Receipt and Invoice AI - Now available in Public Preview!

Hi,

Many thanks for the detailed description, Tried to create the same workflow but here is a catch.

I have three files added to my folder but still getting error as “Input document not found”.
why is this so?

Hey @satish5.prasad

You have a folder called docs here. And in the file path, I dont see a slash after the docs folder… add it and it will be all good

Hi Fernando

Thanks for the quick response, The previous error seems resolved. Now its throwing error as below

“RemoteException wrapping System.Exception: An unexpected error has occurred —> RemoteException wrapping UiPath.SmartData.Digitization.Tokenization.TokenizationException: Invoice _ 987654.pdf —> RemoteException wrapping System.InvalidOperationException: This build of Docotic.Pdf library requires maintenance through 11 July 2019.
Your license includes maintenance through 14 June 2017.
Please update your license at http://bitmiracle.com/purchase.aspx

Does the Digitize document activity require purchasing any licence, I hope not. How could I fix this?

Thanks
Satish

It is not related to digitize document, it is related to Machine Learning Extractor. It doesn’t need any license, However, it requires an API key which should be obtained from the Orchestrator.

Login to orchestrator cloud platform (if you are using the community), and under licenses, you will see a section where you have API keys for invoice processing, computer vision, and invoice processing. Get the key from the required category.

Under machine learning extractor, there is a property where you have to add the API key. Paste the API key there along with the endpoint URL.
end point URL for invoices is

https://invoices.uipath.com
for receipts
https://receipts.uipath.com

Setting this property will resolve your error. Also make sure to upgrade the UIAutomation activities to a newer version…

Hi Fernando,

Done with the steps indicted above, however still getting the same error message.

RemoteException wrapping System.Exception: An unexpected error has occurred —> RemoteException wrapping UiPath.SmartData.Digitization.Tokenization.TokenizationException: Invoice _ 987654.pdf —> RemoteException wrapping System.InvalidOperationException: This build of Docotic.Pdf library requires maintenance through 11 July 2019.
Your license includes maintenance through 14 June 2017.
Please update your license at http://bitmiracle.com/purchase.aspx screenshot

UI automation activities updated.

What should I try next here?

Thanks
Satish

Hello @satish5.prasad

In the machine learning extractor activity, can you see a property called “API Key” and “End point”? You should update those properties as I explained in the previous post…

check whether you can locate those two

Yes, it was all set as you explained, Here is the snapshot.screenshot

i was confused On validation . How can increase the Percentage in validation . And it is asking validation for every doc and for every run .

Is there any syntax for using the fields under item;For example how to use receipt field “line-amount”?

Hello @satish5.prasad,

Please upgrade the IntelligentOCR package and the MachineLearningExtractor package to the latest version so that this error goes away.

Please let me know if it works.

Thanks,

Ioana

Hi @alexcabuz
This is a very good addition. I have been running the taxonomy manager for invoice processing and found that the MLExtractor is running is not learning the post processing the invoices.
E.g. if some field is not extracted and if we manually map the same. ideally it should be able to learn but its not happening so.

In data extraction scope if we set the threshold of confidence score and each field extract from the invoice is above the set threshold then invoice should ideally be a straight through processing but it is not happening. Can you please help.

Also if field name E.g. Invoice Number is in different names like Invoice No, Invoice, Folio No. How can you handle this? Do you have any NLP running behind?

Can you please give some pointers.

Hello @amitbhatt55,

the Machine Learning Extractor uses a pre-trained model not open for training by the community.

We will keep you posted when this feature will be available!

Thanks,

Ioana

1 Like

Hi @Ioana_Gligan, Thanks for your reply. Is it fair to assume that the predefined Taxonomy is trained for Invoice and Receipt for community edition. Let us know.

What is specification is required to onboard onpremise MLExtractor

Thanks,
Amit

Yes.

@alexcabuz can give you more insight on how to access the on-prem version.

Thanks,

Ioana

1 Like

I followed the instructions and tried to install UiPath.MachineLearningExtractor. But it looks like the activity is not available in Uipath Beta Feed. Could you please confirm the same.

2 Likes

Hey @Ioana_Gligan

Just like @Anamika15 mentioned, we cannot find the MachineLearningExtractor activity pack in beta or any other feed. I also checked now and it is the same for me… Any idea why?

1 Like

Hi @Anamika15 @Lahiru.Fernando

We looked into it, and it seems like there was an issue with myget feed that caused it to not display when you search for it.

It does show up when you scroll down the list though:

2 Likes

Thank you and wish you a very happy birthday!!:birthday:

2 Likes