Researchers Are Making Memes Accessible to the Blind

Everyone loves a good meme. If you’re blind or visually impaired though, most internet memes are not accessible. Researchers at Carnegie Mellon University are trying to change that.


For the 1 million Americans who are legally blind, an increasingly large part of the internet is cut off to them.

A young woman who is visually impaired uses a refreshable braille display to access the web on her laptop.

A young woman who is visually impaired uses a refreshable braille display to access the web on her laptop. [Credit: BSIP/Getty Images]

Scroll through Twitter, Instagram, Facebook, or TikTok and you’ll find them littered with memes, coded images, and in-jokes utilizing pop culture references. Now imagine you’re blind. How would you see and understand them?

That’s a challenge that researchers at Carnegie Mellon University are trying to tackle. In a recent paper titled “Making Memes Accessible,” the group trained a system to classify and parse memes with up to 92% accuracy, making it possible for visually impaired users to share the joke, too.

It’s an unenviable task. “Memes are often in-jokes that are supposed to exclude outsiders,” says Scott Wark, a meme researcher at the University of Warwick. “They’re inaccessible by design.”

Many visually impaired people use screen readers and built-in accessibility tools within operating systems to help navigate the screens sighted people have no problem with. Such screen readers rely, in the case of images, on alt-text, or written descriptions of what an image depicts. The use of alt-text is limited at the best of times, and even more so in the world of memes.

“The stereotypical image we might have of a meme is the image with captions at the top and bottom,” says Wark. “But memes have gotten a lot weirder over the last few years. Many don’t really have a punchline like a joke does. A meme like “Is this a pigeon?” seems incredibly hard to parse as text, because it isn’t linear.”

Jemma Gilboy, a senior lecturer at Nottingham Trent University, is equally heartened by the research. “A program that can classify these [memes] with a 92% accuracy rate could be extremely useful for meme consumers with visual impairment,” she says.

Wark is happy to see researchers start to think more about how to make memes accessible. “The web — and memes — isn’t always accessible because it relies on users to generate content, for free,” he says. “Just as the web is poorly moderated because most companies have no incentive to do it better, online culture isn’t always accessible because this labor isn’t accounted for.”

That’s the fundamental reason the researchers decided to tackle the problem. “Back in the day, Twitter was all text when it launched,” says Cole Gleason, one of the authors of the paper. “It was pretty accessible for visual impairment. But the visual content on Twitter has been increasing over time — images, videos, GIFs, things like that.”

Read the full story about Making Memes Accessible at article source here.