Student Projects suggestions (Suzanne Little)

My interests are in data analytics, computer vision (images or video data), machine learning applications in general and in fair and responsible AI. I've worked on projects involving instrumented vehicles ("self-driving" cars); smart cities (how crowds move); medical image data and bias in AI models.

Please consider the following questions before approaching me to discuss a project.

1. What is your topic, problem or domain?
2. What datasets might you use? (how big?, is it labelled?, is it ethical?)
3. How will you know if you have solved or contributed to the problem? (or how will you measure the performance of your methods? how will you test your implementation?)

Frequently Asked Questions

• Will you supervise my project?  Maybe. Projects that I'm less likely to supervise (and there are people with more experience) - text analytics or natural language processing; games; cloud systems; security.
• What data should I use?  It should be large enough to test and train systems. It should be labelled (eg. with the objects or actions you're interested in). You should have permission to use it.
• I want to do deep learning. Can you give me a GPU?  Maybe. There are some options for local access to high-performance systems that will require experience in linux and python programming. Students should also look at options from Google Colab and Amazon web services. Note that these systems are not enough to train large video datasets (insufficient RAM or process timeout constraints). It takes time to learn, implement, train and test CNN/Deep learning systems so consider carefully if this is what your project depends upon.
• I want to train a text-to-image system or LLM from scratch!  The main companies (OpenAI, Meta, Microsoft, Google etc.) have considerably more compute power and data than it is possible for you to achieve during your practicum. Consider what you want to achieve and how you will get results.

Project ideas, 2026/2027

• Analysis of Historical Images:  Using data from The British Library (Book Images), which contains images taken from material published from 1510s to 1890s, analyse and cluster the visual content to identify trends or patterns. Create a classifer that estimates the publication date given the image with and without context. Evaluate an off-the-shelf image classifier to estimate it's performance (no groundtruth is provided). There is also associated text data available. The datasets are large and will need to be processed using big data techniques.
• Historical document analysis:  Utilise Churro (or something similar) which performs analysis of historical documents and apply to a new collection such as the Irish National Archives records of the Dáil Éireann from 1919-1922. Genealogy records could also be interesting.
• Generating Constraint Satisfying Physical Layouts:   Generative AI images are currently fairly poor at understanding spatial instructions and at considering physical contraints. Can you devise ways to generate physical layouts (a 3D model to print, a room layout, a poster layout) that exploits GenAI but also captures proper physical contraints or rules?
• Alternative communication channels:   Most data apps will use a dashboard or a mobile phone interface to communicate or signal information. What other communication channels could be used to create data-driven visualisations? For example, audio "sonifications" rather than colour graphs; or mechanical signage (like the old split-flap departure boards) to relay information in the physical world; or novel ways of using colour or motion.
• Browse available image or video datasets:  Computer Vision Challenges or Hugging Face Datasets can be a good starting place to explore a problem in image or video analysis (e.g., Detection, Classification, Recognition , Tracking, Segmentation , Foreground/Background, Saliency detection, video Surveillance, Multi view, Action, Human Pose/Expression). Be aware that there's not much time in the project to use a very large dataset!