Bank Statement OCR to JSON

I have a batch of scanned bank statements that must be run through OCR and parsed so I receive clean, structured JSON containing only the account-holder information (name, address, account number and any other clearly labeled personal identifiers that appear on the statements).

Accuracy and consistency are critical. I will verify the JSON against the original PDFs, so the output must match every field exactly as it appears, without spelling mistakes or truncation. Feel free to employ any combination of Tesseract, ABBYY FineReader, AWS Textract, or another OCR engine that can handle mixed fonts and occasional background noise—use whatever delivers the best character recognition rate.

Deliverables
• A single JSON file per statement, named to match the original document
• A brief summary of the processing workflow (tool versions, key settings, and any post-processing scripts) so I can reproduce the results if needed

Acceptance criteria
• 100 % of account-holder fields captured for each statement
• No extra data (transactions, balances, etc.) present in the JSON
• All files delivered via the shared drive I will provide, with folder structure preserved

Turnaround is flexible over the next few days, but please indicate how many statements you can process per hour so I can schedule the hand-off. Data is confidential, so NDA compliance is assumed.