You’re offline. This is a read only version of the page.
Announcements
60
I have a Microsoft Flow which is triggered by a user submitting a Microsoft Form. The form contains questions as follows: name and also submitting a screenshot. The screenshot would contain import data which I was hoping to extract. I tried doing it with the following: Parsing the screenshot using OCR and then applying string manipulation to extract relevant information. However, thanks to the help of the community, I was told to build an AI model and use confidence scores to disregard incorrect data.
Since the screenshots are device captures (from Samsung, Google, iPhone, etc - examples shown below), they all look somewhat different (i.e. placement, amount of information, etc). To configure this, I created collections in my AI model, with each collection corresponding to a different device.
Now I would like to develop a Flow which would take in the screenshot and apply my custom AI model to retrieve the relevant information. Currently my AI Model looks like:
Now I realize that submitting a response with an image requires parsing JSON. Therefore, my current flow is as follows, I'm just having difficulty with implementing it in order to send an email containing those two fields (PCI and EarFCN).
My current flow looks like:
However, I don't see an option to add the values from my AI Model. Is there something I'm doing wrong? Additionally, how would I use confidence intervals to ensure that my model isn't reading incorrect values?
--------------------------------------------------------------------------------------------------------------------------------
Device Captures: These are just two of the possible screenshots that could happen. Initially I was doing an OCR and text parsing (just because of the endless possibilities of the text positions and various devices). If you guys think that a better approach is better, let me know!
Share
Report
All responses (
Answers (
