What is OCR and how does it read text from images?
OCR stands for Optical Character Recognition. It is a technology that scans the pixels in an image, recognises the shapes of letters, and converts them into real, editable text.
When you take a photo of a page, your phone saves it as a picture. The computer only sees colours and shapes. OCR uses an engine (like Tesseract) to figure out which shapes are the letter "A" and which are "B", turning that picture into a paragraph you can copy, paste, and search.
How to extract text from an image with MentorMakers
The MentorMakers Image to Text tool converts photos and screenshots into copyable text right in your browser. It supports JPG, PNG, WebP, BMP, TIFF, and GIF files up to 15 MB.
- Open the Image to Text tool on your phone or computer.
- Upload your image. You can also paste a screenshot directly by pressing Ctrl+V (or Cmd+V on Mac).
- Select the language from the dropdown (English or Urdu).
- Click Convert or Extract.
- Copy the text from the output box.
The tool uses ImageMagick to straighten and sharpen the image, and Tesseract to read the text. For multi-page scanned PDFs, use the OCR PDF tool instead.
Worked example. A student takes a photo of a textbook page. The image has some glare and is slightly tilted. They crop the image tightly around the text using the Crop Image tool and then upload it to the Image to Text tool. The tool reads the text and gives them a clean paragraph to paste into their notes.
Does the tool support Urdu and English?
Yes, the MentorMakers Image to Text tool currently reads both English and Urdu.
However, Urdu OCR is technically more difficult than English because the letters are connected. To get the best result in Urdu, the image must be very clear, straight, and high-resolution. If the photo is blurry or has shadows, the Urdu text will likely come out garbled. [CHECK: test with a clear Urdu image to confirm accuracy]
How do I get the best OCR accuracy?
To get the best results, you need to give the OCR engine a clean image to work with. Follow these rules:
- Use screenshots when possible. Screenshots are perfectly sharp and have no shadows, so they give the best accuracy.
- Take photos from straight above. Angled photos confuse the text-recognition engine.
- Use good lighting. Avoid shadows from your hand or phone.
- Crop tightly. Cut away the background, table edges, or anything that is not text. Use the Crop Image tool before uploading.
- Do not use handwriting. OCR is built for printed or typed text. Handwriting recognition is very poor.
Why is my text garbled or out of order?
OCR reads text linearly, from top to bottom. If your image has two columns of text, or a complex table, the OCR engine might mix the lines together. It does not understand columns. You will need to manually rearrange the text after copying it.
Also, OCR often confuses similar characters, such as 0 (zero) and O (letter O), or 1 (one) and l (lowercase L). Always double-check numbers and names before using the text.
Key takeaways
- OCR turns images of text into real, copyable text.
- The tool supports JPG, PNG, WebP, and screenshots up to 15 MB.
- It supports both English and Urdu, but Urdu requires a very clear image.
- Screenshots give the best results. Handwriting does not work well.
- Always check the output for mixed-up columns and confused characters.
Ready to copy text from your image? Extract text from photos and screenshots with the free MentorMakers Image to Text tool in seconds.