Hello people.
So, I have a form in a PDF format. There is a kind of situation and according to that there are two check boxes. The AI Builder is able to correctly get information from the check boxes. So, below the check boxes, there is a signature part. With document processing, I am able to get the text value from the signature, sometimes correctly and sometimes incorrect and sometimes it is not able to detect the signature properly. So, what was I thinking, was to extract the image of the signature and store it in the database. So, how can I do that? [Note: The form would be handwritten(Signatures, ticks in check boxes)]
First way, I was thinking was to use Power Automate Desktop and just extract the image of that certain part with the help of snapping tool and then store it in the database.
Second way, I have the coordinates of the signature area, which I got from document processing. So, is there anyway I can extract with the help of those coordinates?
Third way: I want you guys to suggest me any other way to do it. To perfectly extract the text or just the image of the signature. I also tried using Azure's Text Recognizer OCR, but its similar to the AI Builders Text Recognizer.
Below I am giving how my form looks and the coordinates that I am getting from the document processing.
Image of Form:
Coordinates I am getting:
Thanks for the name. Hehe.
Actually we preferred Form Recognizer. It was decided by the seniors as well as the client.
So, I tested the signature part in Form Recognizer. And in most cases it was able to extract the data out of signatures. But at the end we decided that we will just take the value, that it is signed or not.
Thanks for the help. 🙂
Hello @Unknown123 - you have a cool username. 🙂
Indeed, signature detection is not yet supported in AI Builder Document Processing. It is in our roadmap for a future update.
A possible option could be to look into Azure Form Recognizer custom template model that supports signature detection: https://docs.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/whats-new?tabs=csharp#form-recognizer-model-data-extraction
Feel free to share which option you end up using. Thanks!