOv12: attention-centric real-time object detectors [C ] // Advances in Neural Information Processing Systems . Red Hook : Curran Associates, Inc. , 2025 , 38 : 78433 - 78457 .

ZHAO Y , LV W , XU S , et al . DETRs beat YOLOs on real-time object detection [C ] // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE Press , 2024 : 16965 - 16974 .

TANG S , ZHANG S , FANG Y . HIC-YOLOv5: improved YOLOv5 for small object detection [C ] // 2024 IEEE International Conference on Robotics and Automation (ICRA) . Piscataway : IEEE Press , 2024 : 6614 - 6619 .

ZHANG Z X . Drone-YOLO: an efficient neural network method for target detection in drone images [J ] . Drones , 2023 , 7 ( 8 ): 526 .

HAN Y , WANG C , LUO H , et al . LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient design [J ] . Scientific Reports , 2025 , 15 : 22627 .

WANG J , HUANG Z , DONG Y , et al . LE-YOLO: a lightweight and effective YOLO model for remote sensing image object detection [J ] . Journal of Real-Time Image Processing , 2026 , 23 : 27 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

An edge-cloud collaborative model evolution and communication paradigm in Internet of low-altitude UAV

Edge intelligence empowered internet of vehicles: concept, framework, issues, implementation, and prospect

Cooperative inference analysis based on DNN convolutional kernel partitioning

Research on power efficient autonomous UAV navigation algorithm: an edge intelligence driven approach

Semantic communication networking for the intelligence of everything

Related Author

ZHANG Hongliang

ZHANG Shuhang

YU Xinbo

Yue CAO

Hai LIN

Yongdong ZHU

Kai JIANG

Huan ZHOU

Related Institution

School of Electronics Engineering and Computer Science, Peking University

Pengcheng Laboratory

School of Electronics, Peking University

Wuhan University

Huali SmartWays Technology Co., Ltd.

⁰