Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

LDSNet: A Lightweight Detail-Sensitive Network for Small Object Detection in Low-Altitude UAV Scenarios.

Tong Tan1,2, Xianrong Peng1, Jianlin Zhang1

  • 1State Key Laboratory of Optical Field Manipulation Science and Technology, Institute of Optics and Electronics, Chinese Academy of Sciences, Chengdu 610209, China.

Journal of Imaging
|May 26, 2026
PubMed
Summary

Related Concept Videos

Difference from Background: Limit of Detection01:05

Difference from Background: Limit of Detection

The limit of detection (LOD) is the smallest amount of analyte that can be distinguished from the background noise. The LOD value corresponds to the concentration at which the analyte signal is three times larger than the standard deviation of the blank signal. Below this value, the analyte signal cannot be differentiated from the background noise. It is calculated by dividing the calibration slope by 3 times the standard deviation of the blank signals.
The LOD indicates the presence or absence...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Nonlinear association between systemic inflammation response index and mortality in adult cardiac surgery-associated acute kidney injury: A retrospective Cohort study based on the MIMIC-IV database.

Medicine·2026
Same author

Microbial organic fertilizer improves soil microbial community and promotes tobacco growth.

Journal of environmental quality·2026
Same author

A cross-cultural comparison of educational video quality on college students' anxiety and depression: a cross-sectional content analysis of YouTube and Bilibili.

Frontiers in public health·2026
Same author

Unified Multi-Modal Object Tracking Through Spatial-Temporal Propagation and Modality Synergy.

Journal of imaging·2025
Same author

Transapical beating-heart approach is advantageous for surgical septal reduction therapy in pediatrics.

The Journal of thoracic and cardiovascular surgery·2025
Same author

Use of the novel virtual myectomy in guiding thoracoscopic myectomy for patients with hypertrophic obstructive cardiomyopathy.

International journal of cardiology. Heart & vasculature·2025

This study introduces the Lightweight Detail-Sensitive Network (LDSNet) for improved object detection in Unmanned Aerial Vehicle (UAV) imagery. LDSNet enhances small object feature representation, boosting accuracy and efficiency in aerial surveillance.

Area of Science:

  • Computer Vision
  • Artificial Intelligence
  • Robotics

Background:

  • Object detection in Unmanned Aerial Vehicle (UAV) imagery is hindered by the aerial perspective, particularly for small objects.
  • Weak feature representation of small objects limits detection accuracy and computational efficiency in UAV systems.

Purpose of the Study:

  • To develop a novel network, the Lightweight Detail-Sensitive Network (LDSNet), to address the challenges of small object detection in UAV imagery.
  • To improve both the accuracy and computational efficiency of object detection algorithms for aerial platforms.

Main Methods:

  • Proposed LDSNet incorporates Lightweight Detail-Sensitive Downsampling (LDSDown) to preserve small object details during downsampling.
  • Utilized Shared Recursive Dilated Convolution (SRDC) to capture multi-scale context efficiently without increasing parameters.
Keywords:
Unmanned Aerial Vehicle (UAV)lightweight networkreal-time inferencesmall object detection

Related Experiment Videos

  • Implemented Deeply Decoupled Grouped Head (DGHead) to reduce computational cost for high-resolution UAV inputs.

    • Main Results:

    • LDSNet demonstrated a significant reduction in parameters (84.6%) and FLOPs (29.2%) compared to the YOLOv11n baseline.
    • Achieved a 2.2% improvement in mAP50 on the VisDrone2019 dataset.
    • Attained 94.5% accuracy on the HIT-UAV dataset, showcasing superior performance.

    Conclusions:

    • LDSNet offers an excellent trade-off between accuracy and efficiency for object detection in UAV imagery.
    • The proposed network effectively enhances the feature representation of small objects from aerial perspectives.
    • LDSNet presents a computationally efficient and accurate solution for real-world UAV applications.