Tested with a random manga page:
https://files.catbox.moe/e2f3zm.jpg
swiftocr "recognizes" 2 occurrences of random English garbage/punctuation; I can open this in Preview and select almost all the Japanese text in the page, even the handwritten characters.