welcomecenterbot/nlp/ocr.py

18 lines
565 B
Python
Raw Normal View History

2024-09-28 07:06:04 +00:00
import easyocr
import logging
2024-09-28 08:17:20 +00:00
logger = logging.getLogger(" ocr ")
2024-09-28 07:06:04 +00:00
# Initialize the EasyOCR reader
reader = easyocr.Reader(['ru']) # Specify the languages you want to support
def ocr_recognize(file_path):
# Use EasyOCR to detect text in the photo
result = reader.readtext(file_path)
2024-09-28 08:23:10 +00:00
logger.debug("OCR Result: %s", result)
2024-09-28 08:17:20 +00:00
2024-09-28 08:23:10 +00:00
# Extract recognized text
recognized_text = ' '.join([text[1] for text in result if isinstance(text, tuple) and len(text) > 1])
2024-09-28 08:17:20 +00:00
2024-09-28 08:23:10 +00:00
logger.debug(f'Recognized Text: {recognized_text}')
2024-09-28 07:06:04 +00:00
return recognized_text