I’m using pytesseract with 99% success. The 1% failed sentences are often caused by a small drawing (B&W) in the left margin. When I invert the image (everything=white on black BG), the problem seems to go away. Is this an explainable feature, or just a coincidence?
and, how is this related to opencv ?
(i’d vote to close this for being off-topic, you’re asking in the wrong place, imo)