Machine learning can be a powerful tool for analysing images, but it typically depends on large volumes of labelled data that are not always available. A University of St Andrews computer science student has been named Young Software Engineer of the Year for a project that takes a different route, reconstructing 3D models of animals from 2D wildlife images to support ecological research.
Lily Martin’s project, “Extracting 3D Models of Animals from Wildlife Image”, combines computer vision, geometric reasoning, biological knowledge and optimisation techniques to estimate an animal’s posture and body structure, even when field photography is limited or parts of the animal are obscured. It uses cliff-nesting birds as its test subject.
The work offers an alternative to data-intensive machine learning approaches and lays the foundation for interactive 3D models that could support future research into animal behaviour and movement. Judges praised its strong presentation, its considered approach to a complex, multidisciplinary challenge and its potential for wide-ranging application.
The first prize is sponsored by Sopra Steria. Martin’s win is the second year in a row that a St Andrews student has taken the top award.
The awards, organised by ScotlandIS, the membership and cluster management organisation for Scotland’s digital technologies industry, recognise outstanding final-year computer science and software engineering students from Scottish universities. The 37th annual awards were presented in Edinburgh on 24 September at ScotSoft, ScotlandIS’s annual flagship conference.
Second prize, sponsored by BCS, went to Neelan Patel of the University of Glasgow for a project that brings symptom assessment and appointment booking together in a single AI-assisted system for general practice.
Third prize, sponsored by Resillion, went to Lea Kotzmuth-Clarke of the University of Stirling. Her software combines technology risk compliance auditing and cybersecurity vulnerability detection, and is designed to accelerate decision-making and action.
The Best Engineering Prize, sponsored by Leidos, was awarded to Danyl Kecha of the University of Dundee for a project exploring a new way to make the outputs of AI reasoning models shorter without sacrificing performance or reasoning density. The approach could help reduce the computational resources AI systems require.
Entries were assessed by a panel of industry experts on criteria including innovation, technical difficulty, depth of knowledge and research, and commercial and social relevance. All four winners received a trophy and a cash prize.
Andrew Weir, chief executive of ScotlandIS, said: “What stood out to me was how relevant so many of the projects are to the issues businesses and wider society are grappling with right now. From AI and cybersecurity to healthcare and environmental research, these students are truly thinking about how technology can be applied to make a difference.”
Martin said: “Machine learning can be incredibly useful, but it eats through huge amounts of labelled data, which isn’t always available. That’s why it’s important to explore other approaches, like geometric reasoning, biological knowledge and classic optimisation techniques, which can be just as effective. I hope my research shows how computing science can be used in a practical way to support future ecology studies of animal behaviour and movement.”
The 13 shortlisted students were Rebekah Leslie (University of Aberdeen), Jack Briggs (Abertay University), Danyl Kecha (University of Dundee), Morgan Daniel (University of Edinburgh), Jacob Lewis (Edinburgh Napier University), Neelan Patel (University of Glasgow), Callum Gourlay (Glasgow Caledonian University), Ivan Meister (Heriot-Watt University), Kobina Ampofi-Forson (Robert Gordon University), Lily Martin (University of St Andrews), Lea Kotzmuth-Clarke (University of Stirling), Eva Blues (University of Strathclyde) and Dominika Stachowska (University of the West of Scotland).
