Yeah definitely need that. Would be cool to see a nip for this for a place to put it in the tags
Discussion
Exactly what I was asking about the other day. Someone please build this!
nostr:note1xf55kkyk8khehc8ha2kxp9jc3dv9huwfq3n7qpv99wkq7mllaugshgh2jx
Great suggestion #[3] !
Added a WIP feature request for image accessibility tags:
https://github.com/damus-io/damus/issues/980
#[5] #[6] can yall
Work on a accessibility image tag NIP 🙏
cc #[7] #[8] will add to Iris, Amethyst github shortly
#[9] what is missing from the above feature request?
We have such an awesome community of problem-solvers. I know we can get this done. 🙌💜
You’re definitely going the best route by asking for/gaining routine feedback from #[2] and any other users that identify as blind or have degrees of blindness foremost.
The NIP as you have it works though 💜 It would say Image Description and a user could drop in an image description. Could even be shorted to ID as the text; people either know or will gain understanding as to what it means.
In the interim, users aiming to make their posts most inclusive can type ID or Image Description above their photo and a screen reader will pick up the visual.
Separate/along these lines: users can Camel Case their hashtags to make them easier to read for people living with low vision or conditions like dyslexia. Like: #ImOnNostrNow v #imonostrnow
> It would say Image Description and a user could drop in an image description
Is “user” here the one that is uploading the image?
> Could even be shorted to ID as the text
By “ID” do you mean the image name (for instance “funny-cat.jpeg”. without the suffix)?
CamelCase hashtags is a new one for me. Good suggestion - I’ll request we can automate this.
Ah cheers!
Yep, the user is the person uploading the image.
ID stands for Image Description. Just a shorter way to write it. So in #[8]‘s example, the text box would have a background prompt of either ID or Image Description =>
then the person uploading the photo adds their description of the photo.
Word, that’d be cool! 💜
What if user does not want to type an image description?
Would it be acceptable to have the image file name as a default/starting point as ID?
I know sometimes image names are random characters. Would have to think through this.
Yeah, only trouble would just be if the image file is mislabeled. May get more pushback for tons mislabeled than just the option of individuals adding IDs.
Definitely overdue in general though and still not in broader consciousness. I think Twitters was like .3 percent of people using IDs when they had metrics/A Director of Accessibility for a shortwhile. Other sites don’t really even track stats yet.
This is a pretty big one. What do you think #[7] ?
Are there any open source ML image to text synthesizers/characterizers we could use?
I don’t know about open source but Google ML Kit supports image labeling for 400+ categories. https://developers.google.com/ml-kit/vision/image-labeling
Actually, iOS supports the same functionality out of the box. https://developer.apple.com/documentation/vision/classifying_images_with_vision_and_core_ml
The biggest challenge would be creating a database of labels for all images being uploaded, currently there is no such infrastructure. I don’t think we can put a proper description in the file_name, especially since we use it to hash.. Unless the app can do it real-time?
Another challenge is the cost to use the labeling software.
I do love the idea and would love to implement somehow.
Wow that sounds like a real challenge. I would be happy to have some sort of pdf that tells me what each button on the damus app does so I don’t feel like a 2 year old just pressing butttons and seeing what happens and hoping I don’t do something seriously wroing. this sadly a methood I have had to use on many apps I have encountered.
#[9] which app that you have tested (if any) labels the buttons in a useful manner for you?
Could it be added to the image metadata (Exif)?
I’m no expert, but EXIF specs have a tag labeled 0x9286 UserComment that might be possible to use. However, I don’t think it supports modern character sets so it’s probably too limiting to work with. Someone more knowledgeable can chime in. https://exiftool.org/TagNames/EXIF.html
There are also different places/ways it is stored from one image type to another.. probably easier and more accurate to make a database.
#[2] 🙏