Hi everyone…
What I want is to be able to output all the questions, options, and, if applicable, the question image in a school exam file, which is currently in PDF format, as a JSON file. My next thought is to add the answer options and solution steps to this information and use it in fine-tuning an open-source model. The biggest challenge is fully understanding the image or shape within the question and ensuring the accuracy of the cropping process. The reason for this is that question booklets don’t have a fixed format. While there might be two questions on a page, there might be eight or ten on a page split in the middle. This could be a math question with pictures, or a physics question without any form. What the n8n agent is asked to do here is adapt to any question booklet format, rather than a simple and standard template like an invoice, with a broad range of expectations. What do you recommend? I’m also attaching an example of a simple three-page file.
@ocr
Hey @ressamendy hope all is good. Welcome to the community.
I don’t think the three-page file was actually attached to your question.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.