A long daily commute for millions of drivers fuels serious health and economic consequences. Vehicle emissions are the leading cause of air pollution in North America; and idling for long periods in traffic each day significantly reduces productivity. In an effort to curb traffic congestion in urban centres, Faisal Qureshi, PhD, Associate Professor of Computer Science, in the Faculty of Science, is developing new techniques to capture and analyze big data involved in traffic camera networks, to better understand traffic patterns and conditions. Using that data, he is developing sensors for next generation smart camera networks to detect incidents, divert vehicles and improve traffic flow in urban centres and on highways. This emerging technology will make more intelligent use of existing road networks, help reduce air pollution, and ultimately mitigate climate change.
His research also has meaningful benefits for older adults who want to live independently at home for as long as possible. Dr. Qureshi is creating smart camera networks to allow family members to monitor elderly relatives, particularly those with dementia. To facilitate smart homes, he is developing sensors that family members could program to give cues to the user, reminding them to complete important daily tasks, and helping them maintain independence.
A computer science expert, Dr. Qureshi is fascinated by understanding how human intelligence works and using it to build novel systems capable of carrying out complex tasks without human intervention. He earned his Bachelor of Science in Mathematics with a Minor in Physics from Punjab University in Lahore, Pakistan in 1992, and his Master of Science in Electronics from Quaid-e-Azam University in Islamabad, Pakistan in 1995.
Dr. Qureshi joined UOIT in July 2008 as an assistant professor. Instrumental in establishing the university’s Visual Computing (VC) Lab, he was named associated professor in July 2013. He gained industry experience as a software engineer with Toronto-based Autodesk Canada Co., and as a contract engineer with MDRobotics in Brampton, Ontario.
Industry Expertise (5)
Areas of Expertise (9)
Behaviour-based Computer Animation
Autonomous Characters for Computer Animation and Games
Autonomous Agent Architectures
Director, Visual Computing (VC) Lab (professional)
Dr. Qureshi established UOIT's state-of-the-art VC Lab which focuses on research problems that reside at the intersection of computer vision, visual sensor networks, and computer graphics.
Co-Chair, 13th Conference on Computer and Robot Vision (professional)
Co-Chair of the 12th Conference on Computer and Robot Vision in Halifax, Nova Scotia in June 2015, Dr. Qureshi will also Co-Chair next year's conference in Victoria British Columbia in June 2016.
University of Toronto: PhD, Computer Science 2007
University of Toronto: MSc, Computer Science 2000
Quaid-e-Azam University: MSc, Electronics 1995
Punjab University: BSc, Mathematics & Physics (Minor) 1992
- Institute of Electrical and Electronics Engineers
- Association for Computing Machinery
- Canadian Image Processing and Pattern Recognition Society
Event Appearances (6)
Towards Efficient Feedback Control in Streaming Computer Vision Pipelines
Workshop on User-Centred Computer Vision Singapore
Accelerating Cost Volume Filtering Using Salient Subvolumes and Robust Occlusion Handling
12th Asian Conference on Computer Vision (ACCV 2014) Singapore
A Stream Algebra for Computer Vision Pipelines
Second Workshop on Web-scale Vision Columbus, Ohio
Topic Models for Image Localization
Tenth Conference on Computer and Robot Vision (CRV 2013) Regina, Saskatchewan
I Remember Seeing This Video: Image Driven Search in Video Collections
Tenth Conference on Computer and Robot Vision Regina, Saskatchewan
Droplet Tracking from Unsynchronized Cameras
2nd International Conference on Pattern Recognition Applications and Methods (ICPRAM 2013) Barcelona, Spain
Research Grants (3)
Intelligent Traffic Control Through Multimodal Vehicle Detection and Classification
Ontario Centres of Excellence Voucher for Innovation and Productivity $43000
This one-year research project focuses on the development of a multimodal vehicular traffic detection and analysis system.
Next Generation Smart Camera Networks
NSERC Discovery Grant $90000
This research aims to develop new techniques to handle big data involved in large scale systems such as traffic camera networks, to identify individuals and detect incidents, and store them for later use.
Camera-based Vehicle Detection and Classification
NSERC Engage Grant $63078
This research examines current camera-based vehicle detection and classification systems, with the goal of developing a multimodal vehicular traffic detection and analysis system.
Principles of Computer Science
2nd Year, Undergraduate Course
This article presents the development of a new method for extracting 3D flight trajectories of droplets using high-speed stereo capture. Results suggest that, even when full stereo information is available, unsynchronized reconstruction using the global motion model can significantly improve the 3D estimation accuracy.
This research presents a distributed virtual vision simulator capable of simulating large-scale camera networks, and pedestrian traffic in different 3D environments. Specifically, this research shows that the proposed virtual vision simulator can model a camera network, comprising more than one hundred active pan/tilt/zoom and passive wide field-of-view cameras, deployed in an upper floor of an office tower in downtown Toronto.
Automatic detection of road boundaries in traffic surveillance imagery can greatly aid subsequent traffic analysis tasks, such as vehicle flow, erratic driving, and stranded vehicles. This paper develops an online technique for identifying the dominant road boundary in video sequences captured by traffic cameras under challenging environmental and lighting conditions, e.g., unlit highways captured at night. The proposed method works in real time of up to 20 frames/s and generates a ranked list of road regions that identify road and lane boundaries. Results show that this method outperforms two state-of-the-art techniques in precision, recall, and runtime.