Driver drowsiness detection system using convolutional neural network

Komal*, P. Sharma, A. Lamba, B. Nagpal, S. Chauhan

*Corresponding author for this work

Research output: Contribution to journalArticle


Rapid increase in the number of automobiles plying on the road has not only increased the traffic jams but the probability of having more road accidents. Sleep deprivation remains the primary cause regarding loss in concentration levels and has progressively increased in modern times. As per the statistics of NHTSA, Drowsiness during driving resulted in 44,000 crashes leaving nearly 800 dead. The methods proposed for detecting the drowsiness are very complex and time consuming due to usage of large mathematical computations. In this paper, an unprecedented Deep learning based approach towards real-time driver drowsiness and distraction detection using object recognition is proposed. The work is done making use of a vision-based cost-effective system. It uses 68 point facial Landmarks to find aspect ratios for eyes and mouth-EAR and MAR as these remain the most prominent features that get affected by drowsiness, to identify the probability of driver to fall asleep. It implements object recognition built using various deep learning algorithms simultaneously to get the most promising results to recognize objects that leads to distraction. On identifying drowsiness or distraction, the system will sound a caution alarm along with a message to notify the driver. The proposed model is capable of achieving accuracy of 90%.

Original languageEnglish
Pages (from-to)502-506
Number of pages5
JournalInternational Journal on Emerging Technologies
Issue number3
Publication statusPublished - 30 Apr 2020


  • Convolutional neural network
  • Deep learning
  • Drowsiness detection
  • Object detection
  • Tensor flow

ASJC Scopus subject areas

  • Engineering (miscellaneous)
  • Agricultural and Biological Sciences (miscellaneous)
  • Management of Technology and Innovation

Fingerprint Dive into the research topics of 'Driver drowsiness detection system using convolutional neural network'. Together they form a unique fingerprint.

  • Cite this