Announcing deprecation of MODI OCR engine and Create Tesseract OCR engine action in Power Automate Desktop

With the latest Power Automate Desktop May 2021 release, we have announced the deprecation of utilizing MODI OCR engine and Create Tesseract OCR engine action in Power Automate Desktop. Please review below to learn about the available replacements and more info about these deprecations.  

• MODI (Microsoft Office Document Imaging) OCR engine:  

Ability to initialize and utilize MODI OCR engine will be removed in desktop flows starting from  the July release of Power Automate Desktop.  

Actions required by you: If you are using the MODI OCR engine in your desktop flow, kindly  remove the create MODI OCR engine action and replace it with the Tesseract engine as  explained below. Please note that changing the OCR engine might cause the OCR actions to  fail, thus it is highly recommended to test the actions before moving the desktop flow to  production.  

• Create Tesseract OCR engine action: 

This action will be removed from the actions list in the August release of Power Automate Desktop. This is done to reduce an additional step of always initializing Tesseract OCR engine before being able to use in your desktop flow. This redundant separate action will be removed and now users can directly initialize the Tesseract OCR engine when it is being used. Having this ability also allows the desktop recorder to use OCR engine actions to extract text from images.  

Actions required by you: If you are using Create Tesseract OCR engine action in your desktop flow, kindly remove that action and directly initialize the Tesseract engine in the OCR related actions where the engine instance is used.  

Like in the example below, previously a separate action had to be used to create a new instance of the Tesseract OCR engine and then the output OCR engine instance variable was used in other OCR related actions. 

Now users can directly create the Tesseract OCR engine in all the other OCR related actions (“Extract text with OCR”, “If text on screen (OCR)”, “Wait for text on screen (OCR)” & “Move mouse to text on screen (OCR)”). Please note that with this replacement there will not be an OCR engine instance variable as an output and the engine will terminated after each action run. 

Other available alternatives for text recognition: 

• Other than the in-built Tesseract OCR engine action you can also use the cognitive operation actions in your desktop flow to recognize text. This includes both the Microsoft Cloud OCR and Google Vision API Text Detection action.  
• AI builder as part of a cloud flow on the Power Automate portal can be used to recognize text with pre-built AI models or even a new custom AI model can be created very easily that uses state-of-the-art Optical Character Recognition (OCR). Learn more about it here.