I use Flameshot combined with Tesseract and zbarimg to quickly clip areas of the screen and either OCR them or decode barcodes, which I then map to hotkey combinations. For example, I have `bash -c 'flameshot gui -s -r | tesseract - - | gxmessage -title "Decoded Data" -fn "Consolas 12" -wrap -geometry 640x480 -file -'` mapped to Super+O, so I can just press the key combo, select a region of the screen, and have the O…
Great idea - I ended up experimenting to improve the ocr accuracy:
#!/bin/bash
screenshot=$(mktemp)
decoded_data=$(mktemp)
processed_data=$(mktemp)
cleanup() {
rm "$screenshot" "$decoded_data" "$processed_data"
}
trap cleanup EXIT
flameshot gui -s -r > "$screenshot"
convert "$screenshot" \
-colorspace Gray \
-scale 1191x2000 \
-unsharp 6.8x2.69+0 \
-resize 500% \
"$screenshot"
tesseract \
--dpi 300 \
--oem 1 "$screenshot" - > "$decoded_data"
grep -v '^\s*$' "$decoded_data" > "$processed_data"
cat "$processed_data" | \
xclip -selection clipboard
yad --text-info --title="Decoded Data" \
--width=940 \
--height=580 \
--wrap \
--fontname="Iosevka 14" \
--editable \
--filename="$processed_data"