The pipeline uses Tesseract OCR with custom preprocessing to handle challenging conditions like perspective distortion, low contrast, and scattered digits.