How do I know who and where OCR is being processed?

Help
, ,
1

I recently had a discussion with another RPA developer about OCR processing. I assumed that since Google and Microsoft offered OCR processing, this was done in their datacenters, more commonly referred to as “The Cloud”.

The other developer insisted it was just a DLL, or other library, and that processing was done locally on the machine running the bot and studio (we’re still building things).

Where are the images processed during development (ie on a machine running studio)? Where are they processed during runtime (ie after shipping the bot to Orchestrator)?

2

Hi there @Robot.Builder.9001,
From my understanding, it ultimately depends on which OCR engine you are leveraging.

In essence, you are both correct.
Using the Abbyy OCR, Microsoft OCR, or tesseract OCR engines, the images will be processed locally.
With that said, the Abbyy Cloud OCR, Google Cloud Vision OCR, Microsoft Azure Computer Vision OCR, and Microsoft Project Oxford Online OCR engines will process the image within the cloud.

Thanks in advance,
Josh

1 Like
3

Hi @Robot.Builder.9001

This depends on the engine used, some run locally and others through an API in a datacenter.

If there is an API key asked by the engine then it is processed in a datacenter like the Microsoft Azure OCR Engine. Otherwise it is processed locally. This is the same for development and during runtime, even when run from the Orchestrator.

You can easily test which one works in a datacenter of locally by disabling the internet.

Hope this helps,

Astin

4

Locally on which machine? The Orchestrator server, or the machine running the bot?

5

The entire robot is run on the machine and not on the Orchestrator, the Orchestrator sends the robot package to the machine and starts it. But this still means that if the robot is build with an engine that does the processing in a datacenter, that part will still be done there.

1 Like
6

@Robot.Builder.9001

If this answers the questions, could you mark my last message as Solution? This will help other people in finding the answer if they have the same question.
Thank you!

7

I don’t mind marking an answer as such, but there are 3 responses that, only when taken together, form the answer to my question. It’s like Voltron.

See what I did there? Robots…Voltron… hahahahaha

1 Like
8

@Robot.Builder.9001

The entire solution:

This depends on the engine used, some run locally and others through an API in a datacenter.

If there is an API key asked by the engine then it is processed in a datacenter like the Microsoft Azure OCR Engine. Otherwise it is processed locally. This is the same for development and during runtime, even when run from the Orchestrator.

You can easily test which one works in a datacenter of locally by disabling the internet.

The entire Robot is run on the machine and not on the Orchestrator, the Orchestrator sends the Robot package to the machine and starts it. But this still means that if the Robot is build with an engine that does the processing in a datacenter, that part will still be done there.

9

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.