Receive an error when you try to access the AI links from the browser

Hello @Slimane_OUZIALA

All data retrieved by the service is available through the Machine Learning Extractor. You cannot modify any data or access the service in any other way.

Thanks.

1 Like

Thanks @Ioana_Gligan! Sorry I didn’t not notice that. Will this limitation be removed later? Also, we are looking for on-premise ML model to test our production data set. When can we expect on-premise model?

Thanks,
Ashok

2 Likes

@loana_Gligan

Thank you for your detail
Other question, Can I create a service and call it via ML?

Thank you

1 Like

Hi @ashoks93, we are aiming to release this on premise by August.

Thank you,
Alex.

1 Like

Hello @Slimane_OUZIALA,

The recommended way to go about it is:

  • create your own ML service according to your needs
  • create a custom activity by using the UiPath.DocumentProcessing.Contracts nupkg which contains both the object definitions as well as abstract classes to implement, for creating your own classifier or extractor.

This way, you can have any request / response format you want, deploy anywhere etc.

As a “starter’s kit”, please see the following:

Have fun!

4 Likes

@loana_Gligan,

Thank you for detail.
I will try you method
I will tet you knew about .

Thanks

1 Like

Hello

Can you help me how to add this extractor into my existing code?
or how to import package.

In my project, I couldn’t find any trainable activity support to Train Extractors Scope activity.

2 Likes

Thanks @alexcabuz !

We have use case where we need to extract line items from vendor statements. Can invoice extraction ML model extract line items in statements also? If there are any columns its missing, is there a way we can train the model to extract additional required column?

Thanks,
Ashok

1 Like

hello @harsha_vardhan,

Try to have a look at this: https://activities.uipath.com/docs/trainable-extractor-code-activity-class and implement such a training activity. it should be complementary to the extraction activity you want to train.

Try putting it in a Train Extractors Scope activity, available in the latest IntelligentOCR package.

1 Like

Hello @Ioana_Gligan

I couldn’t make it can you please guide where I need to add these classes/activities.

Thanks!

1 Like

Hello @harsha_vardhan,

You will need to build your own custom activity that implements one of the abstract classes in the UiPath.DocumentProcessing.Contracts public package.

Something like https://activities.uipath.com/docs/trainable-extractor-code-activity-class .

To understand how to use it, maybe you can have a look at a sample project building custom activities for data extraction, which is here: GitHub - UiPath/Document-Processing-Code-Samples: Code samples for document processing activities. .

The sample project only implements an Extractor activity, not a TrainableExtractor (which means a training activity for an extractor), but you will see how to build your own custom activity and how to use the InArguments, and what each contains.

As a note, to clear any confusion, there is NO training activity for the Machine Learning Extractor published in preview. If you want to build your own Machine Learning extraction algorithm, you will need to build two custom activities: an extractor and a trainer.

1 Like

Oh got it, thanks for the quick response.

3 Likes

Hi @alexcabuz are the limitations with regards to document size and number of pages will be increased/removed when this goes to production?

Hi @bazanl,

Yes, absolutely. These are in place only for Community Public Preview.

Best,
Alex.

Hello,
During the execution of Machine learning extractor activity an error message pops up which says "An error occured while sending the request.
can anyone help me with this?
Thanks in advance

I cannot access the following packages within UiPath Studio when it tries to connect to the UiPath servers.

Invoices: https://cv.uipath.com
Receipts: https://invoices.uipath.com
Computer vision: https://receipts.uipath.com

Proxy related issue maybe? I can access cv.uipath from the browser but not within the UiPath

I get an error from the machine learning extractor saying there was an error sending the request.

Any suggestions?

Hello @AdiPopa,

I have the opposite problem. I cannot access the following packages from within UiPath Studio as it tries to connect to their corresponding services on the UiPath servers on the Intenet.

-Invoices: https://invoices.uipath.com
-Receipts: https://invoices.uipath.com
Computer vision: https://cv.uipath.com

The reason is I get a machine learning extractor error: An error occurred while sending the request.

Could this simply be a proxy server issue?

Hello @alejohnson
Are the activities not working for you, or is accessing the links directly from the browser not working?

Please make sure that, directly from activities, you are setting the right ApiKey to both cases (CV or the Machine Learning Extractor).

Please give us more details (a screenshot, detailed error message etc) to see if we can reproduce your issue.

Thank you,

Ioana

Hello all,

Here is a new sample workflow, with the new IntelligentOCR and Machine Learning Extractor functionality in place.

Don’t forget to put in your own ApiKey for the Machine Learning Extractor :slight_smile:

Hi @alexcabuz,

We are still waiting on the on-premise version. Is there any updates on this.

Thanks,
Ashok