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

Ongoing Progress on Pervaporation Membranes for Ethanol Separation.

Membranes·2023
Same author

Deep Learning-Based NMPC for Local Motion Planning of Last-Mile Delivery Robot.

Sensors (Basel, Switzerland)·2022
Same author

Deep Reinforcement Learning for End-to-End Local Motion Planning of Autonomous Aerial Robots in Unknown Outdoor Environments: Real-Time Flight Experiments.

Sensors (Basel, Switzerland)·2021
Same author

An End-to-End Deep Reinforcement Learning-Based Intelligent Agent Capable of Autonomous Exploration in Unknown Environments.

Sensors (Basel, Switzerland)·2018

Related Experiment Video

Updated: Oct 30, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

728

Transfer Learning Based Semantic Segmentation for 3D Object Detection from Point Cloud.

Muhammad Imad1, Oualid Doukhi1, Deok-Jin Lee2

  • 1Center for Artificial Intelligence & Autonomous Systems, Kunsan National University, 558 Daehak-ro, Naun 2(i)-dong, Gunsan 54150, Korea.

Sensors (Basel, Switzerland)
|July 2, 2021
PubMed
Summary

This study introduces a transfer learning method for 3D object detection using LiDAR data, reducing the need for large datasets and training time. The approach achieves competitive accuracy and high speed for autonomous driving perception.

Keywords:
3D object detectionpoint cloud processingsemantic segmentationtransfer learning

More Related Videos

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

3.1K
Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

9.8K

Related Experiment Videos

Last Updated: Oct 30, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

728
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

3.1K
Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

9.8K

Area of Science:

  • Computer Vision
  • Robotics
  • Autonomous Systems

Background:

  • 3D object detection using LiDAR is crucial for autonomous driving, outperforming cameras at night.
  • Supervised methods require extensive labeled data, which is costly and limited.
  • Existing transfer learning for object detection primarily focuses on 2D, not 3D.

Purpose of the Study:

  • To develop an efficient transfer learning approach for 3D object detection using LiDAR point clouds.
  • To reduce reliance on large-scale annotated datasets and decrease training duration.
  • To improve the utilization of 3D point cloud data through Bird's-Eye-View (BEV) representation.

Main Methods:

  • Preprocessing raw LiDAR data into a BEV map.
  • Employing transfer learning from a classification task to a semantic segmentation-based 2D object detection task.
  • Back-projecting 2D detection results from BEV to 3D in a postprocessing step.

Main Results:

  • Achieved competitive mean average precision (mAP) up to 70% on KITTI and Ouster LiDAR-64 datasets.
  • Maintained high processing speeds exceeding 30 frames per second (FPS).
  • Demonstrated significant reduction in the need for large-scale training datasets.

Conclusions:

  • The proposed transfer learning method effectively addresses the data requirements for 3D object detection.
  • The BEV representation and semantic segmentation approach enhance LiDAR data utilization.
  • The model offers a practical and efficient solution for real-time autonomous driving perception.