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 Video

Updated: Jun 3, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

3D object detection for vehicle-mounted LiDAR based on deep learning and euclidean clustering algorithm.

Nan Zhang1, Maolong Xi1, Juan Fang1

  • 1School of Control Engineering, Wuxi University of Technology, Wuxi, China.

Plos One
|June 1, 2026
PubMed
Summary

This study presents a novel 3D object detection (OD) method for autonomous driving, enhancing point cloud segmentation and OD accuracy using deep learning and improved clustering. The approach achieves high precision and real-time performance, outperforming existing methods.

Related Concept Videos

You might also read

Related Articles

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

Sort by
Same author

A Community-Based Health Literacy Intervention to Improve Health Literacy, Health Behavior Self-Efficacy, and Quality of Life Among Older Adults With Frailty: A Quasi-Experimental Study.

Journal of gerontological nursing·2026
Same author

Leveraging multimodal cancer immunotherapy to amplify the efficacy of oncolytic viruses.

Experimental hematology & oncology·2026
Same author

Genomic and clinical insights into multidrug-resistant <i>Corynebacterium striatum</i> in a psychiatric hospital: development of an exploratory infection probability score.

Frontiers in microbiology·2026
Same author

Screening and identification of key genes related to the immune microenvironment of rectal cancer influenced by radiotherapy based on bioinformatics methods.

Discover oncology·2026
Same author

Patterns and predictors of cognitive function and post-traumatic growth in cancer patients: a latent profile analysis.

Frontiers in psychiatry·2026
Same author

Dual-system engineered bacteria and dendritic cells enable precise intratumoral IL-12 delivery and potent antitumor immunity.

Cell reports. Medicine·2026

Area of Science:

  • Computer Vision
  • Robotics
  • Artificial Intelligence

Background:

  • Object Detection (OD) is crucial for autonomous driving perception.
  • Existing methods struggle with sparse and uneven LiDAR data, impacting accuracy and efficiency.
  • Deep learning and clustering algorithms are key areas for improvement.

Purpose of the Study:

  • To develop an accurate and efficient 3D OD method for autonomous driving.
  • To enhance point cloud segmentation and object detection performance.
  • To address limitations in handling challenging LiDAR data.

Main Methods:

  • Integration of Cloth Simulation Filter (CSF) for ground/non-ground separation.
  • Utilized K-Dimensional Tree (KD-Tree) with adaptive parameters for robust clustering.

Related Experiment Videos

Last Updated: Jun 3, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

  • Enhanced PointNet architecture with multi-scale grouping (MSG), multi-resolution grouping (MRG), and skip connections for feature extraction and fusion.

    • Main Results:

    • Achieved segmentation accuracies of 94.96% (KITTI) and 93.12% (NuScenes) with processing times under 20ms.
    • 3D OD task achieved average accuracies of 94.36% (KITTI) and 92.68% (NuScenes).
    • Demonstrated statistically significant improvements (p < 0.001) over standard PointNet, with detection speeds of 34 fps and 31 fps.

    Conclusions:

    • The proposed framework offers a robust, efficient, and generalizable solution for 3D environmental perception.
    • The method excels in challenging scenarios with occluded and multi-scale objects.
    • Achieved a superior balance of speed and precision compared to traditional methods.