Tenured Assistant Professor of Computer Science at the University of Bath, and Fellow of the Higher Education Academy.
I teach machines to understand, restore and reason about visual culture, from comics and film archives to sign language.
My group works on generative and multimodal models, 3D reconstruction, depth, saliency and multitask learning. We apply them to the visual arts and archives, where pictures carry culture; to British Sign Language, where they carry access for Deaf communities; and to medical imaging. I care about AI that is responsible and inclusive, and about how it meets human cognition.
Before Bath I did my PhD and a postdoc at the Image and Visual Representation Lab, EPFL, and worked at the Samsung–POSTECH Intelligent Media Research Centre and Google. I am Chief AI Advisor to Silence Speaks and lead the AI for Visual Arts workshop series.
Four strands, one question: how should machines look at the things people make?
Comics, manga, paintings and film archives break the assumptions of models trained on photographs. We build datasets and methods for depth, segmentation, saliency, retargeting and sketch-to-image generation in stylised art, and generative restoration for archives. I lead the AI for Visual Arts (AI4VA) workshop series at ECCV, ICCV and CVPR.
With Silence Speaks we are prototyping real-time, photorealistic 3D animatable British Sign Language avatars and a 3D dataset of 5,000 BSL signs, evaluated with Deaf end-users. Methods from our earlier work power the company’s avatar pipeline in use at Leeds Teaching Hospitals NHS Trust and with Cambridgeshire County Council.
Editing-first 3D face models whose shape bases are learned per facial region, transformers that share attention across dense tasks, adapters that transfer to new tasks without fine-tuning, instance-aware image translation, and diffusion pipelines that ground language in the right region.
Physics-informed latent graphs that turn cardiac vortices into interacting nodes, with graph entropy as a marker of disease severity, developed with Dr Katharine Fraser’s group in Mechanical Engineering. Related work asks why retinal graders trained on colour fundus images fail on near-infrared ones.
Further projects, including a causal audit of LLM-generated image-editing instructions, are under review and will appear here on publication.
Upcoming: talk and live demo at the Sherborne Alan Turing centenary Tech Fest on 16 October; BMVA technical session on AI for the Visual Arts in December.
Edit3DMM, an editing-first 3D morphable face model, is accepted at CVMP 2026, the ACM SIGGRAPH European Conference on Visual Media Production.
Our paper with Yin Zhang (University of Geneva) on author-stigmatising hazards in LLM literary translation is accepted to the Findings of EMNLP 2026, Budapest.
Chaired the 4th AI for Visual Arts workshop at ECCV 2026, Malmö.
Innovate UK Accelerated Knowledge Transfer grant (£35,000) with Silence Speaks for 3D animatable BSL avatars; Silence Speaks is also co-funding a URSA 50:50 PhD studentship in my group.
Appointed Broadening Participation Chair and Publicity Chair of CVPR 2027, and Area Chair for WACV 2026.
Live on BBC Radio for World Wide Web Day (1 August); invited tutorial at the Bath Machine Learning Summer School.
Faculty of Science Recognising Excellence Award. Quoted in SurreyLive on governing AI in NHS trusts. Invited tutorial at the 29th BMVA Computer Vision Summer School, Durham.
At CVPR 2026, Denver: keynote at the SINT4CH workshop on spatial intelligence for cultural heritage, lead organiser of the 3rd AI4VA workshop, and two student-led workshop papers, one awarded Best Poster among 113. Publicity Chair of the conference.
Led the Faculty of Science workshops at the University of Bath Doctoral Festival of Ideas.
Invited delegate at the United Nations Office at Geneva (UNRISD) and invited speaker at The National Archives, Kew, on AI for archival restoration.
New preprint with Katharine Fraser’s group: Learning Disease-Sensitive Latent Interaction Graphs From Noisy Cardiac Flow Measurements.
Panellist at the BCSWomen Lovelace Colloquium 2026.
Organised the 2nd AI4VA workshop and co-organised COMIQ at ICCV 2025, Honolulu.
Chair of the Department’s Athena SWAN Self-Assessment Team.
Perplexity AI Business Fellowship, Elevate Leadership Fellowship (Ubele Foundation) and Impact and Knowledge Exchange Fellowship; Fellow of the Higher Education Academy.
Co-organised Women in Computer Vision at CVPR 2025, Nashville; Publicity Chair of CVPR 2025; panellist on research impact at CVPR 2025 and at RBC Borealis, Toronto.
Advisory engagement with Silence Speaks; methods from this work form the BSL avatar pipeline deployed at Leeds Teaching Hospitals NHS Trust and with Cambridgeshire County Council.
General Chair of the 1st AI4VA workshop and challenges at ECCV 2024, Milan; three ECCV 2024 papers; co-organiser of WiCV at CVPR 2024, Seattle.
Joined the University of Bath as Lecturer (Assistant Professor) in Visual Intelligence.
Guest on the SuperDataScience podcast: Deep Learning for Machine Vision and Solving an Optimisation Problem with a Custom-Built Algorithm.
CVMP 2026, ACM SIGGRAPH European Conference on Visual Media Production
Findings of EMNLP 2026, Budapest
CVPR 2026 Workshops (AI for Visual Arts)
CVPR 2026 Workshops (AI for Visual Arts)
arXiv 2602.23035, under review
ECCV 2024, Milan
arXiv 2403.06546
Workshop report, arXiv 2411.02445
PhD thesis, EPFL
Transactions on Machine Learning Research
IEEE Signal Processing Letters, vol. 28
American Journal of Computer Science and Technology, 4(4)
Computers & Electrical Engineering, vol. 65, 2018
AAAI 2017, San Francisco
NeurIPS 2017 Women in Machine Learning Workshop, Long Beach
Preprints.org, 2017
MSc thesis, Kyungpook National University, 2017. Best MSc Thesis Award
International Conference on Swarm Intelligence (Springer LNCS), Bali, 2016
SpringerPlus, vol. 5, 2016
ACM Symposium on Applied Computing, Student Research Competition, Pisa, 2016
PlatCon 2016, Jeju
Across the UK, US, Switzerland and China. I currently supervise four PhD students at Bath and co-supervise a doctoral researcher at the University of Geneva.
I am always open to doctoral students who want to work at the confluence of AI and the creative domains, accessibility or responsible AI. Write to me.
