May Developer Challenge - SAP AI Services - Week 2

[noravonthenen]

CHECK OUT WEEK 3 OF THIS CHALLENGE

Thank you so much for your participation last week and the great results! I am so happy to see your drive and experimental mindsets! I also really love the inspiration I got from all your favorite recipes! I will definitely try them out! And I loved the M-AI Developer Challenge pun comment!  

This week we want to use the Document Information Extraction Python SDK to extract information from the same recipe that you used in week 1 using the custom schema you created in the Document Information Extraction UI. With the Python SDK you can implement end-to-end use cases and process documents on a large scale.

1. Use the Python IDE of your choice, create and activate a virtual environment and install the Document Information Extraction SDK: pip install sap-business-document-processing
2. Get your Document Information Extraction service key from your BTP trial account
3. You will need your Schema ID which you can find in the URL of the UI when navigating to the schema you want to use.
4. Send your document from week 1 to the Document Information Extraction instance via the Python script (you can use the default client)
5. To SUBMIT your result simply reply to this discussion by pasting an image with your code and returned json like this:

Additional information:

!!! EXAMPLE CODE !!!

Repository with more examples

Week 1 challenge

[geek]

Now that was an interesting challenge, learnt a lot of stuff. Changed the recipe but:

[harsh_itaverma]

Answer for Week 2: 

[Nagarajan-K]

Not a python expert. So, used the swagger UI to extract the same json result output of the recipe document already loaded in week 1 to the default client. Tried from curl as well. 

[xavisanse]

this time the result is perfect! 

[xavisanse]

by the way, today i changed a little bit the code and pushed again the same doc.  and still having some issues with the recognition. Could be interesting to now if there is an estimation of how many documents you have to train by template (i assume that depends of each one) but to get an estimation cheers

[noravonthenen]

@xavisanse As always with AI models, it depends on the document and the use case but the more documents you assign to your template with perfect extraction results the better 🙂 Also recipes are a particular difficult case as the layout can have immense differences! 

[M-K]

Here's my submission:

[Ruthiel]

Hello @noravonthenen !
The code:

And the output:

[MioYasutake]

My submission for week 2.

[gphadnis2000]

Here is my submission.

[Alpesa1990]

My submission for this this task.

Code:

Response:

[emiliocampo]

My submission:

I'm getting the same result as when using the DoX UI

[moh_ali_square]

Hello everyone, 

Thanks for the opportunity, I had an issue with default client, I had to change it to the current client c_03, and it workd 😊

[martaseq]

Here's my submission:

Instead of using all the header fields defined in my schema for the header_fields parameter, I tried to filter which fields I wanted to fetch, but it seems to make no difference. It seems to always gives me all the fields in the schema.

[jasperdebie]

Hi!

My submission for this week:

Even the optional ingredient is in there 😄

[Hira]

Hi, 

Apologies for delay, Due to this exercise, I have installed Python on my personal computer for the first time. And face few small challenges. But it makes me realize I should learn Python at least for basic task. 🙂

Attaching the results of this challenge.

[Vitaliy-R]

Welcome to the Python Clud @Hira 🐍

Hopefully, you decide to polish your Python skills with the next developer challenge next month

[sainithesh21]

Here is my Week 2 Submission

[CameronWilson]

Here is my week 2 Submission

[CameronWilson]

[Jordi_C]

Great learning experience!

[JoshuaLaw]

Was a great learning experience, thanks!

[Vitaliy-R]

I must say that 100% confidence of the AI model doesn't make me confident in the results 🤓

[Mikkelj]

My Submission.

2 weeks away from putting AI Expert on my CV 😉

[sabarna17]

🐍 asked for the recipe using sap_business_document_processing -