Accessibility for the colour-blind
Commons has a lot of graphs and charts used on Wikipedia and elsewhere, but few consider how they look with colour blindness, mostly because the creator/uploader has no idea. m:Accessibility#Colour-blind-friendly images lists some tools that can be used to automatically transform images into how they are seen by colour blind people. We could run such automated tools on all Commons graphs and charts and reporting the results, ideally after assessing automatically in some way that the resulting images are not discernible enough, lower than some score. The warnings can be relayed with some template on the file description or directly to the authors and can havhe a huge impact on the usefulness of Commons media.
Depending on skills and time constraint, the project taker would do 1, 1-2 or 1-3 of these three steps: 1) develop the code for such an automatic analysis based on free software, 2) identify what are the images to check on the whole Commons dataset and run the analysis on it producing raw results, 3) publish such results on Commons via bot in a way that authors/users can notice and act upon.
Nemo bis added this.
GSoC / Outreachy
- Primary mentor: (Phabricator username)
- Co-mentor: (Phabricator username)
- Other mentors: (optional, Phabricator username)
- Skills: PHP or Python; perhaps ability to handle large datasets like media tarballs; colour blindness is a plus!
- Estimated project time for a senior contributor: 2-3 weeks
- Microtasks: T48921 or a simple pywikibot task like T57062, T57309 (to get familiar with uploading infrastructure and bots); or an accessibility issue if you're sure.