Deep Learning Analysis for Missing Tooth Detection in Mining Monitoring Systems

This project is aimed at using machine learning algorithms and techniques to enhance the current state of the art of missing tooth detection in mining monitoring systems. Unlike heuristic approaches that follow strictly static program instructions, machine learning techniques operate by building a model from example inputs in order to make data-driven predictions or decisions. We use machine learning techniques to identify the bucket and its teeth within the video frames taken by a camera located on the mining device. We keep track of the detected objects within the images to monitor the status of the teeth over time and detect a potentially missing tooth. We train our object recognition model based on a comprehensive database of over 200 hours of video footage, and evaluate our algorithm in the end by means of image benchmarks including various teeth locations/ scales and various weather conditions.

Faculty Supervisor:

Guy Dumont

Student:

Nazanin Hamzei

Partner:

Motion Metrics International Corp

Discipline:

Engineering - computer / electrical

Sector:

Advanced manufacturing

University:

University of British Columbia