alt-text-backend's People
alt-text-backend's Issues
the list of images
Lines 298 to 319 in 7afe104
I think the list of images is enough.
push or pull?
Line 114 in 7afe104
you assume a "push" with a zip file. Why??? It's more efficient to send a URL and have the server pull in the file. This is because
the html is much smaller than most images, and you want to process images one by one. You don't need to get sound files, or css files or font files.
context rather description
Line 109 in 7afe104
maybe what you want is context rather than 'description', as some contextual info might be useful to the model. genre maybe? emphasis on helping the ai, not on discovery!
"Books"
Line 20 in 7afe104
it's worth noting that these objects are html pages, not really "books". "document" would have been a better word.
API user stories
When building software applications, I have found that it is useful to have user stories to guide the application design to help me focus on the important functions of the application.
User stories for alt-text AI API
-
As a client system, I have a document with images. I would like to get suggested alt-text descriptions for each of the images.
-
As a client system, I have received alt-text descriptions from you for a document with images . I would like a way to improve your suggestions.
linking of images to documents
Lines 298 to 304 in 7afe104
this bit seems to imply that you're doing something other than using foreign keys to link images to books.
...and we're keeping the images around, too? Maybe leave them where they came from, or at least have that option using redirects?
Lines 331 to 347 in 7afe104
sorting is expensive!
Line 46 in 7afe104
don't bother
what is a title?
Line 32 in 7afe104
where are you getting the title from? from a document, another metadat feed or from a PUT?
Why would we reanalyse?
Line 196 in 7afe104
We might add to context data to improve the result, or we might want to see the results of a newer model. In that case, wouldn't I want to compare the results? I think you need to keep track of sessions.
string or binary
Line 177 in 7afe104
This is probably just my lack of understanding of the spec semantics but what does it mean that a field is binary AND a string?
Authors are too messy to deal with
sorting is messy too
Line 55 in 7afe104
most applications would rather have fast results
maybe separate image from a session
Lines 464 to 496 in 7afe104
it seems to me you'll want the results of an analysis session separate from the image, espesially if you imagine re-running the model. you can store processing times (and dates) on the session object.
same question about title
Line 103 in 7afe104
titles can come from the doumnet, from the PUT or from an API. Which takes precedence.
Titles can have multiple uses, display, (mostly for human users) context, for the AI, and retrieval.
Maybe add a field for a user-generated document identifier to facilitate retrieval.
where is get by submitter?
Line 28 in 7afe104
also, seems to me I'd want a handle returned with each document I submit, and be able to review the result for multiple sessions
Are we keeping the book around?
Line 174 in 7afe104
you say I can overwrite a book. How is it identified? are you keeping it around. How many Terabytes will I need? (I've seen 100MB books!)
why a single book or document?
Line 95 in 7afe104
Why not a list of books?, with urls to pull from?
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. ๐๐๐
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google โค๏ธ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.