How to use table detection pre-trained weights to train a custom dataset

[Aditi81]

I have created a custom table detection dataset that has different class labels. Instead of training a model from scratch, I want to use the pre-trained weights for the table detection model trained on PubTables-1M. Please help me to proceed with this task.

[bsmock]

This should be possible with probably just a few changes to the code depending on how closely what you're trying to do matches what we did. You said you have a custom dataset for table detection with different class labels. So you have class labels besides "table" and "table rotated"?

[Aditi81]

I have different types of tables as a class.

In addition, how should I use the pre-trained weights for the table detection model trained on PubTables-1M?