Deep Learning Dataset Augmentation for Detecting Rare and Severe Contaminants in Waste Management Collection.

This project strives to identify rare and severe contaminants found in residential curbside recycling. Using computer vision techniques and machine learning, cameras analyze contaminants as they are tipped into recycling trucks. Some of these contaminants are rare, however they can pose severe risks to operators and the general public. As these contaminants are not common, they are difficult to detect. By enabling municipalities the means to collect data on previously unidentifiable items, a multitude of new approaches will be developed for education and outreach. The ultimate effect of these activities will be proper recycling practices which promotes a circular economy, thereby reducing our dependence on virgin plastics and fossil fuels.

Faculty Supervisor:

Ezeddin Shirif;Mohamed El-darieby

Student:

Partner:

Prairie Robotics Inc.

Discipline:

Engineering

Sector:

Administrative and support, waste management and remediation services

University:

University of Regina

Program: