Hi, Lately I’ve been studying and testing the whole IntelligentOCR Document Processing Framework that UiPath offers.
After trying everything I found, I could not train an extractor because I could not find a “trainable activity”.
My question is how can I train my own model, I already have everything else the Train Extractors Scope asks for: Document Object Model, Document Path, Document Text and Human Validated Data.
Thank you in advance!
Any kind of help is appreciated.
This is my question too. Is it something provided for future activities?
I am also still waiting for any updates regarding trainable extractor activity… any updates?
As of now, there are no Trainable activities provided by UiPath. I have gone through the documentation https://docs.uipath.com/activities/docs/train-extractors-scope and it is suggested to create custom activity for training extraction scope.
For creating custom trainable activity you can used interface and class provided at https://docs.uipath.com/activities/docs/i-trainable-activity-interface
I followed the above-mentioned steps by @Akshay_Ahire. I created a Custom Trainable activity and used that activity in UiPath but nothing happened. It’s is also not showing any error and the 0% improvements in extracting PO number and extracting Invoice items.
Any idea how long it will take to train the bot to extract the accurate results?
I implemented the “TrainableExtractorCodeActivity” Class fro custom trainable activity.
I’m attaching Files please check if you can find any issue.
InvoicesReceipts Extractor: UiPath Solution => Invoice.xaml (Working on this file)
Trainable Custom Activity: Visual Studio Solution for custom activity
[InvoicesReceipts Extractor.zip|attachment]Trainable Custom Activity.zip (2.2 MB) (upload://wBeLXBOzMFVH2o4pDEz00mEq4kn.zip) (1.1 MB)
I have also developed a BOT with ML Skill used as AU Invoices. After a few runs again there is no improvements with intelligent OCR. can someone please suggest.
I am on enterprise version with all valid licensing.