search

Psarpei / Multi-Type-TD-TSR

Extracting Tables from Document Images using a Multi-stage Pipeline for Table Detection and Table Structure Recognition:
MIT License
258 stars 51 forks source link

If you provide the training code? #1

Open BeyondYourself opened 3 years ago

BeyondYourself commented 3 years ago

About the model :Multi-Type-TD-TSR, whether you can share the implemtent code

as-stevens commented 3 years ago

This looks to be a promising work, thanks for making it open source! Is there a way to further fine tune the model on our own data set. Model training code, and if not , please share the architecture. Thank you

anishdulal commented 2 years ago

Could you please provide training code to train this model on our own data?

tucachmo2202 commented 2 years ago

They just use opencv technique so no training code here

as-stevens commented 2 years ago

@tucachmo2202 thank you for replying. I am talking about the table detection model, the RCNN based model. I was able to find the link to the public dataset(~20 gb), it was from some Chinese university. Though I have lost the link now :( Please let me know if you find the link, I would need to fine-tune/retrain the table detection model.

tucachmo2202 commented 2 years ago

@as-stevens, about task table detection. Their paper said that they use table bank dataset and Detectron to train detection model. As I experienced, you should create your own dataset for both detection and recognition for better performance.

as-stevens commented 2 years ago

@tucachmo2202 thank you for the dataset link, this is what I referred to earlier!