|
|
 |
Under & Postgraduate (Taught) Projects
For several years, the Centre for Digital Video Processing
has been running a programme of support for undergrad & taught postgrad projects in
the Schools of Computing and of Electronic Engineering.
Projects under this programme receive support in the form of access
to resources and expertise from the CDVP staff.
Back to Centre member list, or
see all past members.
| Year 2011/2012 |
| Proposals |
|
| |
Video analysis from a UAV drone [Taken by Mike Clarke] -
We have access to UAV drones with built in stability and video camera transmitted back to an iPhone/Android handset, described in http://www.expansys.ie/parrot-ar-drone-green-outdoor-hull-indoor-hull-battery-charger-194641/. This project is about analysing the video streamed from the drone, in order to locate something known. So take a large open area and send the drone out to 'find' something. In an open indoor lab, it might be to find where I left my phone or keys. Its a bit of overkill for an application like that but its a proof of concept and would involve route planning, route tracking, etc.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Who is that student ? -
Using a smartphone with video, an app grabs each frame and does face extraction, uploads to a backend server which uses the API from face.com to match against a known DB of student faces. Useful for a lecturer in class to identify who is that student.
Supervisor: Prof. Alan Smeaton / Dr. Cathal Gurrin
SCHOOL OF COMPUTING |
|
|
| |
Celebrity faces -
Download multiple photos for known celebrities, using IMDB for example. Run face extraction, then upload to face.com and measure the 'coherency' of each celeb, i.e. how much his/her faces differ, or in other words how much does a celeb's face change.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Viral videos -
We are constantly collecting usage data on a set of c.2,000 vidoes released on YouTube from a known set of channels. Over time, some of these will go viral. This project is to detect the characteristics that can predict what makes a video go viral and there are several aspects that can be explored:
- some videos are syndicated across multiple channels - can we combine these into one 'release' by crawling the same video from multiple sources;
- our definition of 'viral' is 15% day-on-day cumulative growth - could we have levels of 'virality' by statifying into different levels;
- can we detect abnormalities in the usage statistics, caused by paid SEO downloads causing false peaks, for example;
- can we segment videos by geography or demographics and analyse by these;
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Video montage of an event [Taken by Zhang Biao (BUPT)] -
Take an event like a concert, where multiple videos from multiple angles, taken by multiple people, are available. This project analyses these videos to determine both quality, and synchronisation points, and selects extracts from videos to be included in a montage. THink of the TV series "24", but generated automatically.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Bump ! -
Detecting potholes in city streets and deciding when they need to be fixed, is a labour-intensive task. This project takes a participative sensing approach to this. An app running on an iPhone and/or Android, is started when the user starts a car journey. The app records the GPS and the accelerometer records motion, and bumps. When the journey is completed the user signals this and the app processes the movement data to detect what are most likely to be potholes and offers the user the option to have these, their 'intensity' and location, perhaps superimposed on a map, uploaded to the community database. At the back end, a database of likely potholes is maintained and a user can view their own suggested potholes while the system manager (the local city council for example) can overview everybody's upload and when there are enough reportings on a given potholes existing then the council can decide to repair it.
Supervisor: Dr. Cathal Gurrin / Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Recommender system using eye-tracking on iPad 2 [Taken by Hu Feiyan (BUPT)] -
Using an IPAD-2 with a front-facing camera, use this to detect the face of the viewer and whether that view is actually watching the iPAD and based on that use it as input to a recommender system.
Supervisor: Dr. Cathal Gurrin / Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Isochronic map [Taken by Matthew Quinlan and Julieann Flemming] -
Given a location in Dublin, calculate a heatmap of the average time taken to travel to each other part of the city by public transport, or by car, known as an isochronic map and this gives an indication of how desirable an area might be to live in.
Supervisor: Dr. Cathal Gurrin / Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
|
*Intellectual Property arising from these projects
The School of Computing has
formulated a policy on
the distribution of intellectual property arising
from student projects (such as
these). This is based on an agreed a priori disbursement agreement
between lecturer and student in the cases where the lecturer has a non-trivial involvement in the
project and where the lecturer wants to put in place such an agreement and the student agrees.
In the case of projects from these there is a considerable amount of background knowledge provided
and a significant amount of direction and facilitation by myself and other members of CLARITY: Centre for Sensor Web Technologies. Consequently we shall require students to sign over intellectual property
development rights for such projects before the commencement of the project.
(Student Agreement Form - DOC; 26K)
|
|
| Year 2010/2011 |
| Proposals |
|
| |
Kinect interface to information browsing [Taken by Dave Staceym (MSSF)] -
Microsoft has recently released Kinect, a gesture-based interface to XBox 360. This can be hacked to allow it to be used as an interface device and so this opens the possibility of gesture-based interaction to ... well to almost anything you can think of. For video navigation, for example, think Minority Report. Colleagues in NUIM will be running a taught masters project similar to MSSF/MSE) to use an XBox 360 Kinect (hacked device) for movement rehabilitation following a stroke. This project proposal is to use a hacked Kinect device as an interface to rapid gesture-based browsing of an image collection, a collection of thousands or tens of thousands of images form a visual lifelog, from a Flickr dataset, or of shots from a video archive. The challenges in the project will be to define the capabilities of Kinect, to determine the best way to incorporate it into a browsing activity, to determine the most appropriate (i.e. effective, accurate, comfortable) gestures to be used, and then to combine all these together into a system which can be evaluated.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
MusiColour: How colour and music can trigger synesthesia [Taken by Evan White] -
Recent advances had shown how people react to colour and to music, in similar ways and how certain colours and colour combinations are regarded by us as harmonious, just as music note combinations are harmonious in music. This project will investigate biometric responses to harmonious presentations of colour, music and colour/music combinations. Subjects will be monitored in a lab environment by sensing their heart rate, breathing rate, galvanic skin response, movement, EEGs and an eye-tracker will record what parts of a color display screen they are actually watching as the synchronised music is played. The generated dataset will then be analysed to see if we can identify correlations between colours, music and human responses.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Sensor-based computer game for skill acquisition in Gaelic football -
Last year we developed a sensor-based computer game that was used by 750 children in the 2010 DCU SUmmer Camp. It is based on an image projected on a wall from a datashow, and a camera fixed on that image, able to detect when an object (a ball) hits a projected image (a target). This has enormous potential for skill acquisition in a multitude of sports e.g., Gaelic football, hurling, soccer, basketball, as well as for physical education programs in primary schools. Our Summer camp protoype can be designed more specifically, tailoring the game for implementation in a team training setting or a P.E. class and I'm looking for one Computing student to work with a student in Engineering and two students in Sports Science to work as a team to re-design the game, test it on a cohort of sportspeople, assess energy expenditure and skill levels when using the game, and document all this. Clearly this is a very collaborative project and I've worked out the scheduling and deliverable needs of sports science and engineering FYPs, but if you're into sport, and want to get your project done a bit earlier than the Computing schedule suggests, then this could be for you.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Analysis of domestic energy usage [Taken by Marc Healy] -
Within CLARITY we have 22 homes gathering data on electricity usage, 24x7, for the last 12 months. This data is stored on a conventional SQL database with an API developed and working. The project is to carry out an analysis of usage patterns, activity levels, and appliance usage and to develop visualisations for this enormous database of data readings.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
A Cartoon version of a meeting [Taken by Cian Scolard] -
We attend meetings and document these with minutes but they are very boring and dry. This project is to take the minutes of meetings and transcode them into the input language used to generate cartoon movies in the xtranormal.com website. That means taking the 'conversation' that takes place in a meeting, adding speaker choice, intonation, gestures, switching camera angles, etc., ad then generating the cartoon version of the meeting. This project may look like a bit of a giggle but to do it properly requires some extensive langauge analysis and language processing, some creativity to be able to determine all those artistic things like camera angles, design of avatars/cartoon characters, and how to pace the delivery of a dialogue. If it is done properly, this project would be a real eye-catcher.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Analysis of World Cup Twitter Dataset [Taken by Final Year project from NEU] -
We have gathered a dataset of over 40 million tweets from the 64 games of the World Cup. Each tweet has information on the sender, and from that we can get further information such as that sender's activity levels, number of followers, number followed, etc. The project is to analyse the tweeters in this dataset and determine the most authoritative, the most important tweets, and the 'structure' of the graph of world cup tweets. If you want to work with big data and you're into social networks and Twitter, then this project should attract you.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Track the Whereabouts of the security bins on campus [Taken by Neil O'Donnell] -
Around the staff corridors in DCU you'll often see large wheelie bins with their lids locked and a slot, like a letterbox, on the top. These re for confidential papers, meant to be taken away and shredded. Technology is available now to sense how full each bin is and to notify when it needs to be collected - this is based on an ultrasound sensor in the bin, inside the lid, pointing down, and a GSM connection. However, apart from the lock on the top of the lid, which anyone with a screwdriver can open, there's little other security on the bins. This project is to see how much value would be added by adding an accelerometer and a digital compass to the bins. These sensors would be able to track when bins are moved, or opened, and by using the accelerometer and compass, possibly even track where the bins have been moved to, between them being installed, and emptied. A conventional iPhone or Android would be used to gather the sensor data, with a few trial runs of the wheelie bins, and ten data analysis would tell us whether the new sensors add value and could be used. If you like getting your hands dirty (no "bin" pun intended here) with data gathering and analysis, you should be interested in this one.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
What makes the city pulse ? [Taken by Yang Li] -
We have access to an archive of rainfall radar and other metrological data from Met Éireann covering the Dublin city area; we also have access to an archive of information on usage of the Dublin rental bikes scheme; we have access to information on events within Dublin, and their location; we have access to traffic flow information. This project is to explore the correlations among all this data. Is traffic really heavier when it rains ? Do people use the Dublin bikes in good weather or in bad weather, or both ? Does bike usage peak when there are events in town ? Many other questions can be answered using this data.
Supervisors: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Visualising home energy usage on a large Multi-Touch Wall [Taken by Jude Higgins] -
Multi-touch Wall is a 3m X 1m large public display with multi-touch capability, here in CLARITY Centre, DCU (see what it looks like). We are interested in utilising this large display to visualise people's home electricity usage. Home electricity usage data from 20+ households is contained in a database, and the system in concern will take the data and display on the Multi-touch Wall in an informative and interesting way. The successful student would need to be good at programming (especially C++ and database)
Supervisors: Dr. Hyowon Lee
SCHOOL OF COMPUTING |
|
|
|
*Intellectual Property arising from these projects
The School of Computing has
formulated a policy on
the distribution of intellectual property arising
from student projects (such as
these). This is based on an agreed a priori disbursement agreement
between lecturer and student in the cases where the lecturer has a non-trivial involvement in the
project and where the lecturer wants to put in place such an agreement and the student agrees.
In the case of projects from these there is a considerable amount of background knowledge provided
and a significant amount of direction and facilitation by myself and other members of CLARITY: Centre for Sensor Web Technologies. Consequently we shall require students to sign over intellectual property
development rights for such projects before the commencement of the project.
(Student Agreement Form - DOC; 26K)
|
|
| Year 2009/2010 |
| Proposals |
|
| |
Determining Project Importance by Reference Tracing [CA3 level] -
An important part of scientific work is writing papers - for journals
and for conferences/workshops - which describe a piece of work
complete. These papers normally have citations to other papers
published previously and the more a published is cited by other
papers, the more important it is regarded as. TRECVid is an annual
international collaboration activity involving almost 400 researchers
worldwide, founded in 2001 by Alan Smeaton. TRECVid leads to many
papers published each year, but we do not track these papers, and we
do not track the importance of TRECVid and its papers in its
scientific field. This project is to analyse the scientific
literature using a number of well-known and freely available online
tools, to compute the importance of TRECVid. Rather than just being
desk research and slogging through the outputs of search engines, the
project will require the student(s) to write scripts to iteratively
perform searches on different libraries, parse the output, and launch
subsequent searches. An ideal implementation language/platform for
this would be Ruby on Rails.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Fitness and Agility Based Wall-Touch Game [CA4 level: Taken by Philip Fox] -
This open-ended and evolutional project allows people to practice and improve their limb-to-eye coordination,
motor-neuron response level, agility and fitness levels. An image is projected onto a wall by a projector. A
participant, wearing special gloves, attempts to touch this image and if successful then the projector's beam
is intercepted. A reflective device, attached to the back of each glove, reflects some of the projected image's
light back to a modified web-cam, which senses this reflected light and signals the computer to which it is
connected. The computer signals the projector to remove the image and to project another image to a different
position on the wall. Applications of the finished system are wide reaching and may include reaction/agility
training for sporting diciplines.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
SenseCam montage [CA4 level] -
SenseCam is a wearable camera for recording a person's life and we
have a lot of expertise in developing software for it in CLARITY. www.cdvp.dcu.ie/SenseCam
has details. This project would involve taking all of the images
which make up a particular 'event' in a wearer's life (giving a
lecture, driving a car, waiting for a bus, sitting at a desk, eating a
meal), an event where the view stays reasonably stable, and
aggregating all those images into a single montage. Think of this as
an application for http://photosynth.net but working with poorer
quality images where you need to decide not only where the images
overlap, but which are the better quality images. The project is
challenging and requires an interest in image processing. Much of the
components needed for this are already in place in some form within
CLARITY so its not building from scratch.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Exploiting Multi-user Shared document creation on a Video Wall [CA4 level: Taken by Daniel Morrow] -
The MultiTouch Wall is a state-of-the-art modular LCD cell-based
interactive display, composed of 5 LCD boxes each containing a near-
infrared camera inside to detect multiple finger touches and movement
on LCD surfaces. Altogether it makes up a 3m X 1m wall video wall
with 5,400 X 1,920 pixel resolution which is multi-user touch-operated
and which provides the biggest computer screen available to us. This
project involves designing and building a multiuser multimedia
document creation application to run on the video wall. The project is
quite open-ended and the final spec would be 'evolved' between student
and supervisor.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Spatial game playing with AV feedback [CA4 level: Taken by Marek Steplewski] -
We're well used to playing board games on an 8x8 grid with pieces -
chechers and chess for example. What about a game on a much larger 8x8
grid where the cells are 1m x 1m and the pieces are ... people. Take 2
teams of 2 or 3 people, all placed in one cell of an 8x8 grid marked
out on a floor, and develop a strategy-based board game with turn-
taking, one team moves a person/piece then the other does, etc., until
some strategic aim is achieved ... like the king is taken (chess) or
all the opponent's pieces are removed. Using the CLARITY installation
of Ubisense in some of the CLARITY space, this can be implemented with
the players each carrying a ubisense tag to give their location (i.e.
which cell) in real time, with visual feedback via the 4 TV screens in
the corridor outside L131. The project would involve design and
refinement of the game, and implementation with inputs being Ubisense
stream data and output being visual display on the screens.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Real-time Synchronous Collaborative Search Engine [CA4 level: Taken by Jian Zhang] -
This project will produce a multi-modal, multi-user, collaborative search engine that will allow multiple people to search together in a group in real
time in order to satisfy the same information need. At present, if multiple people want to search together on the web, the only real solution is for
each user to open up a separate Google window and search together in parallel. As the group are searching to satisfy the same information need, this will
result in much duplication of effort and redundancy across the searchers. What is needed is a search engine that is implicitly aware that there are many
people searching at the same time, under this assumption the search engine can then coordinate the search task across the users so that the actions of one
user in the group affects all other members of the collaborating group. One example of such a system-mediated coordination would be the removal of links
from a search results page given that another user has already viewed the page.
Furthermore in order to increase awareness across collaborative searchers the search engine should provide a user with a view of their collaborators
activities, this will allow the group as a whole to coordinate the search task. This view could be realised in the form of an integrated toolbar which
would display all significant actions by other members of a group or allow users to chat to one another.
The system would be web-based and should use the Google API as the underlying search-engine. The system should be available to people either on a desktop
machine or on a mobile computing platform such as an Apple iPhone to allow for fully multi-modal collaborative search.
Supervisors: Dr. Colum Foley and Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
|
*Intellectual Property arising from these projects
The School of Computing has
formulated a policy on
the distribution of intellectual property arising
from student projects (such as
these). This is based on an agreed a priori disbursement agreement
between lecturer and student in the cases where the lecturer has a non-trivial involvement in the
project and where the lecturer wants to put in place such an agreement and the student agrees.
In the case of projects from these there is a considerable amount of background knowledge provided
and a significant amount of direction and facilitation by myself and other members of the Centre
for Digital Video Processing. Consequently we shall require students to sign over intellectual property
development rights for such projects before the commencement of the project.
(Student Agreement Form - DOC; 26K)
|
|
| Year 2008/9 |
| Proposals |
|
| |
Synched playback of virtual golf swings [CA4 level: COMPLETED BY INTERN STUDENT] -
we have access to a dataset
of 3D point recordings of golf swings. 10 recorded swings from each of
30 people, each swing lasting about 10 seconds with 46 data points
sampled at 250 Hz, 250 samples per sec. That's 10 x 30 x 46 x 250 data
points, in 3D, corresponding to players hitting a 5 iron shot,
straight. Each swing is marked up into 8 different phases
corresponding to addressing the ball, backswing, downswing to where
the ball is struck, etc. The thing is, that when we replay 2 swings,
even from the same person, we have different durations of each phase;
some people have a slow backswing and a fast downswing, and others the
reverse, and when we compare these two swings they take the same
amount of time, but are out of sync. This project is to visualise 2
swings from the same or different people, using a 3D wire model but to
join the 46 data points (which we've already done), but to do so on a
way where there the times taken for each phase are aligned, and if
necessary normalised so they are the same length. The net result will
be a visualisation of 2 golf swings in a way that allows easy
comparison.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
|
| Taken |
| |
User-specific location-sensitive audio tour guide [CA4: Eoin Hurrell] -
This project will produce a system using Ubisense tags to provide
location-specific audio content to known users. The project will allow
users to follow a virtual tour providing audio commentary on items
located in the real world. Each item has a predetermined area of
effect in which a user hears audio content related to the item. A
virtual tour designed with a flowing theme is implemented, but should
the user stray from the path to look at items in an order of their own
preference the system will compensate.
The system will be designed with an environment such as an art gallery
with a high turnover of items or frequent exhibitions. A user
registers, is given wireless headphones and an Ubisense tag. When they
come within range of an exhibit audio commentary describes the history
of the piece,
in content of the show.
Recommendations will be given based on the user's preferences,
demonstrated in prior visits. Each visit will be recorded and the
user's path for the visit will be visible from a website.
This website interface will also provide the ability for a user to
review the exhibits they viewed and missed on the path they took.
Expanded with a group in mind the system will be able to generate a
single page detailing the group's experience, such that school classes
would have a visual diary of their journey and individuals a digital
memento.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Fusion of multiple location-aware sensor networks, namely camera,
UbiSense and launcher networks. (Dr. No's big red button) [CA4: Jim O'Shea] -
This project will be a showcase of the fusion of multiple sensor
technologies to provide a multi-functional tracking and sentry
network. The system will be highly extensible as new devices can be
added and interact with existing devices via the central controlling
platform. The project's main application is as a security system where a single
target can be tracked and monitored by a network of devices. It would
be especially useful in a crowded location such as an airport or train
station where a tagged target would be automatically tracked without
getting lost in a crowd.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Using FusionBrain to integrate multiple sensors for room control [CA4: Jeff Finn] -
This project is to prove that Ambient Assisted living can help people
in their home, by building patterns of how people interact with their
living environment during their daily life. This data would be
collected by using discrete sensors to monitor their home.
The patterns developed would be built from monitoring different things
from the persons living area, like the fridge and microwave and other
appliances which would help determine what the person is doing in
their home at a certain time. From this data we would be able to make
an intelligent guess at what time a person does certain tasks at. When
this data is collected over an extended period of time, patterns
should start to emerge. Some of these patterns could deal with at what
time they get up at each day, how long it takes until they turn on the
kettle after they get up.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Navigating virtual maze on video wall [CA4 level: Marc Gowing] -
Within the CLARITY/CDVP group we will shortly install a touch
sensitive LCD video wall. The enviroment will also have an
installation of Ubisense, meaning the location of people within the
room can be tracked in real time. This project is to build a 4-person
tag game for the room where the inputs are the 4 people's Ubisense
tags, the task is to navigate a (virtual) maze in the room, and the
output, and some inputs like setup and winner, are via the video wall.
This will require some creative thinking on gaming strategies, as well
as integrating ubisense and video wall software. The target audience
for this will ultimately be obese children, and a slimmed down version
of this project could run in a larger ubisense environment such as the
tennis court in Tennis Ireland where we have UbiSense running.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
(Semi-)Automatically Adjust my Photos [Guangyu Wu] -
We have available to us several implementations of region
segmentation of still images where areas of an image which are the
same in terms of colour and texture can be identified. So all the
grass area or the sky area or the shrubbery in a landscape, for
example, can be identified and isolated or segmented from the rest of
the image. We have applied this segmentation to a collection of about
100,000 images and indexed each region by its colour, texture, and
shape. In this project we can take a query image and run our
segmentation on it, and ask a user to choose one or more region(s)
which are to be replaced. This could correspond to, say, not liking
the tarmac surface in a picture of a road, or the type of shrubbery
in a landscape. For selected regions in the query image, we could
locate the most similar regions in the database of regions based on
colour, texture, and shape, and then replace the query image region
with one from the database, with some Poisson smoothing around the
edges. This effectively gives a tool for re-compositing your personal
photos.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Colour-coded Google Map based on type of accommodation [CA4 level: Brendan Cleary] -
the Private Residences Tenancy Board is a statutory board charged with mediating
between tenants and landlords in cases of conflict. Landlords are
supposed to register their premises with the PRTB and the PRTB website
publishes these details http://www.prtb.ie/pubregister.htm. This
project is to scrape this data regularly and overlay it on some
mapping tool (Google Earth, Google Maps, whatever), showing a
breakdown of accommodation types in urban areas. A gazetteer would be
used to turn roadnames into GPS coordinates and the regular scraping
would show long term trends in accommodation type shifts. A company
like DAFT.ie might ultimately be interested in the output of this
project.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
iPhone sports event tracker [CA4 level: Daniel McElligott] -
there are tools on the market to allow a
viewer to record the events that happen in a sporting event - such as
goals, points, frees, wides, sidelines, fouls, penalties etc. in a GAA
match. As part of one of our CLARITY projects (see news item on GAA
refs at http://www.clarity-centre.org/news) this CA4 project is to
develop a modifiable sports tracker program for an iPhone/iTouch. For
each sports event type (GAA, soccer, rugby, tennis) there will be a
pre-defined template of event types and a user will watch an event and
tap buttons corresponding to sports events as they happen, and these
will be time-coded. At any point a user should be able to view
statistics on the match to date - so it is a commentator's dream tool
- and the data should be uploaded to a server at the end of the
event. We anticipate developing this to be used in conjunction with
our CLARITY recording of the forthcoming GAA Football National League
campaign, particularly tracking the Dublin team ... God help us ;-)
That means this tool would be used in conjunction with our video,
movement, breathing, etc. recording of those matches, so there could
be a requirement to go to Parnell Park for a match !
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING
|
|
|
| Year 2007/8 |
| Proposals |
|
| |
[EE] [ICE] [EEI] [DME]
Content-based classification of news web-page images -
Usually a news web-page will contain a number of images along with the text of the news story being
reported. Whilst it is effectively impossible to develop a machine learning approach that can understand
the content of these images (i.e. why, what, when, who, where, is depicted), different types of images do
share certain characteristics that can be recognised. For example, it has been suggested that if the news
story is about a particular person (e.g. Bertie Ahern) or place, then the images tend be well-lit, well framed,
correctly focused, with sharp colour and edges. On the other hand, if the news story is about a more abstract
concept (e.g. trade union talks) then the images tend not to exhibit these characteristics. The goal of this
project is to investigate this hypothesis and on this basis design an approach for distinguishing between
these kinds of images. Project steps will involve: (1) Gathering training/testing sets & classifying the news
images manually; (2) extracting a variety of image features from the image (e.g. colour depth/saturation, brightness,
how centred in the frame the main subject is); (3) using an "off the shelf" classifier (e.g. Support Vector Machine) to
determine the feasibility of such an approach.
Skills: C/C++, Matlab, Digital signal processing
Supervisor: Dr. Noel O'Connor
SCHOOL OF ELECTRONIC ENGINEERING |
|
|
| |
[EE] [ICE] [EEI] [DME]
Comic-like layouts of SenseCam images for visual summaries -
Visual summaries are used to allow viewers to skim through large amounts of image or video content. For
video, for example, much research has focused on identifying important events automatically (e.g. goals, close
misses, sending offs in sports content, dialogues/action/montage scenes in movies) and using these as the basis
for the summaries. This type of approach allows a short highlight video clip to be constructed that can convey
the most relevant information to the viewer. However, such an approach has significant drawbacks if there is a
very large amount of content to be summarised. For example, video playback (even accelerated playback) is not
suitable in the case of a wearable camera such as Microsoft's SenseCam (http://research.microsoft.com/sendev/project_sensecam.aspx),
which takes thousands of photographs each day. A more suitable approach in this case is to display multiple representative
images to the user at the same time, but with the placement, size and ordering of these images determined based on
their relative importance to the viewer. The result is a page of images that looks like a page from a comic-book.
The aim of this project is to investigate ways of determining the optimal spatial placement of a day/week/month's
worth of SenseCam images given the most important images and their relative importance. This will require a
geometric optimisation approach based on predefined layout templates.
Skills: Matlab, Java, Digital signal processing
Supervisor: Dr. Noel O'Connor
SCHOOL OF ELECTRONIC ENGINEERING |
|
|
| |
[EE] [ICE] [EEI] [DME]
Prototyping a low-cost visual sensing platform -
Wireless sensor networks are currently a hot topic in the research community. These are networks of low-cost,
long-life sensing devices that can be deployed over large areas for applications from security to environmental
sensing. To date, wireless sensor nodes (i.e. the individual sensing platforms in the network) have tended to be
very simple devices consisting of simple chemical/physical sensors, an RF radio and circuitry to glue all this
together (see the Tyndall "motes" - http://www.tyndall.ie/research/mai-group/25cube_mai.html). However, in the
future, there's no reason why such sensor nodes could not contain more sophisticated sensors, such as low-cost,
low-resolution visible/infrared imagers, mems microphones, and a variety of positional sensors. Such a sensing
platform could find applications in areas as diverse as handheld imaging to smart camera networks. However, existing
wireless sensor platforms, like the "motes", are not up to supporting this level of sensing yet. The objective of
this project is to design a prototype of a visual sensing platform that incorporates a variety of sensors, the output
of which can be communicated to a "base station" (a PC over a wired link, in this case), with a view to understanding
the design and power constraints involved.
Skills: Digital signal processing
Supervisor: Dr. Noel O'Connor
SCHOOL OF ELECTRONIC ENGINEERING |
|
|
|
| Taken |
| |
UbiSense dance partner [2x MSE students] -
UbiSense is a location-tracking sensor network which is to be
installed in several labs in the Computing building in Autumn 2007.
It operates by tracking a tag, about the size of a small wallet, in 3-
D, in real time. This project is to use UbiSense to gather location
information on the 4 limbs and torso of a person (by having the
person wear 5 tags strapped to their arms, legs, and torso), and to
generate a virtual dance partner, in real time. So the wearer
"dances" alone, and a virtual partner is created to follow their
movements, as would a dance partner. The virtual dance partner is
then visualised, also in real time, as a projected video,
superimposed on a video of the real partner (you!). The applications
would be learning dance steps, without the embarrassment of having a
dance partner.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
RFID activity tracking -
In work completed in Summer 2007 by Hristo Novatchkov, he built a
mobile RFID reader which logs interaction with RFID tags. The reader
is connected to a Gumstix computer and has a WLAN connection and the
whole package can be built into a wearable device. The application we
envisaged is that a home - kitchen, living room, bathroom, etc.-
would have RFID tags fixed to devices like kettle, oven, press where
saucepans are, TV remote control, light switches, etc. During daily
activities, a wearer's interaction with these devices would be logged
and uploaded to a server via WLAN, and a backend process would
analyse these interactions to determine what the overall task is ...
I am making tea hence I interact with the kettle, fridge, sink, sugar
press, etc., with certain constraints on the ordering (sink before
kettle for example. Details of the hardware built - and working - can
be seen in the slideshow at the end of this page
http://www.odcsss.com/students/4/hristo-novatchkov/.
This project is to
write the backend software which can analyse RFID tag readings, in
real-time, and determine which, from a pre-defined set of activities,
is being followed.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Harry Potter's Talking Portraits -
You've (probably) seen one of the Harry Potter movies where the
portraits on the staircase aren't still images but clips of people
which move as if they are alive. You've also probably got still
pictures of people on your computer or LCD-based photo frame, but
wouldn't it be nice to add some life to these. The idea here is to
take a (short) video clip of a person, head-and-shoulders shot, and
have them move, smile, frown, laugh, etc. Then take each frame of
the video and compare it to each other one and fine points in the
video where the person is in exactly the same position and the
preceding frame(s) are either the same or very similar. Such points
mark places where the video playback could jump, seamlessly. With a
long enough video clip (60 seconds ?) and enough jump points back and
forward, you could play a never-ending video clip, with seamless
jumping ... not lust a loop but randomly jumping.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Automatically Mashing my iTunes Songs - Every Car you Chase -
Take all the songs on my iTunes and for each gather supplementary
information covering things like beat, lyrics, karaoke version, MIDI
version, anything that can be scraped from the web. Then analyse each
song to find points where it could jump to another, similar song.
When this is done, then pair together similar songs and generate an
automatic mash-up of them, just like the Snow Patrol/Police mash of
"Every Car You Chase". This allows truly personalised playback of my
songs, for me, at playback time, based on my songs .... truly unique !
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
1 or 2 x [CA4]
Cartoon Layout of the SenseCam Images -
This project is related to
the other project on Mosaicing of SenseCam images (see next project for a description of the SenseCam, segmenting a day into
"events", the set of images for an event, and the idea of a landmark
image). The Mosaicing project below is to take all images from a single event
and generate a visual mosaic while this project is to automatically
process each SenseCam image to detect the Region of Interest (RoI)
for each (using our CDVP salience detection) and then to crop each
image to just that RoI, and then to combine all/many/most of the
SenseCam images from an event into a comic book layout like this
http://www.savvyhobo.com/images/RiverRatettes.JPG. The net result
will be a comic-book layout of the important things which constitute
an "event" for the wearer of a SenseCam.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
2 x [MSSF]
What are the Possibilities for Securing Video Streaming - Media streaming to the home uses a rolling key for security
and there is a only small bandwidth channel for rolling key exchange. This transmission takes place to/from the IP
set top box, but this technology may be vulnerable for high security applications where even a rolling key may
be decipherable. Obvious applications are security, and military. The idea for this project is to investigate
various approaches for encryption within the MPEG-2 transport level, or perhaps encryption using RTSP, and
to evaluate and compare as necessary. An implementation to back up theory would be expected so this is a
hard project idea.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
|
| Year 2006/7 |
| Taken |
| |
Kenneth Farrell [CA4]
Online Map-based Blogging Tool -
The objective of this project is to create an online map based blogging tool.
The system will be aimed primarily at people who use GPS (Global Positioning System) devices and who want to keep a record of their travels.
The system will allow the user to load a series of GPS readings and will then sort the readings based the longitude and latitude.
It will then group the sorted readings together into events based on the location that the reading was recorded, with readings taken at the same location within a certain time frame grouped together.
Recurring events such as driving to work would be grouped into one aggregate event to avoid cluttering the map.
The system will also present the user with the option of adding annotations to the groups of events or individual events, and will
then add a marker to the map for each event. The user can then filter and view the events on the map based on location or time or both.
The system will also provide the ability to load photos to the diary.
The main part of the project will be implemented using the Google Maps API (Application Programming Interface).
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Adrian Bratten and Barry Tallon [CA4]
Navigating a world of Cartoon Geo-tagged photos on a handheld device - The GeoTaggr is a device that couples
with a camera and records GPS location and compass direction. See details at http://www.geotagger.co.uk/ Photosynth is
a Microsoft Research project which "takes a large collection of photos of a place or object, analyzes them for
similarities, and displays them in a reconstructed 3-Dimensional space". One of the pre-requisites is a collection of
geo-tagged photos. This project is to take an area, probably within campus, possibly the mall, and take literally
thousands of geo-tagged photos and store them in a database. Version 1 is to develop an application on a PDA with
a wireless LAN connection and a GPS device, to allow a user walk around the mall and to display the nearest photo
from the database, nearest in terms of position and orientation. We could process each photo to turn it into a cartoon-like
rendition for display on the PDA, and that would allow Version 2 to be built, which would be to display merged or
overlapped photos, just like Photosynth but with cartoon-like images from real photos.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Adam McCrossan and Aidan Nolan [CA4]
Scraping Rainfall Progression in Ireland from Radar Images -
The Met Eireann website displays a map of rainfall in Ireland,
sampled using radar every 30 minutes. See
http://www.met.ie/latest/rainfall_radar.asp?ani=n. Met Eireann also string a few of these
together to give a kind of animation, but the interface to this isn't
very intuitive (try viewing it and watching both the rain progess,
and the time ... its impossible). This project will do several
things ... firstly we will scrape the images from the Met Eireann
website on a regular basis and store these locally. Once we have an
archive of these we will process each individual image, and then
process sequences of images to determine the notion of a 'rain
shower' passing across the country where each shower is a cluster of
rainfall pixels, with dimensions of size (how widespread), intensity,
direction (N, S, E, W, etc.) and speed of progress. This will give us
a database of rainfall progression in Ireland. We can then let this
grow over time so this can grow into a reference for rainfall in
Ireland.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Graham Healy and Daniel Kersten [CA3]
Power Consumption for Audio Analysis on Wireless Sensor Network -
Within the CDVP we have identified 3 types of audio analysis which
can be used to detect "events" in a home, office, corridor, street,
etc. These are zero-crossing rate, FFT and volume envelope. We have
gathered and alaysed a large amount of ambient audio from these kinds
of environments and now we wish to implement this audio analysis
directly on a mote, a small, autonomous wireless computer which forms
an ad hoc network with other motes to give a wireless sensor network.
The audio analysis will be done by hardwiring it onto an FPGA, but
before we do this we would like to ascertain the power consumption
(in terms of energy from a battery) of the three kinds of audio
analysis, when implemented directly on the mote, as opposed to on an
FPGA. This project is to implement the three audio analysis
techniques directly on a mote device running TinyOS, and to use the
enprofiler tool to profile the energy consumption of each application.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Séan Carrick and David Caviston [CA4]
Real-Time Vehicle Tracking and Querying -
The CDVP have acquired two motor vehicle tracking devices which give
longitude and latitude and speed in real time over GPRS when the
vehicle to which it is fitted, is moving. These are installed in two
cars (one of which is mine). This data is sent to a server here in
DCU and stored on a SQL Server database. This project is to provide
a (Google) Map based interface to viewing single and multiple vehicle
routes and trips, plotted with colour coded speed ratings, with the
data drawn from the database. The second part of the project is to
take a standard coarse grained gazetteer (a lookup table which turns
Long/Lat values into placenames) and allow a web lookup of "where is
my car now?" -> it is in DCU; "where has it been today?" -> left
Malahide at 07:10 drove to Swords, onto M1, exit Santry, through
Santry Ave, Shanown Ave, to DCU, arrived 08:20 and is there still ...
", in other words use some simple natural language templates, and
values from the gazetteer to generate a text description of the route.
Supervisor: Prof. Alan Smeaton
SCHOOL OF COMPUTING |
|
|
| |
Jing Chen
A search engine for 3D models of musuem artefacts -
I propose to develop the 3D search engine that will be at the core of the web-based retreival and inspection demonstration
system that is one of the key objectives of DigiFact Project (Digitization of Museum Artefacts Facilitating Search and Retrieval),
an on-going project within the AIC/CDVP at Dublin City University with the aims to develop a low-cost solution for digization and
indexing of multiple views of museum artefacts. (see full description (2-page; PDF))
Supervisor: Dr. Noel O'Connor
SCHOOL OF ELECTRONIC ENGINEERING |
|
|
| |
|
*Intellectual Property arising from these projects
The School of Computing has
formulated a policy on
the distribution of intellectual property arising
from student projects (such as these). This is based on an agreed a priori disbursement agreement
between lecturer and student in the cases where the lecturer has a non-trivial involvement in the
project and where the lecturer wants to put in place such an agreement and the student agrees.
In the case of projects from these there is a considerable amount of background knowledge provided
and a significant amount of direction and facilitation by myself and other members of the Centre
for Digital Video Processing. Consequently we shall require students to sign over intellectual property
development rights for such projects before the commencement of the project.
(Student Agreement Form - DOC; 26K)
|
|
| Year 2005/6 |
| |
Peter Kehoe [MSE]
Using Graphics Processor Units (GPUs) for Image Analysis: A Comparison between GPU, Grid and Conventional
Software Engineering (MSE) - I've recently been reading up on how some people have been able to use the
graphics proceeding units (GPUs) in a PC to do things other than rendering for computer games. 'Brook' is a
generic system which extends C and there's a language called gC which provides a programming environment.
Only certain problems are suitable for implementing in this environment, those that are SIMD (single
instruction, multile data) obviously, as that's the way that GPUs work. From this I have 2 suggestions
for projects - the first is for MSE Practicum students and is to examine this as an evironment for software
development and to compare it to grid computing, and the second is for MSSF practicum students and is to see
if you can use this for something in security, an implementation of encryption or something. Both are a bit
open-ended but would allow us to get familiar with this fascinating concept of using the graphics processor
for something useful (not that computer games are not useful ;-)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
| |
Dan Xu and HaiYun Wu [CA4]
Using Google Earth for Location-based Photo Mappting - As part of a larger project in the CDVP I have
a collection of c.12,000 personal photos, each with location information (i.e. GPS location), taken from
around the world. This project would involve taking those photos, or a subset, and adding them as a layer
to the Google Earth application, where thumbnails of the
photos or groups of photos at a given location pop up as one flys over that location using the application.
Selecting a photo should reveal characteristics of the photo such as a detailed version or metadata (date, time,
etc.). It might also be possible to connect photos temporally, i.e. draw lines between the photos based on which
is next, indicating a route I took when I took the photos. Other additions will depend on yours (and my)
creativity ! This will be a really smart demo if it works as it should, and experience with Google Earth will be interesting.
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
| |
Declan Flynn and David Cavanagh [CA4]
Using Microsoft Virtual Earth for Location-based Photo Mappting - As part of a larger project in the CDVP
I have a collection of c.12,000 personal photos, each with location information (i.e. GPS location), taken from
around the world. This project would involve taking those photos, or a subset, and adding them as a layer to the Microsoft
Virtual Earth application, where thumbnails of the photos or groups
of photos at a given location pop up as one flys over that location using the application. Selecting a photo
should reveal characteristics of the photo such as a detailed version or metadata (date, time, etc.). It might
also be possible to connect photos temporally, i.e. draw lines between the photos based on which is next,
indicating a route I took when I took the photos. Other additions will depend on yours (and my) creativity !
This will be a really smart demo if it works as it should, and experience with Microsoft
Virtual earth will be interesting.
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
| |
Conor Brady and Gary Craig [CA4]
Automating the Generation of Movie Trailers - One of the technologies we have in the CDVP is to analyse
a move and detect things like speech, music, silence, other sounds, faces on screen, shot change frequencies,
motion of objects, camera motion, colours and colour changes, and a range of other video "features". WE can
use these to generate what we believe will be action sequences, or person-person dialogies, but it would be
interesting - and novel - to use these analyses to generate movie trailers. This project involves taking a number of
blockbuster movies ... 10, 20, 50 maybe, using our tools to analyse them in the ways described above, then locate
their "trailers" from the Sony or other website, analyse them and locate them in the original full movie, and
then see if you can automate the generation of a movie trailer from our analysis for another set of movies,
not in the original development set. The movie trailer provides a ground truth against which you'd have to
measure a technique we would have developed. The project requires no great expertise in video/audio analysis
since these are all done but if it works, or rather depending on how well it works, it will be impressive.
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
| |
Neil Lundy and Thomas Greely [CA4]
Using the Shape of the Hand as a Biometric Identification Charcteristic - While I was visiting the University of Buffalo
in 2005 I saw a project which captured the shape of a person's hand, with fingers splayed, as a means for recognising or
identifying people. There is lots of info on this on the web. It was a cardboard box with a backlit glass plate and a
webcam mounted over to capture the shape of the hand. The shape was captured and analysed and entered into a database,
and subsequently the shape of peoples' hands were used to identify people in the database, just like a fingerprint or
an iris. I'm curious to try to scale this up to many dozens of people and see how unique is the shape of the hand as
a biometric recognition trait. The project will require a student to build a rig to capture the hand shape (old shoe
box, light, plate of glass, webcam), analyse shapes and match them using a patented shape matching algorithm of
ours, and then capture hand shapes from many people, perform an analysis, and see how well the hand shape works.
This will lead to a real working demo system.
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
| |
Jennifer Smith and David Stacey [CA4]
The Dancing Robot - I have an ER1 robot which is a laptop with wheels (so its mobile), a wireless network connection,
speaker, microphone, etc. (all the usual laptop stuff), a webcam mounted as an "eye" and a pair of infra-red sensors
which, when it is moving, can be used to warn it that it might bump into something. We also have (in CDVP) tools to
analyse audio, specifically music, in order to determine beat and rhythm. Put the two together and we have an idea
for a dancing robot which takes a piece of music, analyses it to determine beat and high points, and then computes
a set of movements (with its wheels) such as spins, turns, forward/back, etc., to go with the music beat analyses.
Then put that on the ER1 and it should play the music, and move in time to it ! You could really go wild with this
and come up with a set of basic dance moves for jive, waltz, tango even, which could be taken and put to music.
Lots of technical problems with this, but really cool if it works, and it wouldn't stand on your toes 'cos of the
IR sensors ;-) Shall we dance ...
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
| |
Denis Kozlov and Shane McKinley [CA3]
Image Comparison Algorithms - There are several ways in which images can be compared using low-level
features including using colour histograms, global and regional, using texture maps, using edge
histograms and so on. This project will investigate a number of variants of this using a large
set of images and will determine differences between different approaches.
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
| |
Eoin Higgins and Darren Hamilton [CA3]
Automated Running Diary and Calculator - A web-based training guide and diary for 10k /
marathon runners, generating individual training schedules for runners depending on past training
and targets, using athletics norms to calculate future training. Dynamically adjusts the calendar
of recommended training based on past training and targets
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
| |
Cormac Fay
Indexing multiple views of museum artefacts
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
Ashley Sterritt
Visual detection of chemical sensing events
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
Nathan Perrem
Using hand silhouettes for person recognition
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
|
| Year 2004/5 |
 |
Ms. Joyce O'Hare
CCTV Image Analyser and Access System (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
 |
Mr. Stephen Whitney
Video Summarisation for RTE News (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Dr. Gareth Jones |
|
 |
Mr. Alan Price
Video Mosaic (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
 |
Mr. Ronan McMahon
Video Mosaic (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
 |
Mr. Oliver Smyth
Unsupervised Video Segmentation (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
 |
Mr. Sean De Burca
Unsupervised Video Segmentation (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
 |
Mr. Shane Carroll
My Average Face (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
 |
Mr. Aidan McGowran
My Average Face (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan Smeaton |
|
 |
Mr. Paul Drumm
Photo Map (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Dr. Cathal Gurrin |
|
 |
Mr. Eoin Brady
Photo Map (2004/2005)
SCHOOL OF COMPUTING |
Supervisor:
Dr. Cathal Gurrin |
|
| |
|
| Year 2003/4 |
 |
Ms. Caroline Toland
Investigation of CBIR using the MUVIS System (2003/2004)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
Mr. Cathal Lathrop
Linking Stories using Icons in News Broadcasts (2003/2004)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
Mr. Alexandre Pascal
EM-based Image Region Segmentation for Retrieval (2003/2004)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
Mr. Robert Power
Cityscape/Landscape Classification for Still Images (2003/2004)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
Mr. Robert Lambe
Real-time Audio Capture for Signal Processing Applications (2003/2004)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
Mr. Jonathan Wynne
Football tracking in MPEG-1 Video (2003/2004)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel Murphy |
|
 |
Ms. Carma O'Connor-Dunne
Touchscreen Interface to Físchlár (2003/2004)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Ms. Rose-Anna Sands
Touchscreen Interface to Físchlár (2003/2004)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Ms. Ruth O'Shea
SPIRIT Digital Image Library (2003/2004)
SCHOOL OF COMPUTING |
Supervisor:
Dr. Cathal Gurrin & Prof. Alan F. Smeaton |
|
 |
Mr. Dennis Kells
SPIRIT Digital Image Library (2003/2004)
SCHOOL OF COMPUTING |
Supervisor:
Dr. Cathal Gurrin & Prof. Alan F. Smeaton |
|
 |
Mr. Damien McCarthy
Vehicle Registration Identification (2003/2004)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Chris Blake
Vehicle Registration Identification (2003/2004)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Andrew Redmond
Evaluation of Variable Speed FF for Video Gisting (2003/2004)
SCHOOL OF COMPUTING SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Prof. Alan F. Smeaton & Dr. Noel Murphy |
|
 |
Mr. Mark McEvoy
Evaluation of Variable Speed FF for Video Gisting (2003/2004)
SCHOOL OF COMPUTING SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Prof. Alan F. Smeaton & Dr. Noel Murphy |
|
 |
Ms. Lindsay Moran
RTE News Annotation Tool (2003/2004)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Ms. Ning Sun
RTE News Annotation Tool (2003/2004)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
| |
|
| Year 2002/3 |
 |
Mr. Scott Lewis
TV schedule browser (2002/2003)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Raymond Wong
Camera motion detection (2002/2003)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Fabio Molle
Multimedia messaging of video images (2002/2003)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Alan Shannon
Multimedia messaging of video images (2002/2003)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Ms. Karen Power
Video shot similarity and retrieval (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
 |
Ms. Jovanka Malobabic
Analysis of text in video (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
 |
Mr. Alan Liu
Java-based experiments in video compression (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
 |
Mr. Ciaran O'Conaire
MPEG-7 feature extraction (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
 |
Mr. Eoin Reilly
Development of a C software library for MPEG-4 video (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
 |
Mr. David Dunne
Detection of exciting sports commentary (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
 |
Mr. Gordon Doyle
Monitoring the on-screen scoreboard in digital TV sports programs (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
 |
Mr. Seán Callery
Modeling power-efficient motion estimation search strategies in Java (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
 |
Mr. John Lyng
Tracking players using word spotting sports commentary (2002/2003)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
| |
|
| Year 2001/2 |
 |
Mr. Colm Dunne
XSL interface to Físchlár (2001/2002)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Cathal Small
Video stream shot comparison (2001/2002)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Karl Podesta
Parallelising Físchlár video analysis (2001/2002)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
| |
|
| Year 2000/1 |
 |
Mr. Jonathan Lundberg
Closed-Caption search on Físchlár (2000/2001)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton
PROJECT LINK |
|
 |
Mr. Brian Delahunty
Cartoon detection on Físchlár (2000/2001)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Andy McGeady
SMS messaging to Físchlár (2000/2001)
SCHOOL OF COMPUTER APPLICATIONS |
Supervisor:
Prof. Alan F. Smeaton
PROJECT LINK |
|
 |
Ms. Jodie Foley
Camera motion detection from digital video (2000/2001)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton
PROJECT LINK |
|
| |
Mr. Derek Williams
Extraction of Camera Motion from Digital Video (2000/2001)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel Murphy |
|
| |
Mr. Tristan Collery
Real-time streaming of MPEG-4 encoded video to an iPAQ (2000/2001)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
Mr. David Davoren
Development of a C/C++ software implementation for audio signal
discrimination Speech-silence detection in audio signals (2000/2001)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel Murphy |
|
| |
Ms. Breda Carroll
Development of a C/C++ software library for MPEG-4 Audio Modeling (2000/2001)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
| |
Mr. Andrew Doyle
Power-efficient DCT Structure in Verilog Modeling (2000/2001)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
| |
Mr. Allan Kerr
Power-efficient Motion Estimation Search Strategies in Verilog (2000/2001)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
| |
Mr. Thomas Schulz
Transfer of a Video Stream over a Serial Port to a PC using an
Embedded ARM Platform (2000/2001)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
| |
Mr. Seán Deasy
(2000/2001)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
| |
|
| Year 1999/2000 |
 |
Mr. Declan Brennan
ASCII animation of video (1999/2000)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton |
|
 |
Mr. Stephen Donnelly
Camera motion detection for MPEG files (1999/2000)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton
PROJECT LINK |
|
 |
Mr. Gabriel Louet-Feisser
Photomantage generator from MPEG (1999/2000)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton
PROJECT LINK |
|
 |
Mr. Kevin Walsh
Photomantage generator from MPEG (1999/2000)
SCHOOL OF COMPUTING |
Supervisor:
Prof. Alan F. Smeaton
PROJECT LINK |
|
 |
Ms. Debbie Carroll
(1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
 |
Ms. Margaret Healy
(1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
|
 |
Mr. Shane Hickey
(1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel Murphy |
|
| |
Mr. Robert Monks
Automatic Extraction of Embedded Subtitles in Television Programmes (1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel Murphy |
|
 |
Mr. Keith Critchley
Development of a API for MPEG-1 Video Bitstreams (1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel Murphy |
|
| |
Mr. John Neary
Implementation of Real-time Video Shot Cut detection (1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel Murphy |
|
| |
Mr. Stephen Kenny
(1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
 |
Mr. Melvyn Sheehan
(1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
| |
Mr. Dave Peyton
Detection of Action Replays in Digital TV Sports Programs (1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
 |
Mr. Patrick Ward
(1999/2000)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Seán Marlow |
|
| |
|
| Year 1998/1999 |
| |
Mr. Liam Brett
Java-based illustration of block-based motion estimation and compensation (1998/1999)
SCHOOL OF ELECTRONIC ENGINEERING |
Supervisor:
Dr. Noel O'Connor |
| |
|
|
|
|
|
Centre is running a number of
major projects.
If you have question on
the Centre's Undergraduate & (Taught) Postgraduate Projects,
contact the relevant staff as referred to in each entry.
Telephone:
+353 -1 -7005262
Fax:
+353 -1 -7005442
Address:
Centre for Digital Video Processing,
Dublin City University,
Glasnevin, Dublin 9
Ireland
|
|
