ocr - select screen portion and recognize text from non text source such as videos

ocr

#!/bin/env bash

input="$(mktemp)"
output="$(mktemp)"

import "$input.png"
tesseract -l eng "$input.png" "$output" 2> /dev/null
cat "$output.txt"

rm -f "$input"
rm -f "$input.png"
rm -f "$output"
rm -f "$output.txt"

Here is a little and unspectacular script to read text from screen. It uses tesseract to ocr the text and import command from ImageMagick to make a screenshot. Then the script outputs the recognized text to stdout. You could replace the screenshot tool with something else you like, but the script expects the created files.