Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Increase the robustness for "large" PDFs #16

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
lolipopshock merged 4 commits into from
Apr 20, 2022
Merged

Increase the robustness for "large" PDFs #16

lolipopshock merged 4 commits into from
Apr 20, 2022

Conversation

@lolipopshock
Copy link
Collaborator

@lolipopshock lolipopshock commented Apr 11, 2022 •
edited
Loading

The fix tries to improve the robustness of the VILA library for "large" PDF -- the width or height of the PDF is more than 1000, and it has tokens with bounding box dimensions larger than 1000. In this case, the input will break the 2D position encoding process used in the base Transformer models, which is fundamentally a lookup table (bbox dimension value -> some embedding values) that only takes input from 0~1000.

I added a normalize function to solve this issue. When the input PDF size is "large" (i.e., either page_width>1000 or page_height>1000), it will normalize all the tokens in this page using the normalize_bbox function that coverts the dimension to the range 0~1000.

However, this solution is not perfect ~ our models hasn't been appropriately tuned for this large PDFs. Ideally, we should retrain such models with normalized inputs.

It will lead to one API change:

import layoutparser as lp # For visualization 

from vila.pdftools.pdf_extractor import PDFExtractor
from vila.predictors import HierarchicalPDFPredictor
# Choose from SimplePDFPredictor,
# LayoutIndicatorPDFPredictor, 
# and HierarchicalPDFPredictor

pdf_extractor = PDFExtractor("pdfplumber")
page_tokens, page_images = pdf_extractor.load_tokens_and_image(f"path-to-your.pdf")

vision_model = lp.EfficientDetLayoutModel("lp://PubLayNet") 
pdf_predictor = HierarchicalPDFPredictor.from_pretrained("allenai/hvila-row-layoutlm-finetuned-docbank")

for idx, page_token in enumerate(page_tokens):
    blocks = vision_model.detect(page_images[idx])
    page_token.annotate(blocks=blocks)
    pdf_data = page_token.to_pagedata().to_dict()
    predicted_tokens = pdf_predictor.predict(pdf_data, page_token.page_size) #<---- you need to specify the page size in the predict function! 
    lp.draw_box(page_images[idx], predicted_tokens, box_width=0, box_alpha=0.25)

@lolipopshock lolipopshock mentioned this pull request Apr 11, 2022
Closed
@lolipopshock
Copy link
Collaborator Author

lolipopshock commented Apr 11, 2022 •
edited
Loading

Now VILA works for large PDFs like poster: page_size is (2304.0, 2448.0) for the example below.

image

yoganandc
yoganandc reviewed Apr 11, 2022
View reviewed changes
src/vila/predictors.py
Comment on lines +50 to +51
# Right now only execute this for only "large" PDFs
# TODO: Change it for all PDFs
Copy link

@yoganandc yoganandc Apr 11, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

you're doing this because there isn't a retrained model at this time, correct?

Copy link
Collaborator Author

@lolipopshock lolipopshock Apr 11, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, that's correct!

@lolipopshock
Copy link
Collaborator Author

lolipopshock commented Apr 19, 2022

For further reference, when we merge this issue, we'll also release v0.3.0 of vila due to changes in API.

yoganandc
yoganandc approved these changes Apr 19, 2022
View reviewed changes
Copy link

@yoganandc yoganandc left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@lolipopshock let's merge, i verified that master crashes with a big pdf and that this branch doesn't.

@lolipopshock lolipopshock merged commit 66b090f into master Apr 20, 2022
@lolipopshock lolipopshock mentioned this pull request Apr 22, 2022
Merged
@lolipopshock lolipopshock deleted the simple-large-pdf-fix branch June 29, 2022 07:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

1 more reviewer

@yoganandc yoganandc yoganandc approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@lolipopshock @yoganandc