table-transformer
pix2struct
table-transformer | pix2struct | |
---|---|---|
9 | 5 | |
1,869 | 553 | |
8.3% | 4.5% | |
6.1 | 4.4 | |
5 months ago | 6 months ago | |
Python | Python | |
MIT License | Apache License 2.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
table-transformer
-
Data extraction from pdf
Saw this last time but never played with it https://github.com/microsoft/table-transformer
- FLaNK Stack Weekly 11 Dec 2023
-
[P] OCR + Table Extraction Advice
Have you tried the SOTA on Table Detection and Extraction with out of the box model weights?
-
How do you parse tables in PDF with langchain? Especially, the context which is few lines above and below the table.
https://huggingface.co/blog/document-ai https://github.com/microsoft/table-transformer https://github.com/google-research/pix2struct https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.6/ppstructure/table/README.md
-
[D] Unimpressive improvement in training speed after upgrading from GTX 980 Ti to RTX 4090
GPU is at 100% although a bit spiky (dropping occasionally to 50%), I expect that to be normal? I use the same configuration as the authors, num_workers is set to 1 https://github.com/microsoft/table-transformer/blob/main/src/structure_config.json . Data is on a separate SSD, C-drive is a NVMe SSD
- Microsoft TableTransformer
- DeepDoctection
pix2struct
-
[D] Is there a good ai model for image-to-text where the images are diagrams and screenshots of interfaces?
Here are a few useful resources you could start with: [Pix2Struct by Google Research](https://github.com/google-research/pix2struct) might be a valuable tool, although it will most likely need some fine-tuning to fit your specifics. You can also find some fine-tuned models on HuggingFace by searching 'pix2struct'. Another option worth considering is [DonutI](https://github.com/clovaai/donut). Like Pix2Struct, fine-tuning likely needed to meet your requirements. Tesseract OCR is another alternative, particularly for handling text. It's primarily designed for pages of text, think books, but with some tweaking and specific flags, it can process tables as well as text chunks in regions of a screenshot. Bit too much tweaking for my taste. As I'm also in search of OCR tools for UI and chart screenshots, so share if you find something else.
-
How would you go about driving contextual data from images?
For images with text, if you want to do visual qa, document classification, table/key information extraction, checkout https://huggingface.co/blog/document-ai https://github.com/philschmid/document-ai-transformers https://github.com/google-research/pix2struct https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.6/ppstructure/README.md
-
How do you parse tables in PDF with langchain? Especially, the context which is few lines above and below the table.
https://huggingface.co/blog/document-ai https://github.com/microsoft/table-transformer https://github.com/google-research/pix2struct https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.6/ppstructure/table/README.md
-
Advice building model for web elements/ browsing specific site
It is related to document AI. Recently google has released a model pix2struct. Some of the tasks they considered and datasets they used include:
-
Best Pipeline for pre-processing text from PDFs for fine-tuning LLMs
Google has a new model pix2struct that can accept image input(charts, documents, book covers, UI screenshos) and answer questions. https://github.com/google-research/pix2struct
What are some alternatives?
CascadeTabNet - This repository contains the code and implementation details of the CascadeTabNet paper "CascadeTabNet: An approach for end to end table detection and structure recognition from image-based documents"
PaddleOCR - Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)
blackmaria - Python package for webscraping in Natural language
FLaNK-Ice - Apache Iceberg - Cloud Data Lakehouse
camelot - Camelot: PDF Table Extraction for Humans
llama - Inference code for Llama models
web2text - Source code for the paper "Web2Text: Deep Structured Boilerplate Removal", full paper @ ECIR'18
document-ai-transformers
doctr - docTR (Document Text Recognition) - a seamless, high-performing & accessible library for OCR-related tasks powered by Deep Learning.
KaNiaFi - Processor and Service Implementation to manage Kanashi request response results using Apache NiFi or Niagara Files