search

dhlab-epfl / dhSegment

Generic framework for historical document processing
https://dhlab-epfl.github.com/dhSegment
GNU General Public License v3.0
370 stars 116 forks source link

Layout Analysis Use Case: DIVA-HisDB #56

Closed doobybug closed 4 years ago

doobybug commented 4 years ago

Hi,

Are there any plans to issue the training and demo for layout analysis on DIVA-HisDB?

solivr commented 4 years ago

Hello,

No there are no plans to do a special example for the DIVA-HisDB dataset. It should be quite similar to the cBAD and the Page demo. If you're experimenting particular problems to have the training running, feel free to ask. With a bit of archeology you can also see what we had implemented here.

doobybug commented 4 years ago

Hi. thank you so much for answering me. I will definitely look into it. I would like to experiment with your code to classify text and handwritten text in the same document and maybe get their boundary boxes. Which route do you think I should take it? Page segmentation or Layout analysis?

solivr commented 4 years ago

Your task looks more similar to the DIVA-HisDB task than to the page segmentation task. So I would go more into the direction of document image analysis. If you need a guide for the principal steps you can have also a look at this notebook. However I would recommend in you case to try to annotate the text at pixel level as it is done for the DIVA-HisDB experiment (and not with bounding boxes around text).