DocSegTr icon indicating copy to clipboard operation
DocSegTr copied to clipboard

Large dataset and single image inference code

Open moured opened this issue 3 years ago • 4 comments

First of all, Thanks for the great work! I wanted to request if you can share minimal-PubLayNet or at least the code for the inference on a single image as an input argument. That is because the dataset is too large and I want to do a quick test.

Best, Omar

moured avatar Mar 22 '22 09:03 moured

I think the publaynet-minival in the author's code is the examples folder from https://github.com/ibm-aur-nlp/PubLayNet

skyerhxx avatar May 08 '22 11:05 skyerhxx

@moured You can try this notebook I created. You may have to edit some codes in detectron2 to make it work (covert to cpu I think).

fjbriones avatar Aug 01 '22 16:08 fjbriones

@moured You can try this notebook I created. You may have to edit some codes in detectron2 to make it work (covert to cpu I think).

exactly what I want! Thanks

moured avatar Aug 02 '22 10:08 moured

@moured You can try this notebook I created. You may have to edit some codes in detectron2 to make it work (covert to cpu I think).

Hello, because the link to the best model provided by the author is invalid. I am looking for it now. I don’t know if you have it here. If so, I hope you can provide me with one.

yexujian avatar Mar 09 '24 03:03 yexujian