Welcome to the first ever workshop on shortcomings in vision and language (SiVL)!
The final program is now available here.
It has been great to be together and speak about each other work. Here is a summary of our stimulating day. Thanks to all the participants who made the workshop a such successful event!
[NEW!] LOCATION: The workshop will be held in the room "Audimax 0980", Technical University of Munich (TUM). Please note that the workshop is held at a different location than the main conference. Directions to getting there can be found here.
The primary purpose of this workshop is to bring together researchers at the intersection of vision and language to discuss shortcomings of modern approaches, tasks, datasets, and evaluation metrics for problems including image and video captioning, visual question answering, visual dialog, activity recognition, image retrieval and referring expressions. By highlighting common shortcomings in these domains, the workshop aims to facilitate discussion of novel research directions and to steer the community towards high-level challenges affecting the vision and language community broadly.
We have an exciting lineup of invited speakers from a diverse set of disciplines related to vision and language. In addition, we are also hosting the organizers of the Visual Dialog Challenge 2018 who will announce the results of their challenge. Visual Dialog is a novel task that requires an AI agent to hold a meaningful dialog with humans in natural, conversational language about visual content. For more information and to participate in the challenge, please visit their challenge page.
We call for papers that focus on analyzing the strengths and weaknesses of current models, datasets, and evaluation metrics as well as novel tasks proposed to overcome or measure current shortfalls.
Registration deadlines: Early registration till 31st of July (expired), Workshop-only early registration till 10th of August, regular registration till 26th of August.
We hope to see you in Munich!
Call for Abstracts and Papers
We call for papers and abstracts exploring shortcomings in current vision and language models covering topics including but not limited to
- Analysis of current vision and language models
- Analysis of current tasks and datasets
- Novel evaluation metrics
- Novel vision and language tasks
- Other pertinent work about shortcomings of vision and language
Papers must describe substantial, original, completed and unpublished work. Paper submissions must be anonymous and will receive at least 2 peer-reviews. They may consist of up to six (6) pages of content, plus unlimited pages for references. Papers may be supplemented with up to two (2) pages of appendices included in the same file as the paper, before the references. Camera-ready versions of accepted papers will be given two additional pages of content (up to 8 pages plus 2 pages appendices) so that reviewers’ comments can be taken into account. They will be presented both in the Spotlight talks and the Poster session. They will be published by Springer after the workshop.
Abstracts can describe work in progress, work under review, accepted to be published elsewhere or already published work. They may consist of up to two (2) page of content, plus additional pages for references. Abstracts will be evaluated for topical relevance by the workshop organizers. Accepted abstracts will be posted online on the workshop website and will be presented both in the Spotlight talks and the Poster session.
Dual submission policy: We welcome dual submissions for both papers and abstracts i.e., work that is/will be on review elsewhere. However, paper submissions must not be published elsewhere and the authors must choose a single venue if accepted in multiple venues. Abstract submissions have no such restrictions.
Submissions should follow the ECCV guidelines:
- LaTeX Templates (zip): eccv2018kit.zip
- Example submission paper with detailed instructions: eccv2018submission.pdf
Additional instructions for the two (2) page Abstract submissions only:
- Authors' names and affiliations should be included in the submitted pdf
- Please do not include a short abstract and keywords
Submission is electronic, using the conference management system.
- Paper submission deadline:
July 9, 2018July 12, 2018 at 12:00 PM in CET
- Abstract submission deadline: July 30, 2018
- Paper and abstract notification: August 4, 2018
- Abstract final version upload: August 25, 2018
- Workshop date: September 8, 2018
- Paper camera-ready: September 30, 2018
All deadlines are 11:59 PM in CET timezone.
- Aaditya Prakash, Brandeis University
- Akos Kadar, Tilburg University
- Albert Gatt, University of Malta
- Aurelie Herbelot, University of Trento
- Carina Silberer, Universitat Pompeu Fabra
- Christopher Kanan, Rochester Institute of Technology
- Damien Teney, The University of Adelaide
- Elia Bruni, University of Amsterdam
- Enver Sangineto, University of Trento
- Ivan Vulić, University of Cambridge
- Jasper Uijlings, Google Research
- Jin-Hwa Kim, Seoul National University
- Katerina Pastra, Cognitive Systems Research Institute
- Lucia Specia, University of Sheffield
- Luowei Zhou, University of Michigan
- Manoj Acharya, Rochester Institute of Technology
- Mario Fritz, Max-Planck-Institut für Informatik
- Ranjay Krishna, Stanford University
- Sandro Pezzelle, University of Trento
- Subhashini Venugopalan, Google Research
- Timothy Hospedales, University of Edinburgh