Date: Friday February 7th, 2020

 

Time:  1:00 PM to 2:00 PM

 

Location: Lambton Tower, 3105

 

 

Now a days use of machine learning is becoming more and more popular in different fields like medical, automotive, finance, etc. and proven to be state of the art in tasks like image processing, decision making and classification. Considering these abilities, this thesis presents an approach to develop a system that can help identify a person intoxicated under influence of marijuana use. And may be helpful to develop a system that can help reducing vehicle accidents caused by impaired driving which is a major criminal cause of vehicle accidents in Canada. To achieve this deep learning techniques like convolutional neural network (CNN) and pattern recognition is used. Using Dlib library which contains a wide range of machine learning algorithms, image of a person can be extracted from a video. This image will be forwarded to a trained CNN to predict if the person’s sobriety based on facial features.

 

 

Internal Reader: Dr Jianguo Lu         

 

External Reader: Dr Dan Xiao

 

Advisor: Dr Dan Wu

 

 

5113 Lambton Tower 401 Sunset Ave. Windsor ON, N9B 3P4 (519) 253-3000 Ext. 3716 csgradinfo@uwindsor.ca