A guide to turning a picture of text into text you can copy, edit and search.
OCR stands for optical character recognition. A computer looks at a picture, finds the shapes that look like letters, and writes out the text it thinks they are. That turns a photo of a page, a screenshot or a scanned PDF into text you can copy, edit and search, instead of typing it out by hand.
It works best on printed text. It is not designed for handwriting, which varies too much from person to person for a reliable result.
Open the Image to Text tool, choose the language in the picture, and add your photo, screenshot or scanned PDF. The tool reads the page and shows the text in a box. Read it through, correct anything that is wrong, and then copy it or save it as a text file.
For a PDF, each page is turned into an image and read in turn. A PDF that already has selectable text does not need OCR, because you can copy the text from it directly.
The reader needs to know which letters to look for. Choose English for English text and Hindi for Hindi. If the page mixes the two, as government notices and exam papers often do, choose English plus Hindi, which reads the page in both languages and keeps the better result for each line. A single language is faster and a little more accurate when the page only uses one.
The quality of the picture matters more than anything else. Take the photo straight on, with the page flat and evenly lit, so the text is sharp and dark against a plain background. Fill the frame with the text and crop away everything else. Avoid shadows, glare from a flash, and photos taken at an angle.
A screenshot is the easiest input of all, because the letters are already clean and straight. If a scanned PDF gives poor results, go back to the original scan if you can, since a compressed copy has already lost detail.
Hindi is harder for a computer than English because letters join into conjuncts and carry marks above and below the line. Clear printed Hindi, such as a notice, a book page or a question paper, reads well. Small, faint or angled Hindi will have more mistakes, so check names, numbers and dates against the original.
No text recognition is perfect. Read the output once before you use it, especially figures, names, dates and anything you will submit. Look for confusions like zero and the letter O, or one and the letter l, which are easy to miss.
Many online OCR sites upload your picture to a server, where it is read and sometimes kept. Documents like ID cards, marksheets and bank letters are not something you want to hand to an unknown service. Tooqly runs the recognition engine inside your browser, using files served from tooqly.com, so the image or PDF you choose never leaves your device.
Not reliably. It is built for printed text. Neat block capitals sometimes work, but most handwriting will come out wrong.
Usually the picture. Try a sharper, straighter, better-lit photo, crop to the text, and make sure the language setting matches the page.
Yes, for clear printed Hindi. Choose Hindi, or English plus Hindi for a mixed page, and check the result against the original.