Drunkenness detection using a CNN with adding Gaussian noise and blur in the thermal infrared images. (5th September 2022)

• Record Type: Journal Article
• Authors: Huynh, Kha Tu; Nguyen, Huynh Phuong Thanh
• Is Part Of: International journal of intelligent information and database systems. Volume 15:Number 4(2022)
• Page Start: 398
• Page End: 419
• Publication Date: 2022-09-05
• Subjects: drunkenness detection -- convolutional neural network -- CNN -- thermal infrared image -- Gaussian noise -- blur -- machine learning -- augmentation; Database management -- Computer programs -- Periodicals; Information retrieval -- Computer programs -- Periodicals; Information storage and retrieval systems -- Computer programs -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Intelligent agents (Computer software) -- Periodicals; 006.33
• Languages: English
• ISSNs: 1751-5858

Abstract

Drunkenness is now often regarded as one of society's most serious issues. The majority of road accidents are caused by drunk driving. This paper proposes a methodology based on evaluating a facial thermal infrared image, adding noise and filters for augmentation, and determining intoxication using machine learning algorithms. In drunkenness detection, most research focus on using RGB image of facial expressions like eye sate, head position, or functional state indicators. Sometimes it is not trusty when attempting to predict on certain people who have certain facial feature patterns that the machine learning algorithm learned to be a factor of drunkenness. The combination of using the thermal infrared image with some noise and filter then predicting by optimised convolution neural network (CNN) model approach 93% on accuracy proves the efficiency as well as the feasibility of the proposed method.