LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient design

Abstract Small object detection in UAV aerial images is challenging due to low contrast, complex backgrounds, and limited computational resources. Traditional methods struggle with high miss detection rates and poor localization accuracy caused by information loss, weak cross-layer feature interacti...

Full description

Saved in:
Bibliographic Details
Main Authors: Yuqi Han, Chengcheng Wang, Hui Luo, Huihua Wang, Zaiqing Chen, Yuelong Xia, Lijun Yun
Format: Article
Language:English
Published: Nature Portfolio 2025-07-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-07021-6
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849335306693640192
author Yuqi Han
Chengcheng Wang
Hui Luo
Huihua Wang
Zaiqing Chen
Yuelong Xia
Lijun Yun
author_facet Yuqi Han
Chengcheng Wang
Hui Luo
Huihua Wang
Zaiqing Chen
Yuelong Xia
Lijun Yun
author_sort Yuqi Han
collection DOAJ
description Abstract Small object detection in UAV aerial images is challenging due to low contrast, complex backgrounds, and limited computational resources. Traditional methods struggle with high miss detection rates and poor localization accuracy caused by information loss, weak cross-layer feature interaction, and rigid detection heads. To address these issues, we propose LRDS-YOLO, a lightweight and efficient model tailored for UAV applications. The model incorporates a Light Adaptive-weight Downsampling (LAD) module to retain fine-grained small object features and reduce information loss. A Re-Calibration Feature Pyramid Network (Re-Calibration FPN) enhances multi-scale feature fusion using bidirectional interactions and resolution-aware hybrid attention. The SegNext Attention mechanism improves target focus while suppressing background noise, and the dynamic detection head (DyHead) optimizes multi-dimensional feature weighting for robust detection. Experiments show that LRDS-YOLO achieves 43.6% mAP50 on VisDrone2019, 11.4% higher than the baseline, with only 4.17M parameters and 24.1 GFLOPs, striking a balance between accuracy and efficiency. On the HIT-UAV infrared dataset, it reaches 84.5% mAP50, demonstrating strong generalization. With its lightweight design and high precision, LRDS-YOLO offers an effective real-time solution for UAV-based small object detection.
format Article
id doaj-art-c2c528748fe645c9a8229d9de2cb539d
institution Kabale University
issn 2045-2322
language English
publishDate 2025-07-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-c2c528748fe645c9a8229d9de2cb539d2025-08-20T03:45:19ZengNature PortfolioScientific Reports2045-23222025-07-0115111910.1038/s41598-025-07021-6LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient designYuqi Han0Chengcheng Wang1Hui Luo2Huihua Wang3Zaiqing Chen4Yuelong Xia5Lijun Yun6School of Information, Yunnan Normal UniversitySchool of Information, Yunnan Normal UniversityKunming Branch of the Third College of PLA Information Engineering UniversityDepartment of Education of Yunnan Province, Engineering Research Center of Computer Vision and Intelligent Control TechnologySchool of Information, Yunnan Normal UniversitySchool of Information, Yunnan Normal UniversitySchool of Information, Yunnan Normal UniversityAbstract Small object detection in UAV aerial images is challenging due to low contrast, complex backgrounds, and limited computational resources. Traditional methods struggle with high miss detection rates and poor localization accuracy caused by information loss, weak cross-layer feature interaction, and rigid detection heads. To address these issues, we propose LRDS-YOLO, a lightweight and efficient model tailored for UAV applications. The model incorporates a Light Adaptive-weight Downsampling (LAD) module to retain fine-grained small object features and reduce information loss. A Re-Calibration Feature Pyramid Network (Re-Calibration FPN) enhances multi-scale feature fusion using bidirectional interactions and resolution-aware hybrid attention. The SegNext Attention mechanism improves target focus while suppressing background noise, and the dynamic detection head (DyHead) optimizes multi-dimensional feature weighting for robust detection. Experiments show that LRDS-YOLO achieves 43.6% mAP50 on VisDrone2019, 11.4% higher than the baseline, with only 4.17M parameters and 24.1 GFLOPs, striking a balance between accuracy and efficiency. On the HIT-UAV infrared dataset, it reaches 84.5% mAP50, demonstrating strong generalization. With its lightweight design and high precision, LRDS-YOLO offers an effective real-time solution for UAV-based small object detection.https://doi.org/10.1038/s41598-025-07021-6UAV (unmanned aerial vehicle)Small object detectionFeature pyramid networkReal timeAttention mechanisms
spellingShingle Yuqi Han
Chengcheng Wang
Hui Luo
Huihua Wang
Zaiqing Chen
Yuelong Xia
Lijun Yun
LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient design
Scientific Reports
UAV (unmanned aerial vehicle)
Small object detection
Feature pyramid network
Real time
Attention mechanisms
title LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient design
title_full LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient design
title_fullStr LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient design
title_full_unstemmed LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient design
title_short LRDS-YOLO enhances small object detection in UAV aerial images with a lightweight and efficient design
title_sort lrds yolo enhances small object detection in uav aerial images with a lightweight and efficient design
topic UAV (unmanned aerial vehicle)
Small object detection
Feature pyramid network
Real time
Attention mechanisms
url https://doi.org/10.1038/s41598-025-07021-6
work_keys_str_mv AT yuqihan lrdsyoloenhancessmallobjectdetectioninuavaerialimageswithalightweightandefficientdesign
AT chengchengwang lrdsyoloenhancessmallobjectdetectioninuavaerialimageswithalightweightandefficientdesign
AT huiluo lrdsyoloenhancessmallobjectdetectioninuavaerialimageswithalightweightandefficientdesign
AT huihuawang lrdsyoloenhancessmallobjectdetectioninuavaerialimageswithalightweightandefficientdesign
AT zaiqingchen lrdsyoloenhancessmallobjectdetectioninuavaerialimageswithalightweightandefficientdesign
AT yuelongxia lrdsyoloenhancessmallobjectdetectioninuavaerialimageswithalightweightandefficientdesign
AT lijunyun lrdsyoloenhancessmallobjectdetectioninuavaerialimageswithalightweightandefficientdesign