Academic year: 2017/2018.
My interests are in data analytics, computer vision (images or video data) or machine learning applications in general.
I'm also part of the Smart Stadium/Smart Cities initiative with Croke Park, Arizona State University, Intel, Microsoft
and DCU. Projects that may have Smart Cities or Internet of Things applications are very welcome.
Identify plant species from leaf photographs:
under supervision of Dr Naresh Yarlapati, this project will implement and evaluate algorithms for identifying plant species from images of folliage (eg. Leafsnap dataset
) using various machine learning approaches. This has applications in environmental monitoring and computer vision.
Video classification and search: under supervision of Dr Houssem Chatbri, this project will implement and evaluate algorithms to classify, index and search video data.
Analysis of racist and offensive tweets A dataset has been created as part of a project with the School of Communication that contains examples of racist and offensive tweets and comments. The project would perform statistical analysis on the dataset to explore possible patterns and clusters and/or apply machine learning methods to create classification models that label tweets on a scale of possibly "racist". The project could also use social network graphs to explore links between accounts.
Finding agreement in different descriptions of
an event. Prof Alan Smeaton has a large dataset of first year computing student
responses describing lectures from the Digital World module. These plain
text email messages can be analysed to find the level of agreement and
difference between how individuals describe and remember a talk. This is
potentially useful for understanding eye-witness accounts and the
different emphasis that people place on content. The project would take
the anonymised email archive, extract the text, clean and process the
descriptions and apply various text analytics and classfication processes.
: using sound level monitoring data from Croke Park and raw audio stream filtering to visualise events (see here
- Monitor pigeon and seagull activity on the pitch at Croke Park from an existing set of timelapse images (Beware!)
- Apply person tracking to classify activity in a scene
- Event classification from photographs - example dataset
- Use the concept of selective forgetting to decide what to store from a CCTV stream
Phone: ext. 6052