The University of Windsor has moved to an “essential service only” model. Learn More.

MSc Thesis Defense Announcement by Ryan Bluteau: Obstacle and Change Detection Using Monocular Vision

Wednesday, June 12, 2019 - 09:30 to 11:30

SCHOOL OF COMPUTER SCIENCE

 

The School of Computer Science at the University of Windsor is pleased to present …

 

MSc Thesis Defense by:

Ryan Bluteau
 
Date:  Wednesday June 12, 2019
Time:  9:30 am – 11:30 am
Location: 3105, Lambton Tower
 

Abstract: 

We explore change detection using videos of change-free paths to detect any changes that occur while traveling the same paths in the future. This approach benefits from learning the background model of the given path as preprocessing, detecting changes starting from the first frame and determining the current location in the path. Two approaches are explored, a vision based approached and an approached based on deep learning. In our vision based approach, we extracted feature points and compared frames through feature matching. This is used to determine the current location in the change-free training video. The matching frames are then processed by first registering the test frame onto the training frame through a homography of the matching feature points. Finally, a comparison of a region of interest (ROI) of the direct path in both frames is made to determine changes. This approach showed good results various test videos with some issues in dynamic colour changes. In our deep learning approach, we use an ensemble of unsupervised dimensionality reduction models. We first extract feature points of a ROI of the frame and extract small frame samples around the feature points. The frame samples are used as training inputs and labels for our unsupervised models. The approach aims at learning a compressed feature representation of the frame samples and using a distribution of the training samples to discriminate between outliers, such as windows containing obstacles or changes. This approach performs well using just two models in the ensemble method.
 

Thesis Committee:

Internal Reader:  Dr. Imran Ahmad
External Reader:  Dr. Jonathan Wu
Advisor:  Dr. Boubakeur Boufama 
Chair: Dr. Saad Ahmed

 

Thesis Defense Announcement

 

(519)253-3000