How to extract image “features” (not the text) from Tesseract / PyTesseract?

Question

I can extract the text from an image from tesseract using the Python binding PyTesseract as follows:

import cv2 
import