Using the Tesseract-OCR library, I have built a Bengali text extraction class, OCR. The class performs 3 simple tasks such as filter images, extract address (modify as your need) and extract text from image. Pytesseact can detect multiple languages, available on official documentation.