diff --git a/parse_pol177.py b/parse_pol177.py index cd56182..53b3f1a 100644 --- a/parse_pol177.py +++ b/parse_pol177.py @@ -1,7 +1,7 @@ #!/usr/bin/env python3 """ Parse PDF document using unstructured-api on 192.168.1.103. -Uses maximum quality settings (hi_res + GPU). +Uses maximum quality settings (hi_res + yolox model). Saves parsed JSON to /tmp/pol177_parsed.json. """ @@ -13,12 +13,12 @@ PDF_PATH = "/opt/documents_xk/АОХКСибцем/07 Служба Вице-пр API_URL = "http://192.168.1.103:8005/general/v0/general" def parse_pdf(): - print("Parsing PDF with maximum quality (hi_res + GPU)...") + print("Parsing PDF with maximum quality (hi_res + yolox)...") # Maximum quality parameters form_data = { 'strategy': 'hi_res', - 'hi_res_model_name': 'chipper', + 'hi_res_model_name': 'yolox', 'pdf_infer_table_structure': 'true', 'extract_images': 'true', 'extract_image_block_types': '["image", "table"]',