Realtime Single-Shot Refinement Neural Network for 3D Obejct Detection from LiDAR Point Cloud

Yutian Wu, Harutoshi Ogai

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

3D object detection from point cloud is an important aspect of environmental perception in intelligent systems such as autonomous driving systems and robot systems. However, efficient 3D feature extraction and accurate object localization is challenging for current algorithms. In this paper, we introduce a new single-shot refinement neural network for fast and accurate 3D object detection. Firstly, we simplify the 3D feature extraction network and use single-shot object detector to increase processing speed. Secondly, we exploit self-attention mechanism in main object detection branch to improve object feature representation. Thirdly, an object refinement branch is introduced to produce a finer regression of objects upon the primary estimation from the main detection branch. Both modifications lead to further improvements in performance without additional computational cost. Our approach is tested on KITTI 3D Car detection benchmark and achieves good results in the validation set. The running speed is around 40 frame per second.

Original languageEnglish
Title of host publication2020 59th Annual Conference of the Society of Instrument and Control Engineers of Japan, SICE 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages332-337
Number of pages6
ISBN (Electronic)9781728110899
Publication statusPublished - 2020 Sep 23
Event59th Annual Conference of the Society of Instrument and Control Engineers of Japan, SICE 2020 - Chiang Mai, Thailand
Duration: 2020 Sep 232020 Sep 26

Publication series

Name2020 59th Annual Conference of the Society of Instrument and Control Engineers of Japan, SICE 2020

Conference

Conference59th Annual Conference of the Society of Instrument and Control Engineers of Japan, SICE 2020
CountryThailand
CityChiang Mai
Period20/9/2320/9/26

Keywords

  • 3D object detection
  • LiDAR point cloud
  • neural network
  • single-shot object detector

ASJC Scopus subject areas

  • Control and Optimization
  • Instrumentation
  • Computer Vision and Pattern Recognition
  • Signal Processing
  • Decision Sciences (miscellaneous)
  • Industrial and Manufacturing Engineering

Fingerprint Dive into the research topics of 'Realtime Single-Shot Refinement Neural Network for 3D Obejct Detection from LiDAR Point Cloud'. Together they form a unique fingerprint.

Cite this