search

ryouchinsa / Rectlabel-support

RectLabel is an offline image annotation tool for object detection and segmentation.
https://rectlabel.com
504 stars 73 forks source link

Using CoreML models for OCR in images #212

Closed Mast3rChief closed 1 year ago

Mast3rChief commented 1 year ago

Hi,

do you have a "best practice" for OCR in RectLabel? Could be bounding box based or segmentation.

Use Case: I would like to label/detect large letters and ID on assets in my images.

Thanks a lot for your help.

ryouchinsa commented 1 year ago

In the latest version 79, you can use "Automatic text recognition using OCR" feature on Core ML menu. Changing the confidence threshold and the language to be recognized, you can adjust the auto labeling results.

We hope this feature will help your labeling work. Please let us know your opinion.

ocr

Mast3rChief commented 1 year ago

Sorry for the late feedback. Thx…. Works great

ryouchinsa commented 1 year ago

If you have any other requests, please let us know.