Hey there, fellow automation enthusiasts! I’m currently working on an AI-driven automation process for generating reports, and I’m stuck on a crucial step: reliably capturing and extracting complex tables from PDF documents. The goal is to convert these tables into structured JSON for further analysis.
I’ve tried various tools, including ChatGPT-4, Gemini 2.5, Google Document AI, and several Python libraries like PyMuPDF and pdfplumber. However, these tools often misinterpret table structures, especially when dealing with merged cells, nested headers, or irregular formatting. This leads to incorrect JSON outputs, which affects subsequent analysis.
So, I’m turning to the community for advice: Has anyone found a reliable process, OCR tool, or AI approach to accurately extract complex tables into JSON? Any tips or advice would be greatly appreciated.
I’m eager to hear about your experiences and learn from them. Let’s work together to overcome this challenge!