Automated Alt Text and Accessibility Auditing
A tool that crawls a website, finds accessibility failures, and generates candidate alt text for images that lack it, with a review workflow before anything is applied. Most alt text tools produce descriptions that are technically accurate and useless in context, which is the problem worth solving.
Intermediate
A small team, or one strong student willing to learn something new.1 semester, 2 students
5 to ship
3 optional extrasSuggested stack
What you should ship
- Crawler collecting pages, images and the surrounding document context
- Automated checks for a defined subset of accessibility criteria, with results mapped to the relevant guideline
- Alt text generation that takes surrounding page context into account, not the image alone
- Review interface where a human accepts, edits or rejects each suggestion
- Evaluation comparing generated alt text against human written alt text on at least 100 images, judged by blind or low vision reviewers if possible
If you have time left
- Distinguishing decorative images, which should have empty alt text, from informative ones
- Colour contrast and heading structure checks alongside images
- A report that prioritises findings by user impact rather than listing everything equally
The problem
Most of the web is inaccessible to screen reader users, largely through missing or unhelpful alt text. Automated description tools exist and mostly produce output like a person standing near a building, which is accurate and communicates nothing about why the image is on the page.
What you build
A crawler, a checker for a defined set of accessibility criteria, a context aware alt text generator, and a human review step. The review step is not optional; publishing generated descriptions unreviewed is how you make a site worse in a way nobody notices.
The insight worth building on
Good alt text depends on the purpose of the image in its context. The same photograph needs a different description in a news article, a product page and a biography. Feeding the surrounding heading, caption and paragraph into the generation is the difference between output that helps and output that fills a field.
How to evaluate it
Compare against human written alt text on a sample, and if at all possible get blind or low vision reviewers to rate both. Automated similarity metrics are close to meaningless for this task, and saying so in your report is a point in your favour.
Scope warning
Pick a subset of accessibility criteria and cover them properly. Attempting the whole specification produces a shallow checker that finds the same three things every other tool finds.
Ideas and guidance, not finished projects
These are project ideas and scoping guidance, published free for students to use as a starting point. I do not build, write, or sell final-year projects, and I do not complete coursework for anyone. Take an idea, make it yours, and build it.
Not sure this one fits you?
I help students pick an idea that matches their skills and their deadline, then scope it down until it is finishable. Guidance only, never done for you.
Stuck on something specific?
Writing only gets you so far. If you want an answer to your situation rather than the general case, book a session and we will work through it together. Every session is free; a few slots open each week.
Follow along
New writing, resources and project ideas land here first.