I need to develop software that can automatically validate documents called “W contracts.” These contracts contain numerous checkboxes and fields to fill out, and my goal is to accurately identify which boxes are checked and validate the values based on specific rules and parameters.

I’ve tried using OpenAI’s API to detect these elements, but I’ve found it struggles with handling the large number of checkboxes and options in the documents. It often misidentifies the structure or fails to recognize the checkboxes correctly.

I’m looking for recommendations for technologies, libraries, or APIs that could help me perform this task more accurately, preferably using OCR and computer vision. The ideal solution should be able to:

Detect checkboxes and input fields in a PDF or image document.
Accurately determine whether the checkboxes are checked or not.
Read fields and validate the data according to certain rules.
Any suggestions on tools or approaches that can facilitate this task would be greatly appreciated. I’m open to AI-based solutions but also interested in traditional OCR approaches.