# Serialize bank statements from PDF to CSV

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fserialize-bank-statements-from-pdf-to-csv%2F19728%3E%5C
**Category:** Beginners\
**Created:** [June 28, 2022, 3:26pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fserialize-bank-statements-from-pdf-to-csv%2F19728 "2022-06-28T15:26:25Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![PackElend](https://avatars.discourse-cdn.com/v4/letter/p/90ced4/32.png) [@PackElend](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FPackElend)%5C
**Post date:** [June 28, 2022, 3:26pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fserialize-bank-statements-from-pdf-to-csv%2F19728%2F1 "2022-06-28T15:26:25Z")

</div>

Hi there,  
to get all my financial history into Firefly III I have to extract the tables in my bank statements, which are available as PDFs, to a CSV.

I started once to build an app with _Camelot: PDF Table Extraction_ similar to [A table detection, cell recognition and text extraction algorithm to convert tables in images to excel files | by Hucker Marius | Towards Data Science](https://towardsdatascience.com/a-table-detection-cell-recognition-and-text-extraction-algorithm-to-convert-tables-to-excel-files-902edcf289ec) but covering all cases it just a nightmare.

Some weeks ago I read [Structured Approach to Extract Information from Unstructured Documents | by Md Sharique | Analytics Vidhya | May, 2022 | Medium](https://medium.com/analytics-vidhya/structured-approach-to-extract-information-from-unstructured-documents-7013d1e83b04) (and others like MaskR-CNN, YOLO).  
The approach there sounds quite promising but I would like to verify if that is reasonable to persuade this idea as accuracy has to be somewhere close 100% (I still can add cross-check sum of all items against the given total and add/modify things manually)?  
If so, how to ensure confidential training as I don’t want to share my statements with a cloud?

Would very kind of some advice would be given.  
thx  
stefan

---

<div class="post-metadata">

**Author:** ![PackElend](https://avatars.discourse-cdn.com/v4/letter/p/90ced4/32.png) [@PackElend](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FPackElend)%5C
**Post date:** [July 6, 2022, 7:42am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fserialize-bank-statements-from-pdf-to-csv%2F19728%2F2 "2022-07-06T07:42:51Z")

</div>

found some resources on TAPAS

- [machine-learning-articles/easy-table-parsing-with-tapas-machine-learning-and-huggingface-transformers.md at main · christianversloot/machine-learning-articles (github.com)](https://github.com/christianversloot/machine-learning-articles/blob/main/easy-table-parsing-with-tapas-machine-learning-and-huggingface-transformers.md)
- [Guide To TAPAS (TAble PArSing) - A Technique To Retrieve Information From Tabular Data Using NLP - Analytics India Magazine](https://analyticsindiamag.com/guide-to-tapas-table-parsing-a-technique-to-retrieve-information-from-tabular-data-using-nlp/)

Is this the right way to go?