[Python] OpenCV produces "OCRTesseract(33): Tesseract not found" error with working Tesseract...

This Python 3.12.7 script with tesserocr 2.8.0 and opencv-python 4.11.0.86 works as expected:

import cv2 as cv
from tesserocr import PyTessBaseAPI
from PIL import Image

img = cv.imread('text.jpg')[:, :, 2]
_, img = cv.threshold(img, 180, 255, cv.THRESH_BINARY_INV)
pimg = Image.fromarray(img)
api = PyTessBaseAPI(path='/usr/share/tesseract-ocr/4.00/tessdata')
api.SetImage(pimg)
print(api.GetUTF8Text())
print(api.AllWordConfidences())


When I try to use Tesseract directly from OpenCV with this script:

import cv2 as cv

img = cv.imread('text.jpg')[:, :, 2]
_, img = cv.threshold(img, 180, 255, cv.THRESH_BINARY_INV)
ocr = cv.text.OCRTesseract_create(datapath='/usr/share/tesseract-ocr/4.00/')
ocr.run(img, 90)


I'm getting this error:

OCRTesseract(33): Tesseract not found.
/usr/share/tesseract-ocr/4.00/


The documentation requires datapath argument to be the name of the parent directory of tessdata ended with "/", which it is. What is the problem here?

Continue reading...