Mark up text as instance segmentation or semantic segmentation?

Question

I have data with a page of handwriting and I want to segment each word individually. If I mark each word for semantic segmentation, then the model will impose a segmentation mask on each word separately, or do I need to mark up instance segmentation for this?

Answer 1

this can be of your help.

https://www.kaggle.com/code/dmitryyemelyanov/receipt-ocr-part-1-image-segmentation-by-opencv

you need to further extend it for words.