Pvalite CLN: Lightweight Object Detection with Classfication and Localization Network

Guo Jiun-In, Tsai Chi-Chi, Tseng Ching-Kan

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

We developed a 2D convolution object detection network called CLN layer to let a network with a small amount of parameters and have a better detection result as well. Compared to YOLOv2 detector and SSD with lightweight Pvalite network, the proposed Classification and Localization Network (CLN) layer has better accuracy and higher recall on the object which is located on the border of image. Additionally, we revise the original open source to make this architecture more platform portable than other architectures. The proposed system is implemented on the embedded platforms with the performance of 640x480@10 fps under nVidia Jetson TX-2 and 480x320@5fps under Renesas R-car H3.

Original languageEnglish
Title of host publicationProceedings - 32nd IEEE International System on Chip Conference, SOCC 2019
EditorsDanella Zhao, Arindam Basu, Magdy Bayoumi, Gwee Bah Hwee, Ge Tong, Ramalingam Sridhar
PublisherIEEE Computer Society
Pages118-121
Number of pages4
ISBN (Electronic)9781728134826
DOIs
StatePublished - Sep 2019
Event32nd IEEE International System on Chip Conference, SOCC 2019 - Singapore, Singapore
Duration: 3 Sep 20196 Sep 2019

Publication series

NameInternational System on Chip Conference
Volume2019-September
ISSN (Print)2164-1676
ISSN (Electronic)2164-1706

Conference

Conference32nd IEEE International System on Chip Conference, SOCC 2019
Country/TerritorySingapore
CitySingapore
Period3/09/196/09/19

Keywords

  • Classification
  • Localization
  • Object Detection

Fingerprint

Dive into the research topics of 'Pvalite CLN: Lightweight Object Detection with Classfication and Localization Network'. Together they form a unique fingerprint.

Cite this