Detection and Classification of Pavement Defects Using Computer Vision

The objective of the project is to automate the detection of pavement defects. Defects can be of different types: cracks, deformations, potholes and others. These defects are cataloged and detailed in standards established by the Quebec Ministry of Transport in Quebec and various authorities in other regions of the world. At present, the inspection of pavement defects (e.g. potholes, cracks, ruts) is mainly done manually. Inspectors crisscross the roads or scrutinize images taken from inspection vehicles. Some existing systems include an element of automation, but these systems are both expensive and not very adaptable to different regional conditions.
This project with Rival Solutions (RSI) seeks to automate this process using Computer Vision methods applied on geotagged images. The subject of study will be mainly Quebec, a region subject to climatic hazards, which have a significant impact on the condition of its road infrastructure.

Faculty Supervisor:

Amin Hammad

Student:

Partner:

Rival Solutions Inc.

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

Concordia University

Program: