Deep Learning-based Electric Kickboard Helmet Detection under Night Driving Conditions

Daun Kim, Jin Woo Jeong

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

As the use of electric kickboards has increased recently, the frequency of safety accidents is also increasing. Even though the government amended the road traffic law to make it compulsory to wear a helmet, the helmet-wearing rate is still remarkably low. In this work, we propose a novel method to monitor and detect whether a user of an electric kickboard is wearing a helmet through a deep learning-based object detection network, thereby increasing the helmet-wearing rate. For continuous driver monitoring, the proposed method captures the user's face in a front view using the user's mobile phone attached to the kickboard. In addition, to handle both day and night driving conditions, various image conversion, enhancement, and augmentation techniques were considered and evaluated. The experimental results showed the feasibility of the proposed method.

Original languageEnglish
Pages (from-to)1411-1419
Number of pages9
JournalTransactions of the Korean Institute of Electrical Engineers
Volume71
Issue number10
DOIs
StatePublished - Oct 2022

Keywords

  • Deep learning
  • Electric Kickboard
  • Helmet detection
  • Night Driving
  • Personal mobility driving safety

Fingerprint

Dive into the research topics of 'Deep Learning-based Electric Kickboard Helmet Detection under Night Driving Conditions'. Together they form a unique fingerprint.

Cite this