Image Description Week

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search


Image Description Week (16-22 May 2022)
The artist has placed frames around everyday objects—a banana, plant, and flask—and labelled them
Banana Plant Flask, Max Gruber, CC-BY 4.0
This image was produced as part of the Better Images of AI project to suggest the detection of objects by computer vision algorithms.

You’re invited to take part in a conversation linked to International Museum Day (18 May) and Global Accessibility Awareness Day (19 May). It’s a week-long call-to-action to make images more accessible and discoverable, in all languages.

Why does this matter? [edit]

One of the most significant internet trends has been the rise in image-led communication. Social media platforms are optimized for engagement and to maximize attention on both user-generated content and advertising. Since images have been found to drive attention, social media platforms have embedded tools to allow their users to more easily add and share images. [1] There are more than five million distinct images on English Wikipedia alone and research into reader interactions found that images on Wikipedia drive 10 times the engagement that citations do.[2] The same study concluded that readers use the images on page previews to decide whether to view a full article, and that readers are more likely to open the media preview of images on shorter or lower-quality articles. These findings suggest that Wikipedia readers use the images illustrating articles to fill information gaps.

Images on Wikipedia are used for more than just illustration, but less than half of them are contextualized by a caption that relates the image to the article, and only 10% have some form of alternative text (‘alt text’) that describes the visual content of the image for people with low or no vision. This is better than social media platforms like Twitter, where an estimated 0.1% of images have alternative text, but far from the 70% of images that are described with alt text on large commercial and news sites.[3] And other platforms are actively improving their alt text features. Most recently, Twitter implemented a prominent ‘ALT’ badge on images that have a description.

Our lack of image description not only limits the accessibility of content on Wikipedia, it also reduces the ease with which our open access images can be discovered through search. While Commons has grown into the world’s largest free-to-use media library—with more than 80 million illustrations, photos, drawings, videos, and music files—this media is not always easy to find and use, especially in languages other than English. But it’s not a simple case of just writing more descriptions. The descriptions we write need to be useful and ethical. Image description can touch on sensitive aspects of appearance and identity that, if described inappropriately or incorrectly, could cause harm to people who are already marginalized by society.[4] We want Wikimedians, museum professionals, and accessibility experts to come together to develop shared practices for image description so that:

  • Editors reflect on how they select and interpret images in different contexts.
  • Organizers understand image description as a creative task that can advance social justice goals.
  • Image uploaders make their image contributions count.
  • More images can be discovered and understood in multiple languages.
  • People with diverse backgrounds and identities feel a sense of belonging in Wikimedia spaces.

By contributing well-described images under open licenses, we can address knowledge, representation and access gaps on the Internet, in the places where the most people are trying to meet their information needs—Google search and Wikipedia.

Events [edit]

This first Image Description Week will introduce the current state of play: product pilots, research projects, and the tools available. We hope that you'll be inspired to include image description in your own events. If you do, please share your session below and let us know at glam@wikimedia.org

During Image Description Week, 16-22 May[edit]

Day and time (UTC) Duration Description Language(s) Joining instructions
Tuesday, May 17, 9AM UTC 30 mins First impressions of the Wikipedia Image/Caption Matching Challenge, by Miriam Redi English Zoom link:
https://wikimedia.zoom.us/j/89367546691
Tuesday, May 17, 4PM UTC 1 hour Five easy tools to improve descriptions of images on Commons: ISA Tool, Depictor, AC/DC, SDC, and Media Data Verification Tool. Demo & try it yourself. For beginners, by Sandra Fauconnier English Zoom link:
https://us02web.zoom.us/j/83992360504
Wednesday, May 18, 1PM UTC 30 mins Evaluation of captions written as a newcomer task, with Michael Raish (Lead Design Researcher) English Google Meet link:
https://meet.google.com/ysu-qzcj-ddp
Thursday, May 19, 10AM UTC 1 hour Sneak preview of the Wikistories product pilot, with Sudhanshu Gautam (Lead UX Designer) English Google Meet link:
https://meet.google.com/meh-yifn-rpj
Thursday, May 19, 12PM UTC 45 mins The basic routine of a wikipedian checking the images uploaded by others (steps to check the status of an image, fixing description, deletion practices, tools), with Florence Devouard -> Goals : to better understand part of the life of an image after you uploaded it (in particular when the original description work was a bit poor... to help you reflect on the best practices. Notes here: User:Anthere/Image Description Week English ZOOM meet link: https://us02web.zoom.us/j/84245950622
Thursday, May 19, 4PM UTC 1 hour Describe images on Wikimedia Commons in batch with OpenRefine: tutorial for beginners, by Sandra Fauconnier English Zoom link:
https://us02web.zoom.us/j/86721746604

After Image Description Week[edit]

Day and time (UTC) Duration Description Language(s) Joining instructions
Tuesday, May 24, 17 UTC 1 hour 30 min Conversations on the role of images in the construction of cultural narratives from a critical, contextualized and accessible perspective, by Wikimedia Argentina + DALAT + ICOM CECA LAC. Demonstration on Wikipedia's ALT text feature. Spanish and english live translation Video recording here.
Thursday, May 26, 13 UTC 1 hour 10 minutes Conversations around Zugasti Collection of indigenous related photographs and their descriptions, by Wikimedia Argentina and No'Lhametwet Indigenous Documentation Center. In addition, we will work with ISA Tool. Spanish and english live translation Zoom link here / Register here to save the date
Saturday, May 28 to Sunday, May 29 The whole weekend Using Wikimedia Commons with young volunteers to document and describe local cultural heritage within the frame of the National Cultural Heritage day, by Wikimedia Chile. Spanish Use the National Cultural Heritage day's category to upload local cultural heritage pictures and describe them.

Editing tasks to make images more findable and accessible [edit]

Wikipedia[edit]

  • Write a caption to contextualize the image within the article or section.
  • Add alt text to describe the visual content of the image.

Wikidata[edit]

  • Add an image to the relevant Wikidata item using the property image (P18).

Commons[edit]

  • Add a unique and descriptive title using plain language with spaces. If relevant, include the name of a person, place, or event. The first 20 characters will be shown on category pages, so put the most essential information there.
  • Add relevant structured data captions in as many languages as possible. This should be a short and simple description of the image. It will be saved as a Wikibase label and published under a CC0 (public domain) waiver.
  • Add a detailed description, describing what the file is about and any other relevant context. This is wikitext and can include mark-up.
  • Add the file to relevant categories. Use the most precise sub-category within each tree.
  • Add depicts statements for the items that are clearly visible in the image. Identify the main features of the image using the ‘mark as prominent’ option.
  • Nominate quality images, where appropriate.

Four challenges[edit]

Here are four challenges you can try straight away:

  1. Help describe the diverse folklore of Ukraine, using the ISA Tool
  2. Verify the items shown in images, using Depictor
  3. Write short captions for the winning images in last year’s Wiki Loves Monuments and Wiki Loves Earth campaigns
  4. Revisit a Wikipedia article you recently edited and check the alt text for the images. Can you improve it?

Working group [edit]

There can be a lot of confusion about the difference between the caption, description, alt text, and depicts forms of image description on our projects. If you want to work on defining these and providing good examples for people to follow, please join our working group. Simply add your username below and tell us what topics you're most interested in.


  • FRomeo (WMF): drawing on the visual literacy of museum staff, improving image accessibility
  • GFontenelle (WMF): I'm interested in all topics but especially alt text and depicts, to help improve image accessibility on the Wikimedia projects, so that people from diverse backgrounds can participate and images can be more discovered and understood in multiple languages.
  • Amuzujoe: I am interested in all aspects
  • Scann: idk if this is the place :) but I'm interested in exploring how sometimes images get used on Wikipedia in wildly inaccurate & inappropriate ways (i.e., pictures that depict smoking as cool despite WHO recommendations to avoid such depictions and other more problematic misuses of pictures inside Wikipedia).
  • SFauconnier: especially interested in making the data modeling of, and distinction between alt text, description, captions... clearer for Wikimedians, taking into account that these are often very context-dependent! Also interested in giving advice for product decisions in this area.
  • User:T Cells as the founder and global coordinator of the Wikipedia Pages Wanting Photos for two consecutive years, I am interested in learning how users can improve their image caption, and what we could do to support them. T CellsTalk 06:42, 22 May 2022 (UTC)Reply[reply]
  • User:Susan Tol Not sure if it's too late but I'd love to join the user group. Would love to include tags and captions with language that is from the community depicted (e.g. queer, indigenous) and allow for terminology to be updated.
  • Aliyu shaba I have a interested to learning how users can improve the image caption and how we support themAliyu shaba]]Talk 07:52, 5 July 2022 (UTC)Reply[reply]

Resources [edit]

Please add useful research papers, guidelines, and tutorials about image description.

Resource Format(s) Language(s)
Analyze a Photograph by National Archives Worksheet EN, ES
The Digital Image Guide (DIG) Method by Dana Thompson PDF EN
Guidelines for Image Description by Sina Bahram, Anna Chiaretta Lavatelli, and Cooper Hewitt Webpage EN
Veronica with Four Eyes alt text archives Webpages EN
Alt Text as Poetry workbook by artists Bojana Coklyat & Shannon Finnegan Website and PDF EN
Decolonizing the Internet's Structured Data by Whose Knowledge? Webpage EN
“It’s Complicated”: Negotiating Accessibility and (Mis)Representation in Image Descriptions of Race, Gender, and Disability by Cynthia L. Bennett and others PDF EN
Reimagine Descriptive Workflows: A Community-informed Agenda for Reparative and Inclusive Descriptive Practice by OCLC Webpages EN
Concadia: Tackling Image Accessibility with Descriptive Texts and Context by Elisa Kreiss, Noah D. Goodman, and Christopher Potts PDF EN
Estructurando datos para revertir brechas by Wikimedia Argentina Video ES
Texto alternativo: la guía definitiva by DALAT Desarrollo Accessible Latinoamericano Webpage ES

References[edit]

  1. Li, Y. and Xie, Y. (2019) Is a Picture Worth a Thousand Words? An Empirical Study of Image Content and Social Media Engagement Journal of Marketing Research
  2. Rama, D., Piccardi, T., Redi, M. et al. (2022) A large scale study of reader interactions with images on Wikipedia EPJ Data Sci. 11, 1
  3. Bennett, C. L., Gleason, C., Scheuerman, M. K., Bigham, J. P., Guo, A., and To, A. (2021) ‘‘It’s Complicated’’: Negotiating Accessibility and (Mis)Representation in Image Descriptions of Race, Gender, and Disability. CHI Conference on Human Factors in Computing Systems (CHI ’21)
  4. Bennett, C. L., Gleason, C., Scheuerman, M. K., Bigham, J. P., Guo, A., and To, A. (2021) ‘‘It’s Complicated’’: Negotiating Accessibility and (Mis)Representation in Image Descriptions of Race, Gender, and Disability. CHI Conference on Human Factors in Computing Systems (CHI ’21)