I need to recognize the text from the screen. For example, I make a screen application in which something happens. Then I drop the image in any folder, and Java takes the image and recognizes the text. There is an assumption that neural networks are involved. If so, then who can know the library, and how to work with it (sample code)

    1 answer 1

    For text recognition, you can use tess4j (tesseract wrapper). It recognizes text from images well, including from scanned copies, but in most cases preliminary processing of documents is required (only horizontal labels are recognized, so you need, for example, to rotate the image, etc.)

    ITesseract ocrEngine = new Tesseract(); ocrEngine.setDatapath("pathToDictionaries"); ocrEngine.setLanguage("usingLangNames"); BufferedImage bImage = <read image code>; String text = ocrEngine.doOCR(bImage);

    Where, pathToDictionaries are trained text recognition models for a specific language (can be taken from a github ), usingLangNames is a list of languages ​​for which models are connected and which need to be used in recognition.