When an image is received, Tesseract OCR extracts text from the image. The system then searches for keywords (like "comprobante", "transferencia", "pago") and bank names in the extracted text. More keyword matches = higher confidence. You can add keywords, bank patterns, and tune scoring below. Review past results to find misclassifications and improve detection.
These keywords are searched in the OCR-extracted text and message captions. More matches = higher confidence the image is a payment receipt.
0 keywords configured