Date of Award

3-10-2021

Publication Type

Master Thesis

Degree Name

M.Sc.

Department

Computer Science

First Advisor

Dan Wu

Rights

info:eu-repo/semantics/openAccess

Abstract

Machine learning is a broad study of computer science, widely used for data analysis and algorithms that has the ability to learn and improve by experience through training. Supervised learning, Unsupervised learning, Dimensionality Reduction, Deep Learning, etc are the methods offered by Machine learning. These techniques are applied in fields like medical, automotive finance, and many more. In this thesis, Convolutional neural network (CNN) which is a part of deep learning techniques is applied to identify if a person is under influence of Marijuana or sober, using facial feature changes like redness in eyes, watery eyes, and drowsiness caused after smoking Marijuana. CNN is a state-of-the-art method in tasks like image classification and pattern recognition. CNN’s ability to learn from training the model using image dataset is a suitable method to be used in the problem of identifying a person’s sobriety based on facial features. The proposed methodology is divided into three components. Which are dataset creation, face detection to extract input image from real-time video, and finally, tuning and training CNN model for making a prediction. The purpose of this thesis is to develop a CNN model that may be helpful if implemented in vehicles in the future to reduce impaired driving incidents. Impaired driving is a major criminal cause of vehicle accidents in Canada. Impaired driving is a serious problem that puts the lives of pedestrians on the road and drivers involved in impaired driving themselves in danger. This thesis presents how Machine Learning can be applied to predict driver’s sobriety that may be helpful in reducing impaired driving incidents in the future by implementing in vehicles.

Share

COinS