can you please tell how to create object_map.csv for images if want use Pytesseract only instead of Commercial OCR techniques.