search

microsoft / table-transformer

Table Transformer (TATR) is a deep learning model for extracting tables from unstructured documents (PDFs and images). This is also the official repository for the PubTables-1M dataset and GriTS evaluation metric.
MIT License
2.02k stars 232 forks source link

Fine-tuning Tutorial #72

Open Iron-man-0 opened 1 year ago

Iron-man-0 commented 1 year ago

I'm confused on how to fine-tune the model on custom dataset for table structure recognition. I had few questions regarding fine-tuning process.

  1. What should be the folder structure for the dataset.
  2. Also, I tried to execute main.py file and got the following error.

image

It would be really helpful if someone provide a fine-tuning example on any sample dataset. Thank You so much. cc'ing @bsmock for visibility

NielsRogge commented 1 year ago

Hi,

See this link for all tutorials: https://github.com/microsoft/table-transformer/issues/68#issuecomment-1282409585

sanprit commented 1 year ago

@Iron-man-0 Have you got any tutorial to fine-tune the model? If yes, please share the link or answer this question please:

  1. What should be the folder structure for the dataset
SentinelReacon commented 1 year ago

@NielsRogge Hi, I was trying to train the model for table detection using the procedure given in the README. But after following all the procedures I am getting and error RuntimeError: Directory 'static/' does not exist. I am attaching an image as well please help me put here. Screenshot 2023-06-27 121016

isarker commented 1 year ago

This (https://github.com/microsoft/table-transformer/issues/125) might help you.