Home | Notifications | New Note | Local | Federated | Search | Logout
Note Detail
Blurry Moon@sun@shitposter.world (2026-09-03 14:59:00)
open source OCR is terrible, I used Google Cloud to translate and that worked a lot better.
Reply
---Replies---
Q.U.I.N.N.@icedquinn@blob.cat (2026-09-03 15:01:02)
@sun tesseract isn't great. there are qwen models and such that do ok.
GLM has one that can read handwriting BUT it requires you to have already pre-processed the input with another model :/
Snacks@snacks@netzsphaere.xyz (2026-09-03 15:05:06)
@sun aren't there ai models that can do it? Iirc reading text was the first thing anyone did with artificial neural networks
mangeurdenuage :gnu: :trisquel: 🌿 :abeshinzo: :ignutius: :descartes: :stargate: @mangeurdenuage@shitposter.world (2026-09-03 16:43:40)
@sun
>open source OCR is terrible
>I used Google Cloud to translate
>to translate
What ? OCR isn't about translating.
All the free/libre OCR I know just work.