Image Captioning with Huggingface.js

@HuggingFace [0] is the collaboration platform for the machine learning community.

Putting this together with Supabase Storage[2] and Database Webhooks[3] we can easily put together a service that automatically generates captions for any image we upload to a storage bucket.

## Chapters
00:00 Intro to Hugging Face.
03:05 Image Captioning.
05:20 Supabase Storage Metadata tables.
06:35 Create a new bucket called `images`.
08:10 Generate TypeScript types from remote Database.
11:04 Implementing the Database Webhook logic.
12:05 Create Database Table to store captions.
14:30 Create signed URL for images in Supabase Storage.
21:02 Write the caption to the Database table.
22:30 Deploy the Supabase Edge Function.
23:35 Set up the Database Webhook.
24:23 Demo time

[ThomasWapps]

Awesome! the more I use supabase, the more I love it

[xiaohaihuang]

I think you should check if the inserted object is coming from the 'images' bucket. Because the 'objects' table contains objects from every bucket.

[hamishirving]

Cool! Is there any caching in functions? Is it loading the hf package on each run?

[zacklau]

not really standing on the grass ...maybe is just the model problem... ?