Field is recognized by ML-Skill in evaluation run but not in UiPath Studio

[Daniel_Wentsch]

I created a machine learning skill for extracting data from a delivery bill. There are three different delivery bills and according to the evaluation run in the AICenter, all fields are recognized by the ML skill (Figure 1). If I use the ML-Skill in a program in UiPath Studio, however, the delivery note number is not recognized each time for a delivery note variant (Figure 2). Can someone tell me where the difference is between the extraction in AiCenter and in UiPath studio ?

[alpaca (Stefan Reutter)]

Possible reasons:

• you’re using an old version of the package for the ML skill
• you configured a different OCR engine or are using different settings
• the document is not actually in the evaluation set
• you didn’t map the field correctly in “Configure Extractors”