ECCV 2026 Workshop · Malmö, Sweden
AI for Visual ArtsThe 4th Edition
--Days
--Hours
--Minutes
--Seconds
Until the camera-ready deadline · August 15, 2026, 23:59 AoE
SCROLL
ECCV 2026 Workshop · Malmö, Sweden
Until the camera-ready deadline · August 15, 2026, 23:59 AoE
Abstraction
Heritage
NarrativeThe 4th AI for Visual Arts workshop at ECCV 2026 explores how computer vision perceives, interprets, models, and generates abstraction across artistic and cultural imagery: paintings, comics, films, typography, sculpture, and heritage artefacts that challenge the assumptions of conventional models.
Building on three successful editions, AI4VA now expands toward the frontier of generative and agentic creativity: world models for cinematic simulation, long-form video generation for storyboarding, neural 3D representations for creative domains, and agentic workflows for art, film, and storytelling, alongside our enduring focus on perception under abstraction, provenance, and the digitisation and restoration of cultural heritage.
Advance robust segmentation, saliency, depth, and multimodal reasoning under stylization and abstraction. Evaluate how vision models interpret artistic imagery and generalize beyond photorealism.
Push long-form, temporally coherent video generation, world models, neural 3D representations, and agentic workflows that give artists cinematic and stylistic control over the creative process.
Benchmark watermarking, authenticity, and provenance in AI-assisted art; digitise historic archives and apply spatial intelligence to the restoration and preservation of cultural heritage.
Closed. This track is no longer accepting submissions
Authors notified by email
11:59 PM AoE · one PDF to OpenReview, max 14 pages excl. references · Springer via Meteor · read the full instructions
Closed. This track is no longer accepting submissions
11:59 PM AoE · authors notified by email · no camera-ready required
Afternoon (PM) session · Malmö, Sweden
Novel, previously unpublished research. Accepted papers appear in the official ECCV 2026 Workshop Proceedings. This track closed on July 27, 2026 and is no longer accepting submissions.
New, previously, or concurrently published research or work-in-progress. Will not appear in proceedings. This track closed on August 6, 2026 and is no longer accepting submissions.
Every accepted paper presents a poster in the MalmöMässan Exhibit Hall during the 15:00–16:00 session — this includes papers selected for an oral spotlight. Oral spotlight authors present both a short talk and a poster; the talk does not replace the poster. Poster boards are 140 × 100 cm; poster printing information will be posted on the website shortly. Room assignments will be posted on the app.
All ECCV 2026 papers, both workshop and main conference papers, that have been approved for inclusion in the proceedings are required to have the full passport/author registration.
Camera-ready instructions for accepted full archival papers are now available and are due August 15, 2026, 23:59 AoE. The main paper is limited to 14 pages excluding references. There are two separate submissions — OpenReview and Meteor — with different rules about supplementary material. Read the full instructions carefully →
These instructions apply to accepted full archival papers only. Extended abstracts are non-archival and require no camera-ready submission. There are two separate submissions — one to OpenReview and one to Meteor — and the rules about supplementary material are different for each. The main paper is limited to 14 pages excluding references in both.
Please read carefully · Due August 15, 2026, 23:59 AoE
Authors of accepted full archival papers must upload their camera-ready papers to OpenReview using the “Camera Ready Revision” button on their submission page. The deadline is 15 August 2026, 23:59 AoE.
Accepted full-paper authors will be invited directly to submit their papers and supplementary material via Meteor — the same procedure as for the main conference. The invitation comes separately from the ECCV and Springer publication chairs; you do not need it in order to complete your OpenReview upload. The 14-page limit excluding references applies here too.
What Meteor will allow you to submit
Springer will allow only ONE submission per paper. The submission link can only be used ONCE. Please make sure everything is final before you submit. Submission instructions are the same as for the main conference: read the ECCV 2026 submission instructions.
The ESM may consist of PDFs, videos, images, data, or other files supporting the paper. The ESM should not be considered an Appendix of the paper. Authors may refer to the ESM generically — e.g. “as described in the Supplementary Material” — but there should be no formal cross-references, e.g. to “Appendix B.3”.
All ECCV 2026 papers, both workshop and main conference papers, that have been approved for inclusion in the proceedings are required to have the full passport/author registration.
Half-day afternoon session · Wednesday, September 9, 2026 · Malmö, Sweden
Deblina Bhattacharjee · Workshop Chair, University of Bath
LMU Munich · latent diffusion models and generative image research
University College London · Adobe Research · generative models for 3D content
University of Washington · Allen Institute for AI · vision-language models and multimodal reasoning
Please make your way to the MalmöMässan Exhibit Hall
All accepted papers, archival and non-archival — including the oral spotlight papers · MalmöMässan Exhibit Hall
Our fourth keynote is awaiting confirmation and will be announced shortly
Four selected papers · 7 minutes each followed by 3 minutes of questions · these authors also present a poster in the 15:00 session
Deblina Bhattacharjee · Workshop Chair, University of Bath
The programme is provisional and subject to change. Room assignments will be posted on the app.
Board size · 140 × 100 cm
All accepted papers present a poster, including those selected for an oral spotlight. An oral spotlight is in addition to the poster, not instead of it.
All posters will be displayed in the MalmöMässan Exhibit Hall during the 15:00–16:00 session. Boards are shared with other workshops and are assigned according to schedule, so please set up and remove your poster within your session window.
Poster printing information will be posted on the website shortly.
Malmö, Sweden · September 9, 2026
Room assignments will be posted on the app. The keynotes and oral spotlights take place in the workshop room; the poster session takes place separately in the Exhibit Hall.
Presenters are asked to load their slides during the preceding break.
BOLMU Munich
Professor of Computer Vision & Learning at Ludwig Maximilian University of Munich. His group's work on latent diffusion models underpins Stable Diffusion and much of today's generative image research.
NMUniversity College London · Adobe Research
Professor of Computer Graphics at UCL, where he leads the Smart Geometry Processing group, and Senior Research Scientist at Adobe Research. Works on geometry processing, shape analysis, and generative models for 3D content.
RKUniversity of Washington · Allen Institute for AI
Assistant Professor at the Paul G. Allen School and research lead at AI2. Creator of Visual Genome; his work spans vision-language models, multimodal reasoning, and human-AI interaction.
Keynote 4
Our fourth keynote is awaiting confirmation and will be announced here shortly.
Lead OrganizerAssistant Professor, University of Bath, UK
Sponsors for the 4th edition will be announced soon. Previous editions of AI4VA were generously supported by Meitu, whose contributions made our awards and challenges possible.
Interested in supporting the intersection of AI and visual arts? We welcome partners who share our vision for trustworthy, human-aligned AI in creative domains.
Become a Sponsor
