Matching the Table Transformer in Hugging Face

#11
by Iamhereforever - opened

Hi,

I am unable to replicate the results from the Table Transformer available in the Hugging Face webpage. I don't get the same results when I use the very straight forward code available in the documentation, I oftentimes get not as good results.

Is there any additional pre-processing done on the image in Hugging Face? If so, what are they?

Thanks,

Microsoft org

Hi,

Refer to this notebook to get one-on-one matching results with the Microsoft implementation (this was manually verified).

@nielsr

still some issues with that notebook code. this image produces no detections on hugging face page but does produce detections using notebook code. obviously it is not a table.

12-29-2023 2-51-54 PM.png

Sign up or log in to comment