Dr. Brian Mac Namee
Home Research Publications Teaching
  DT228/4 Graphics & Image
  Processing
  DT217 Problem Solving,
  Communication & Innovation
  Final Year Projects
  DT286 Machine Learning   MSc Visualisation   DT286 Probability &
  Statistical Inference


The following is a list of a some suggested final year/MSc project topics for the coming year. I have tried to suggest which ones would be appropriate for undergraduate (BSc) and postgraduate (MSc) students, although this is just a suggestion. If you are interested in doing any of these projects (or any project in these or related areas) please contact me by e-mail.

Data Analytics

  • Academic Expertise Viewer (BSc): One challenge faced by researchers is to determine the expertise of colleagues or researchers in other institutions. This project would seek to create a topic map based on the publications of a researcher in which various terms representing topics would be shown connected to the researcher to indicate that this researcher was an expert in that particular field.

  • Literature Review Assistant (BSc/MSc): All good research is built upon a solid and comprehensive literature review. Often though finding papers and keeping track of the relationships between papers is a difficult task. This project will build an interactive visual tool that will help researchers discover and manage papers used in a literature. One of the key features of the tool will be that it will illustrate relationships between papers on a network. Important relationships include citations, common authors, common content vocabulary and common publishing location.

  • Traffic Camera Car Tracker (BSc): South Dublin County Council, amongst others, make traffic camera feeds available online (see here for a selection). This is a useful service that allows people estimate journey times, check for congestion etc. It would be interesting, however, to investigate if this camera network could be used to track individual vehicles across the city. This project will investigate the development of a tracking system that will identify cars in traffic camera images and track these cars on a map as they cross the city.

  • Email Connections (BSc/MSc): www.datahug.com are an exciting Irish technology start-up that allow companies exploit the social graph contained within email records. An interesting addition to a service such as that offered by Data Hug would be to augment links between people who exchange emails with word clouds featuring the keywords that they typically exchange emails about. In this way connections could be categorised based on topics/types. This project will build a system to do this.

  • Tell It Once (MSc): Ireland has a problem that citizens do not have a unique personal identifier that is used in our interactions with government agencies etc. There are a range of reasons why this has turned out to be the case, including security issues, public resistance, privacy issues, practical problems and beaurocratic resistance. while many of these issues are valid there are many frustrations that arise through not having a universal ID system - for example repeatedly having to give name and address information to various agencies. This project will investigate the possibility of a citizen-led tell-it-once to which citizens could opt in in order to have their information stored and then control the access to this information that various agencies could have. This project will involve technical development of such a system and research into the feasibility of operating such a scheme.

  • Open Data Privacy Spectrum (MSc): The open data movement pushes for government agencies (an others) to make all national data available to the public for scrutiny and to form the basis of useful applications - see opendata.ie and the London Data Store for examples. While the open data movement offers incredible possibilities for better governance, economic development and greater public engagement it brings with it many issues surrounding privacy and security - what data is it okay to make public? who decides what data is made public? does the format in which data is made available affect its sensitivity? what control does a citizen have over their own data? This project will attempt to develop a spectrum of data types of increasing sensitivity so that data owners could easily determine which datasets, data aggregates or data subsets it is reasonable to make public.

  • Future Recommender Systems (MSc): A recent AAAI AI Magazine article (The Big Promise of Recommender Systems, Matin et al) lamented the fact that the current batch of recommender systems worked on behalf of organisations rather than users - for example the amazon.com recommender system is working for Amazon not for you. This project will attempt to create a cultural interest recommender system that crosses books, theatre, music and cinema. By using incoming social network feeds a profile of a user could be created and this profile could be used to select items of interest from suitable cultural sources - e.g. the Ticket from the Irish Times.

  • Dail Monitor (BSc): All Dail proceedings are now broadcast on Oireachtas TV. This offers an interesting opportunity to monitor the attendance of TDs using the Oireachtas TV video feed and image processing techniques. This would act as an interesting data feed from which constituents could monitor the Dail attendance of their representatives.

  • Biography Assistant (BSc): Reading the modern celebrity biography is an interesting experience as the reader constantly find themselves turning to online media sources to seek material (e.g. videos and pictures) to augment their reading experience. This project would develop a tablet application that would automatically retrieve and suggest online media artefacts based on the page of a celebrity biography ebook a user is reading.

  • MCFC Analytics (BSc/MSc): Manchester City Football club have recently made freely available online a large collection of sabremetric data (http://www.mcfc.co.uk/home/the%20club/mcfc%20analytics) covering every "on the ball" event from the 2011 - 2012 Premier League sesaon. A larger dataset is also available on request if the requestor has a particularly compelling idea for which to use it. This project will leverage this dataset to create a novel data driven application that could be used by sports fans of professionals.

  • Activity Monitor (BSc/MSc): What do we do at work all day? This project will seek to answer this question by building a system that records the activites a worker is performing over the course of their work day. This might involve, amongst other things, using computer vision techniques to recognise broad activities (e.g. using the phone, using a PC, talking to colleagues), software monitoring to determine which software applications a worker is using and pupillometry based techniques to determine the level of cognitive load a worker is experiencing. All of this information will be gathered together into easy to consume reports on activity durations etc.

Augmented Reality (AR)

  • AR Scalextric (BSc): Anyone who has played Scalextrix will remember that one of the real problems with it is that the cars keep falling off the track and it is a real pain to put them back on again. However, there is something great about physically building the track to play on. This project will develop an augmented reality solution in which virtual cars will race around a real Scalextrix track viewed through a Head Mounted Display. There will also be many opportunities to add other augmentations such as realistic crowds, and to provide racing experiences that wouldn't be possible with physical Scalextrix cars - jumping through rings of fire for example!

  • 3D AR (BSc): In 2008 - 2009 a very successful final year project looked at using a real robotic forklift to manipulate virtual crates. A very interesting extension to this project would be to view it in true 3D using multiple cameras and displays. This should make interactions with the system more intuitive as users will be better able to appreciate the positions of virtual objects with relation to real ones. An image of the project is shown to the right for illustration.

  • AR Baseball Replays (BSc): Reams of sports data are recorded on most professional sports. Baseball, in particular, has a long association with data and is particularly suited to data capture due to the restart nature of the game. The full details of a modern baseball game are publicly available immediately after a game (e.g. http://www.baseball-reference.com/boxes/DET/DET201204050.shtml). This project will seek to create an augmented reality based reproduction of a baseball game from the publicly available play-by-play data. An existing system that has done something very similar is described in "Visual Enhancement for Sports Entertainment by Vision-Based Augmented Reality", Y. Uematsu & H. Saito, Advances in Human-Computer Interaction, 2008.

Other

  • Research Student Management iPad App (BSc/MSc): Research supervisors need to keep track o the activities of multiple research students all working on different topics and at different stages of their research degrees. Events such as transfer exams, annual evaluations and thesis submission need to be kept track of along with open research questions, answers to past research questions, paper submission deadlines and experiment deadlines. This project will develop an iPad app that supervisiors can use to easily keep track of the progress of their students.

Past Projects (maybe you have an idea for an extension to one of these...)

  • Create a Sandbot Robot Using a Lego NXT Mindstorm (BSc): Building robots that can navigate uneven terrain such as sand is difficult using wheeled approaches as wheels, just like in cars, become stuck in soft sand. One solution to this problem that has emerged is the Sandbot which uses semicircular legs to walk across uneven surfaces (more information on the Sadnbot is available here. This project will look at building a replica of the Sandbot using a Lego Mindstorm NXT kit and testing it on various terrains.

  • Weights of Evidence for General Classification: In many financial classification applications, rather than using raw attribute values, weights of evidence values are used as classifier inputs. This generally improves classifier performance. This project would consider whether this behaviour can be seen in general classification problems, rather than simply in financial ones.

  • Grangegorman AR: Grangegorman VR is a system developed by Hugh McAtamney to allow people view a virtual reality (VR) representation of the planned new DIT campus at Grangegorman. Grangegoman AR will be an augmented reality (AR) based application which will allow users view the real Grangegorman environment augmented with virtual representations of planned new buildings. The environment would be viewed through a telescope like device similar to those seen at tourist spots all over the world.

  • Brainstorming Tool: A tool to aid in the brainstorming process by recording the ideas generated by participants and displaying these through a projector. This would involve a speech recognition engine which would recognise ideas spoken to it, convert this speech to text and display all ideas in a readable format.

  • Wiimote-Based Interface for a Brainstorming Tool: This project would attempt to use a Nintendo Wiimote pointing device to manipulate idea bubbles created during a brainstorming session. Ideas would be projected in large scale onto a wall and the Wiimote could be used to move them around, rearrange them to the front or back etc. This porject would involve a large amount of work in explaring the Wiimote interface.

  • Mobile Phone Based In-Class Voting System: In class student voting has been shown to be a useful teaching tool. However, required hardware can be expensive making it unattainable for many institutions. This project would develop an in class student voting system in which students use their mobile phones. Most problably this would involve mobile phones connecting to a server via bluetooth as expecting students to pay for calls/SMS messages is unrealisitc.

  • AR Route System: Rather than having signposts at the side of the road which direct everybody, signposts could be left blank and be augmented with the information required by a particular driver which would be viewed through an AR headset. For example, rather than seeing a signpost on approaching a roundabout which shows where all of the exits go, the driver would only see the exit they need to take highlighted.

  • MySpace for Academics: Allow academic researchers post information about published papers, research interests etc in the fashion of mySpace using networks, linking etc. Especially link the networking aspects of mySpace with publication author lists.

  • Applications Using the Google Maps API: Any applications that make use of the Google Maps API. For example, a surf site which allows users mark good surf beaches and post surf reports to these via SMS messages. Good surf would be easily located on the Google map.

  • Runner's Training Centre: Create a community web site for people doing running training. This would include sections such as time tables, training plans, links to other runners and maps of routes. For the map section Google maps could be used - an interesting application in this area is the Gmaps Pedometer which can be seen here.

  • Snooker Visualiser When watching snooker on TV it is often hard to see whether a particular shot is possible because the camera man cannot disturb the player to get the correct viewing angle. This project suggests that a computer vision system is used to identify all of the balls etc from a top down view of a snooker table and based on this a 3D model of the current situation is built. This model would then allow the table to be viewed from any viewpoint.

  • Jigsaw Visualiser As part of a larger project to build a robotic jigsaw system for disabled children, this system would build a model of a block based jigsaw using images of a scene. The system would need to be capable of identifying the various pieces in the jigsaw scene and determining which pieces are connected to each other. It would also be very nice if the system could determine whether the jigsaw had been completed or not.

  • Neural Net/Genetic Algorithm Based Blackjack Player: Using genetic programming and neural nets how successful a Blackjack player could be evolved?

  • Robot Receptionist: In this system a small mobile robot would be positioned at the entrance to a building (Kevin St. in this case) beside an information kiosk. A user would select from the kiosk the person/room they wanted to visit and the robot would lead them to that location. This project would involve developing a navigation system for the robot, the kisok application and a communication mechanism between the kiosk and the robot itself. For those interested there is a tenuous Star Wars link to this project too!

  • Plane Seating Simulation: Develop a simulation to test out different plane seating schemes to see wich would save most time in plane turn-around. There are a number of existing schemes and these could be tested. Furthermoe could you come up with a new scheme? There is an article from wired.com which discusses the issues involved here.