This is an old revision of the document!
Table of Contents
Currently offered Projects, Fall 2011 (updated September 3, 2011)
(Listed in order received.)
Building an autonomous motorboat
Supervisor: Michael Jenkin
Required Background: General CSE408x prerequisites
Recommended Background: Robotics
Description An opportunity exists for a small number of students to build an autonomous motorboat using a RC motorboat as a base and integrating computation and control in the form of a Beagleboard. Students will participate in lectures and labs associated with CSE6324 (Part I). Interested students should attend the first lecture of CSE6324. See the departmental schedule for time and place.
:
Athenians Data Project
Supervisor: Nick Cercone
Required Background: General CSE408x prerequisites
Recommended Background: Data Mining
Description The Athenians Project is a multi-year, ongoing project of compiling, computerizing and studying data about the persons of ancient Athens. Possible project ideas for this term span from simpler ones such as how to present data in the best possible way, add spatial characteristics to existing data, add multimedia data, improve text searching, etc. to more complex ideas such as filling missing parts for the “broken” words on the existing inscriptions. Filling text for the broken words has been done in the past using expert knowledge. Those experts have establish certain rules/guidelines that may be possible to extrapolate in some kind of expert system when talking in IT terminology. Furthermore, any hypotheses on word completion enters the database with some likelihood. Associating probabilities with hypotheses introduces another opportunity for research projects.
:
Three-Dimensional Context from Linear Perspective for Video Surveillance Systems
Supervisor: James Elder
Requirements: Good facility with applied mathematics
Description
To provide visual surveillance over a large environment, many surveillance cameras are typically deployed at widely dispersed locations. Making sense of activities within the monitored space requires security personnel to map multiple events observed on two-dimensional security monitors to the three-dimensional scene under surveillance. The cognitive load entailed rises quickly as the number of cameras, complexity of the scene and amount of traffic increases.
This problem can be addressed by automatically pre-mapping two-dimensional surveillance video data into three-dimensional coordinates. Rendering the data directly in three dimensions can potentially lighten the cognitive load of security personnel and make human activities more immediately interpretable.
Mapping surveillance video to three-dimensional coordinates requires construction of a virtual model of the three-dimensional scene. Such a model could be obtained by survey (e.g., using LIDAR), but the cost and time required for each site would severely limit deployment. Wide-baseline uncalibrated stereo methods are developing and have potential utility, but require careful sensor placement, and the difficulty of the correspondence problem limits reliability.
This project will investigate a monocular method for inferring three-dimensional context for video surveillance. The method will make use of the fact that most urban scenes obey the so-called “Manhattan-world” assumption, viz., a large proportion of the major surfaces in the scene are rectangles aligned with a three-dimensional Cartesian grid (Coughlan & Yuille, 2003). This regularity provides strong linear perspective cues that can potentially be used to automatically infer three-dimensional models of the major surfaces in the scene (up to a scale factor). These models can then be used to construct a virtual environment in which to render models of human activities in the scene.
Although the Manhattan world assumption provides powerful constraints, there are many technical challenges that must be overcome before a working prototype can be demonstrated. The prototype requires six stages of processing: 1)The major lines in each video frame are detected. 2) These lines are grouped into quadrilaterals projecting from the major surface rectangles of the scene. 3) The geometry of linear perspective and the Manhattan world constraint are exploited to estimate the three-dimensional attitude of the rectangles from which these quadrilaterals project. 4) Trihedral junctions are used to infer three-dimensional surface contact and ordinal depth relationships between these surfaces. 5) The estimated surfaces are rendered in three-dimensions. 6) Human activities are tracked and rendered within this virtual three-dimensional world.
The student will work closely with graduate students and postdoctoral fellows at York University, as well as researchers at other institutions involved in the project. The student will develop skills in using MATLAB, a very useful mathematical programming environment, and develop an understanding of basic topics in image processing and vision.
For more information on the laboratory: http://www.elderlab.yorku.ca
:
Estimating Pedestrian and Vehicle Flows from Surveillance Video
Supervisor: James Elder
Requirements: Good facility with applied mathematics
Description
Facilities planning at both city (e.g., Toronto) and institutional (e.g., York University) scales requires accurate data on the flow of people and vehicles throughout the environment. Acquiring these data can require the costly deployment of specialized equipment and people, and this effort must be renewed at regular intervals for the data to be relevant.
The density of permanent urban video surveillance camera installations has increased dramatically over the last several years. These systems provide a potential source of low-cost data from which flows can be estimated for planning purposes.
This project will explore the use of computer vision algorithms for the automatic estimation of pedestrian and vehicle flows from video surveillance data. The ultimate goal is to provide planners with accurate, continuous, up-to-date information on facility usage to help guide planning.
The student will work closely with graduate students and postdoctoral fellows at York University, as well as researchers at other institutions involved in the project. The student will develop skills in using MATLAB, a very useful mathematical programming environment, and develop an understanding of basic topics in image processing and vision.
For more information on the laboratory: http://www.elderlab.yorku.ca
—- :
Tandem repeat detection using spectral methods
Supervisor: Suprakash Datta
Required Background: The student should have completed undergraduate courses in Algorithms and Signals and Systems.
Recommended Background: Some background in Statistics is desirable but not essential.
Description DNA sequences of organisms have many repeated substrings. These are called repeats in Biology, and include both exact as well as approximate repeats. Repeats are of two main types: interspersed repeats (which are spread across a genome) and tandem repeats, which occur next to each other. Tandem repeats play important roles in gene regulation and are also used as markers that have several important uses, including human identity testing.
Finding tandem repeats is an important problem in Computational Biology. The techniques that have been proposed for it fall into two classes: string matching algorithms and signal processing techniques. In this project, we will explore fast, accurate algorithms for detecting tandem repeats and evaluate the outputs of the algorithms studied by comparing their outputs with those of available packages, including mreps (http://bioinfo.lifl.fr/mreps/), SRF (http://www.imtech.res.in/raghava/srf/) and TRF (http://tandem.bu.edu/trf/trf.html).
The student will implement existing spectral algorithms based on Fourier Transforms and on an autoregressive model. He will then make changes suggested by the supervisor, and evaluate the effect of the modifications. Throughout the course, the student is required to maintain a course Web site to report any progress and details about the project.
:
Touch- and Gesture-based Text Entry With Automatic Error Correction
Supervisor: Scott Mackenzie
Required Background: CSE3461 (or equivalent), CSE3311 (or equivalent), CSE4441 (or equivalent) A student wishing to do this project must be well versed in Java, Eclipse, and developing java code for the Android operating system.
Recommended Background: Possession of an Android touch-based phone or tablet would be an asset, but is not essential.
Description This project involves extending a touch-based text entry method to include automatic error correction. The method, as is, uses Graffiti strokes entered via a finger on a touch-based Android tablet. The stroke recognizer works fine, but it is not perfect. Some strokes are mis-recognized while others are un-recognized. The fault is sometimes attributable to the recognizer, but, often, the fault is simply that the user's input was sloppy. The work involves developing, integrating, and testing software. The core software is already written, but automatic error correction is lacking. The primary task of the added software is to receive a sequence of characters representing a word and matching the sequence with words in a dictionary. If a match is found, all is well (presumably). If a match is not found, the search is extended to find a set of candidate words that are “close” to the inputted sequence. “Close”, here, involves using a minimum string distance algorithm (provided). The user interface must be modified to present the user with alternative words in the event an error occurred. The user selects the desired word by tapping on a word in the list. The project will involve testing the new input method in a small user study and writing up a report describing the work and presenting the results of the user study.
:
Early Breast Cancer Detection based on MRI’s.
Supervisor: Amir Asif
Required Background: General CSE408x prerequisites
Recommended background: Signal processing
Project Description: This research will develop advanced computer-aided, signal processing techniques for early detection of breast cancer using the available modalities. In particular, we propose to develop time reversal beamforming imager, based on our earlier work in time reversal signal processing, for detecting early stage breast cancer tumours from MRI data. Our preliminary work has illustrated the type of results that are possible for breast cancer detection by applying time reversal signal processing on MRI breast data. In this research, we propose to extend these results to provide a quantitative understanding of the practical gains provided by time reversal in MRI based breast cancer detection and its limitations. This will be accomplished a local hospital, and running our algorithms on these datasets. The first step is important to check the validity of our algorithms. The next step is to compare the estimated locations of the tumours (as derived with our algorithms) to their precise locations as identified by the pathologists. The second step will quantify the accuracy of our estimation algorithms.
: