Research Group

DECaDE: UKRI/EPSRC Next Stage Centre for the Decentralised Digital Economy

I am founder and director of DECaDE, a multi-disciplinary research centre exploring the future digital economy, with a focus on the creative industries. We live in a decentralised digital economy where anyone can produce or consume digital goods and services (e.g. Uber, AirBnB), yet this activity is mediated by centralised platforms with closed governance.

DECaDE investigates how data-centric technologies such as AI and Distributed Ledger Technology (DLT) can enable decentralised platforms that support fairer governance, build trust and resilience, and open new models of value creation. A core theme is digital supply chains, using decentralised technologies to track the provenance of digital assets (or digital twins of physical assets).

Our current emphasis is on the creative supply chain: provenance of media to establish authenticity (combatting misinformation and deepfakes) and to improve attribution and reward for creators - issues made more urgent by Generative AI.

DECaDE not only explores technological aspects but also incorporates researchers from the Surrey Business School and Surrey Centre for Cyber Security, alongside academics from Edinburgh’s School of Law and Institute for Design Informatics, reflecting the deeply socio-technical nature of this space. DECaDE partners with the Digital Catapult, whose DLT Field Labs help scale our prototypes into real-world trials. Read more on the DECaDE website and in the DECaDE Impact brochure (2024).

Copyright and AI: Provenance for Media Rights and Monetization

Tracing the origins (provenance) of media can not only help establish authenticity, it can help establish ownership. This in turn can help establish frameworks for consent and rights over re-use and ultimately a means to create value (monetary, reputation, etc.) for rightsholders as content is re-used. This is particularly important now in the age of Generative AI, where content is being re-used at massive scale in many cases without consideration of consent or compensation to creative rightsholders.

The provenance of a synthetic image is the AI model that made it, and ultimately the training data used to create that model. As such, we consider provenance backed content licensing to be a promising direction to resolving the issues arising today around Copyright and AI. In support of this goal, we published early solutions for attributing synthetic images to the AI models that made them [e.g. RepMix, ECCV 2022] and for determining the most responsible training data for a generative AI image [e.g. EKILA, CVPRW 2023]. Later works have explored the use of data and model watermarking to establish creator attribution [e.g. ProMark, CVPR 2024; CustomMark CVPRW 2025].

Crucially we have shown how such solutions can be unified with open provenance standards (C2PA) and distributed ledger technologies (DLT) to create automatic ‘copyright exchanges’ that empower creatives to express consent and receive compensation for re-use of their work. Examples include [DECORAIT, CVMP 2023] and [EKILA, CVPRW 2023]. We describe these applications for provenance ‘to authenticity and beyond’ in our [IEEE CG&A, 2024] paper, which presents solutions both for durable media provenance markers and outlines a technical framework (ORA) for monetizing content at the time of re-use. This work was developed further to outline end-to-end frameworks for what are now referred to as ‘creative copyright exchanges’ – e.g. [ContentARCs, CADE 2025) and the ACCCT framework [Time to ACCCT, CoStar/DECaDE report]. The latter has impacted policy in the emerging Copyright and AI debate, being presented to the UK Department for Culture Media and Sport (DCMS), and the House of Lords in the wake of the UK government’s consultation on Copyright and AI.

Sketch-based Visual Search

Our group was among the earliest to explore sketch based visual search of image and video databases. We developed the first technique for sketch based video search [ICCV 2009] and popular datasets and descriptors for sketch based image search (noably GF-HoG [CVIU, 2013]). We explored other interesting forms of visual search in this domain, such as sketch based choreography search as part of the AHRC funded Digital Dance Archives (DDA) project. Several PhDs were run in this space (e.g. Rui Hu, Stuart James, Tu Bui) over those years exploring novel descriptors and indexing strategies for scalable sketch based search. Our group is no longer active in sketch based research however the Sketch-X group run by Prof. Song at Surrey remains very active in this space.