[INFO] 14:20:01 Pipeline initialization successful [INFO] 14:20:01 Document format detected: image/pdf [INFO] 14:20:02 Starting Tesseract Engine v5.3.0 [INFO] 14:20:03 OCR Scan: 124 words detected with 0.94 avg confidence [DEBUG] 14:20:03 Normalizing geometry for extraction [DEBUG] 14:20:04 Calling Classification Endpoint: /api/v1/classify [WAIT] 14:20:04 Inference in progress... (Current latency: 450ms) [RUN] 14:20:05 Document Type identified as 'INVOICE' (0.99) [RUN] 14:20:05 Switching to extraction model: 'layoutlm-v3-base' [INFO] 14:20:05 Beginning key-value pair extraction...