drbh/candle-ocr-server

wip

Svelte

0

3 commits

updated Nov 21, 2023

See the code

README

candle-ocr-server

This is a small server and web app that leverages the open source transformer based OCR model microsoft/trocr-base-handwritten for optical character recognition from freeform text and photos.

This app is built on candle, poem, svelte and fly.

TLDR; 🕹️ Play with it now

Try it out at https://ocr.dholtz.com

**Note: the app performance may be impacted by other users since the model is running on a small CPU bound VM.

oooo, show me what it does 🚀

Detect from freeform

Note: the model has been trained on single line text and works best with similar input

handwritedetect

Select handwriting subset

Using select mode we can can analyze specific sections of handwriting

handwriteselect

Drop in a photo

In a more practical example we can drag in a screenshot.

dropfileselect

Select sections

and now we can select sections of text to run detection on

selectscreen

Contributors

drbh

3 commits

drbh/candle-ocr-server

wip

Svelte

0

3 commits

updated Nov 21, 2023

See the code

README

candle-ocr-server

This is a small server and web app that leverages the open source transformer based OCR model microsoft/trocr-base-handwritten for optical character recognition from freeform text and photos.

This app is built on candle, poem, svelte and fly.

TLDR; 🕹️ Play with it now

Try it out at https://ocr.dholtz.com

**Note: the app performance may be impacted by other users since the model is running on a small CPU bound VM.

oooo, show me what it does 🚀

Detect from freeform

Note: the model has been trained on single line text and works best with similar input

handwritedetect

Select handwriting subset

Using select mode we can can analyze specific sections of handwriting

handwriteselect

Drop in a photo

In a more practical example we can drag in a screenshot.

dropfileselect

Select sections

and now we can select sections of text to run detection on

selectscreen

Contributors

drbh

3 commits

Languages

Svelte

61.1%

Rust

24.8%

Makefile

6.3%

CSS

2.9%

Dockerfile

2.0%

JavaScript

1.6%