search

DevashishPrasad / CascadeTabNet

This repository contains the code and implementation details of the CascadeTabNet paper "CascadeTabNet: An approach for end to end table detection and structure recognition from image-based documents"
MIT License
1.46k stars 422 forks source link

match table caption #92

Closed Sunnycheey closed 3 years ago

Sunnycheey commented 3 years ago

Hello, first thank you for your wonderful work.

I wonder How can I detect the caption of the table and match it back.

May I need construct a brand new dataset contains not only table but also caption and train your model on it?

AyanGadpal commented 3 years ago

Hey @Sunnycheey , Surely that is one way you can solve it, but I would suggest using heuristics on the detected table bounding box to detect the caption (assuming that caption is present either at the top or below the table). In the well-formatted document, the distance between caption and table will be less than other text. However if you are considering to generalize it across a wide variety of documents, heuristics won't give an accurate result as training the model would.