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 Concept Videos

You might also read

Related Articles

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

Sort by
Same author

EPGNet: Enhanced Point Cloud Generation for 3D Object Detection.

Sensors (Basel, Switzerland)·2020
Same author

Homoharringtonine, aclarubicin and cytarabine (HAA) regimen as the first course of induction therapy is highly effective for acute myeloid leukemia with t (8;21).

Leukemia research·2016
Same author

[Morphology and Spectral Properties Study of LaCeF3:Tb Microcrystalline].

Guang pu xue yu guang pu fen xi = Guang pu·2016
Same author

A highly selective fluorogenic probe for the detection and in vivo imaging of Cu/Zn superoxide dismutase.

Chemical communications (Cambridge, England)·2016
Same author

Development of a disaggregation-induced emission probe for the detection of RecA inteins from Mycobacterium tuberculosis.

Chemical communications (Cambridge, England)·2016
Same author

Bias-polarity-dependent resistance switching in W/SiO2/Pt and W/SiO2/Si/Pt structures.

Scientific reports·2016

Related Experiment Video

Updated: Nov 23, 2025

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

655

PSANet: Pyramid Splitting and Aggregation Network for 3D Object Detection in Point Cloud.

Fangyu Li1, Weizheng Jin1, Cien Fan1

  • 1School of Electronic Information, Wuhan University, Wuhan 430072, China.

Sensors (Basel, Switzerland)
|December 31, 2020
PubMed
Summary
This summary is machine-generated.

This study introduces a novel backbone network for 3D object detection using LiDAR point clouds, enhancing bird's eye view (BEV) information utilization. The proposed method significantly improves detection performance by effectively fusing multi-scale BEV features.

Keywords:
3D object detectionLiDARautonomous drivingconvolutional neural networksvoxel

More Related Videos

Measuring the Structure, Composition, and Change of Underwater Environments with Large-area Imaging
09:19

Measuring the Structure, Composition, and Change of Underwater Environments with Large-area Imaging

Published on: April 18, 2025

1.2K
A Method for 3D Reconstruction and Virtual Reality Analysis of Glial and Neuronal Cells
12:49

A Method for 3D Reconstruction and Virtual Reality Analysis of Glial and Neuronal Cells

Published on: September 28, 2019

13.1K

Related Experiment Videos

Last Updated: Nov 23, 2025

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

655
Measuring the Structure, Composition, and Change of Underwater Environments with Large-area Imaging
09:19

Measuring the Structure, Composition, and Change of Underwater Environments with Large-area Imaging

Published on: April 18, 2025

1.2K
A Method for 3D Reconstruction and Virtual Reality Analysis of Glial and Neuronal Cells
12:49

A Method for 3D Reconstruction and Virtual Reality Analysis of Glial and Neuronal Cells

Published on: September 28, 2019

13.1K

Area of Science:

  • Computer Vision
  • Robotics
  • Autonomous Systems

Background:

  • 3D object detection in LiDAR point clouds is crucial for autonomous driving and robotics.
  • Existing one-stage 3D detectors face performance limitations due to suboptimal bird's eye view (BEV) information utilization.

Purpose of the Study:

  • To propose a new backbone network for enhancing 3D object detection performance.
  • To improve the utilization of multi-scale BEV feature maps in LiDAR point clouds.

Main Methods:

  • Introduced a novel backbone network with a coarse branch utilizing pyramidal feature hierarchy (PFH) and a fine branch with a pyramid splitting and aggregation (PSA) module.
  • Implemented cross-layer fusion of multi-scale BEV feature maps to enhance feature expressiveness.

Main Results:

  • The proposed method demonstrated superior performance in both 3D and BEV object detection on the KITTI-3D benchmark.
  • Experimental results, measured by average precision (AP), validated the effectiveness of the new network architecture.

Conclusions:

  • The novel backbone network effectively fuses multi-scale BEV features, leading to improved 3D object detection.
  • The proposed approach offers a significant advancement for applications requiring accurate object detection from LiDAR data.