AI-Powered Art Analysis and Discovery
Art Analysis
The global art market remains fragmented and heavily reliant on manual evaluations by curators, collectors, and investors, a process that is subjective, time-consuming, and difficult to scale. Traditional image-recognition systems often fail to interpret the symbolic, emotional, and historical depth of fine art.
Big Look Art is on a mission to provide an AI platform capable of understanding and describing artworks with expert-level nuance – automating analytical tasks such as identifying style, genre, emotion, and contextual relationships across large and diverse collections.
In collaboration with the Gateway and funded by Data2Sustain, Big Look Art designed and built a multimodal AI platform that combines advanced visual analysis with natural language understanding.
AI-powered platform
The project utilised the Enterprise Ireland funded High Performance Computing (HPC) infrastructure in the Atlantic Technological University (ATU). At the heart of the system is a Large Multimodal Model (LMM) that integrates:
- A Vision Transformer (ViT): Extracts detailed visual features from artworks.
- A Task-Specific Attribute Prediction System (TAPS): Analyses style, emotion, and artistic technique.
- An autoregressive language model: Generates contextualised, natural language descriptions of each artwork.
The model includes advanced classification, clustering, and similarity tools, using an Art Influence Graph and Graph Neural Networks to uncover meaningful relationships between artworks and movements. A curated dataset of thousands of public-domain artworks, annotated through a hybrid AI + expert process, underpins the entire system. Deployed through scalable cloud APIs, the platform delivers expert-level art analysis for museums, collectors, and researchers.
1
Irish Company
1
EDIH Project
1
Multimodal AI platform
“Data2Sustain’s support has been transformative for our project. Their funding, computing infrastructure (JANUS HPC), and expertise allowed us to develop Big Look Art’s initial AI model and pipeline far beyond what we could have achieved alone. With their guidance, we built a sophisticated multimodal analysis system, complete with custom taxonomies and large training sets, and validated it with real-world art professionals. We truly could not have made this level of technical progress or attracted follow-on funding without their collaboration“.
Fintan O’Byrne, Founder, Big Look Art
Big Look Art & WiSAR Technology Gateway partnership
The collaboration between Big Look Art and Data2Sustain produced a pilot AI platform that demonstrates how artificial intelligence can transform art discovery, research, and curation.
Key outcomes included:
- Automation and efficiency: The platform demonstrates the ability to generate consistent, richly detailed descriptions and metadata, reducing manual effort and improving accuracy.
- Deeper insights: Offers historical, stylistic, and emotional context to support academic and curatorial work.
- Enhanced discovery: Enables galleries and collectors to uncover hidden links between artworks, artists, and movements.
- Curated dataset: A robust, legally compliant datasetthat can support ongoing research, education, and innovation.
- Economic and social impact: The technology helps democratise access to art knowledge, bringing expert-level insights to a wider audience.
- Data-driven decision-making: Demonstrated potential to reduce inefficiencies in the art market, and opens new opportunities for art-tech services and cultural digitisation.
This project has positioned Big Look Art as a pioneering AI innovator in the creative industries, showcasing how high-performance computing and artificial intelligence can advance cultural heritage and modernise the global art ecosystem.