Participation Instructions

• Submission page:  https://openreview.net/group?id=thecvf.com/ECCV/2024/Workshop/AI4VA

• July 25th, 2024 August 7th (AoE): Paper submission deadline and extended abstract submission deadline

• August 21st, 2024 (AoE): Acceptance Notification

• September 18, 2024 (AoE): Camera-ready deadline. The ECCV Workshop proceedings will be published by Springer, in the LNCS (Lecture Notes in Computer Science) series. They will send the author a personalised link to upload the paper.

• September 29th, 2024 (AM session): Workshop event

All submission deadlines are AoE (Anywhere on Earth).

This workshop, in conjunction with ECCV 2024, calls for high-quality works related to AI and visual arts. Submissions for both tracks should conform to the ECCV 2024 proceedings style and will be double-blind peer-reviewed by three reviewers. Full-length paper submissions should be novel works and previously unpublished. Extended abstract can be new, previously, or concurrently published research or work-in-progress. However, extended abstracts will not appear in the conference proceedings. All submissions (full-length papers and extended abstracts) must be submitted as anonymous submissions through the OpenReview system here.

TOPICS include but are not limited to:

• Art History and AI

• 3D scene/ object reconstruction from historical art/ existing modern artworks/ storybook illustrations

• Novel approaches for generative art

• Generative arts, ethics, and safety

• 2D and 3D human pose in art and animations

• Multi-modal multimedia systems and human-machine interaction

• Multimedia databases and digital libraries for artistic research

• Visual Question Answering (VQA) or Captioning for Art

• Interactive 3D media and immersive AR/VR for visual art/ films/ storytelling/ comics

• Media content analysis and search

• Any other topic at the intersection of AI and Visual Art.

SUBMISSION DETAILS

All submissions will be handled electronically via the OpenReview Website: https://openreview.net/group?id=thecvf.com/ECCV/2024/Workshop/AI4VA. The format for paper submission follows the guidelines of the ECCV 2024 main conference. Please refer to ECCV 2024 submission policies (https://eccv.ecva.net/Conferences/2024/SubmissionPolicies)

There are 2 tracks to submit your workshop paper.

Full-Length Papers: 

1. The workshop will offer the opportunity to publish full-length papers in the workshop proceedings (up to 14 pages excluding references). These papers should describe new work that has not been previously published, accepted for publication, or submitted for review at another venue during our review period. Please follow the standard ECCV template for submissions in this track.  

2. Supplementary material is allowed ONLY for full-length papers and will follow the same submission deadline of August 7th.

3. The review for AI4VA 2024 will be double-blinded; Authors will not know the names of the area reviewers of their papers, and area reviewers will not know the names of the authors. Each submitted paper will be reviewed by three experts.

4. The paper length should match that intended for final publication.

5. Papers are limited to 14 pages, including figures and tables, in the LNCS style of Springer. Additional pages containing only cited references are allowed.

6. Please download the ECCV 2024 Author Kit for detailed formatting instructions. Note also that the template has changed since ECCV 2022. Authors must, thus use this new template instead of templates from older conferences.

7. While there is no limit to the number of references, reviewers are instructed to check for pertinence. Citation of any of the co-authors’ published work must be done in the third person. Reference to unpublished work can be made by anonymising it, including it as supplementary material, and referencing it as Anonymous.

8. Authors may optionally submit additional material that could not be included in the main paper due to constraints of format e.g., a video, project website with interactive demos, art demos, etc as supplementary material. However, these should be provided as anonymous links at the end of the paper and the deadline to submit these are the same as that of the full-length paper.

9. Papers that are not blinded, do not use the template or have more than 14 pages (excluding references) will be rejected without review.

Extended Abstracts: 

4 pages excluding references, NOT appearing in proceedings. The abstract will describe new, previously, or concurrently published research or work-in-progress with a length of up to 4 pages excluding references. The abstracts will NOT appear in proceedings. However, the accepted abstracts will be published online on the workshop website upon consensus agreement provided by the authors. Further submission instructions are below.

WORKSHOP PROCEEDINGS

• All accepted full-length papers will appear in the ECCV 2024 workshop proceedings.

• Authors of all accepted papers will be invited to present their work in a poster session, and a selected few in an oral session.

• The presenter need not be the first author of a submission.

Submission page:  https://openreview.net/group?id=thecvf.com/ECCV/2024/Workshop/AI4VA

Paper template:  For both types of submissions please use the ECCV paper template.

Our workshop features two challenges focusing on applying computer vision (depth, and saliency) to visual arts, centered around the AI4VA dataset (see Figure below). This dataset, a first-of-its-kind benchmark, is designed to evaluate vision models’ ability to interpret comic art, focusing on depth and saliency. It includes detailed annotations of diverse comic styles, such as the toon-styled ’Placid & Muzo’ and the realistic ’Yves le loup’, digitized from weekly papers published in the mid-20th century. These annotations cover various aspects like segments of graphical elements, ordinal depth, and character actions, enhancing understanding of artistic nuances. This is also an effort to aid artists in diffusing their work’s reach by integrating the AI4VA benchmark annotations with automated remediation algorithms. The dataset, already released, will be evaluated using standard metrics on a CodaLab-hosted server. 

All instructions related to the challenge: https://github.com/IVRL/AI4VA

Evaluation: A single-phase evaluation with the competition hosted on Codalab.

Important Dates for the Challenge

June 18th, 2024 (AoE): All data and development kit release

September 15th, 2024 (AoE): Final submission deadline

September 20th, 2024 11:59PM (AoE): Challenge report and code deadline. Please write within your challenge report a short introduction, training and implementation details,  any extra data used, and evaluation results. The report can be as per the ECCV 2024 format and there are no page limits. You can include images. Any methods used or referenced should be cited. Please note that challenge report will not be considered for proceedings but we ask the report to assess the submissions fairly along with the leaderboard standings. 

Challenge winner presentations: The winning team of each challenge will be invited for a short oral presentation and will likely get a prize. Certificates will be awarded to the top 2 winners of the challenges. No posters required from challenge winners.

NEW !! Few FAQs about the challenges: 

1) The final score of the competition is determined by the rankings in both the Development stage and the Code submission stage, which should be identical. In other words, the rankings should not differ between these two stages.

2) The top three teams from the Development stage will be required to submit their code through a public GitHub repository. The link to this repository should be sent to us via email, along with a technical report detailing your model. Specifics on the report can be found at: https://sites.google.com/view/ai4vaeccv2024/participation, under Challenge Participation. We will review and run the submitted code to verify that the rankings in the Development and Code Review stages match. To ensure fairness and replicate your original environment, it is crucial that the technical report includes information on all packages, libraries with version details, the platform used, the number of GPUs employed, and their specifications. Alternatively, you can send us all your code and report in a zipped file, if you prefer.

3) The leaderboard visible on CodaLab reflects the rankings. Ideally, the leaderboard rankings in the Development stage should match those in the Code Review stage. In the event of a tie, both top-performing teams will be awarded.

4) Certificates will be awarded to the top two three teams, with additional prizes given only to the top-performing team.

Congratulations to the leaderboard toppers of the AI4VA saliency and depth challenges. The top 3 from the leaderboard are invited to submit their report and codes. Please email a zipped file or a GitHub repository with your code and technical report by Sept 20 (AoE) 11:59 pm. Upon review we will finalize all winners.

1. All posters should be portrait (vertical), with a maximum size of 90 x 180 cm.

2. Oral presenters have to also present their posters during the poster session. Oral presenters please make a presentation, in any template you like while following the standard Powerpoint aspect ratio, and send it to the AI4VA organisers email. You will have exactly 7 minutes to present (no exception) and a 3 minute Q&A.

3. Challenge winners (to be announced on 22nd September) please make a presentation in any template you like and send it to the AI4VA organisers email. You will have 5 minutes to present. 

All presentations from oral presenters and challenge winners should reach us by 27th September, 2024 11:59PM (UTC). Please do not send us your posters. Please print them prior to your arrival. If you want to print them locally in Milan you might do so too. However, please note that we do not have more information about this and cannot guarantee your posters will be printed on time if you choose to do so locally in Milan.