OCR - Extract Specific Text

[Boston23]

We've been testing the "recognize text in an image" and it's working well, but the output includes all text from the image when we're really just looking for the serial number. 

Is there any way to set it up to only extract that type of information?

[SimnaJ]

Here are the general steps:

1. Identify the area of the image where the desired text is located.
2. Use OCR to recognize the text in that specific area.
3. Filter the OCR output to only include the desired text.

To accomplish this, you can use a combination of Power Automate and Cognitive Services. Here's an overview of the process:

1. Create a Power Automate flow and add a "Get file content" action to get the image file.
2. Add a "Recognize text in an image" action from the Cognitive Services connector.
3. Configure the OCR action to recognize text in the specific area of the image where the desired text is located. You can do this by specifying the x,y coordinates and width,height of the area in the "Region of interest" field.
4. Add a "Filter array" action to the flow and configure it to filter the OCR output to only include the desired text. You can use an expression like "contains()" or "equals()" to specify the text you want to filter for.
5. Add a final action, such as sending an email or writing the text to a SharePoint list.

With these steps, you can extract specific text from an image using OCR. Keep in mind that the accuracy of OCR can vary depending on the quality of the image and the complexity of the text, so it's a good idea to test the flow with different images and adjust the OCR configuration as needed.

If you found my previous response helpful, please consider marking it as the solution and giving it a thumbs up. This will help other community members to quickly find the solution to similar issues. Thank you!

[Cam_14]

Hi SimnaJ, 

Is there an option to set the coordinates for an online flow on Power Automate? I have done so for a PAD flow before but now I'm building an online flow, and with every run the flow generates strings in different order.

The image concerned is actually the screenshot of a powerbi report consists of multiple tables with irregular number of rows and columns (I basically just put a bunch of textboxes and cards together along with tables of different sizes) that is refreshed automatically and I use the flow to extract 3 dynamic numbers next to a fixed string (item names).

Thanks a lot!