How to extract text from images with OCR
In short: Capture a single image, check the recognised text, restore useful structure, and save a note you can find again.
To extract text from an image, start with a readable photo or screenshot, run OCR, and review the result before saving it. In MemoFlow, you can photograph a page or select one image from your photo library. Standard recognition runs on the device and advanced recognition runs in the cloud; either way, starting the operation requires an online Credits authorization, and standard OCR is free on Plus and Pro. The result opens in a draft editor where you can check the original image, correct names and numbers, choose a title and folder, and save the text as a note. OCR does not promise a correct transcript, table layout or handwriting result. Keep your own source copy if you will need it after review: the temporary OCR image is removed when the review session ends.
Prepare a readable source image
Hold the phone parallel to the page, use even lighting and check that the words are legible in the photo. For a curved book page, flatten it where permitted or capture a smaller area. Inspect blur, glare and cut-off lines before starting. These are capture checks, not a guarantee of recognition quality.
Capture one image in MemoFlow
Open OCR, choose a mode, then take a photo or select one image from the photo library. Standard recognition runs on the device after an online Credits authorization and does not upload the image (1 Credit per image on Basic, free on Plus and Pro). Advanced recognition is for complex layouts such as multi-column pages and tables: it uploads the image to the cloud for recognition, which you confirm before first use, and it costs 2 Credits per image on every plan. You do not need to select an OCR language. The workflow is for a single image, not a batch or PDF converter.
Review words and structure separately
Use the original image during review to check names, amounts, dates and quoted passages. Then restore headings, paragraph breaks and reading order. Inspect multi-column text and table cells individually rather than assuming the original layout survived. Handwriting and decorative type need source comparison; no measured quality guarantee is published.
- Check meaningful words before polishing punctuation
- Compare each number you plan to use with the image
- Restore headings and paragraph order
- Remove page numbers and unrelated marks
Save the recognised text into a note
Choose a title that describes the source and the reason you captured it, then select a folder and save. Note search can retrieve the saved text; it does not search pixels in an unsaved photograph. If you use AI later, select the relevant note and check the answer against the source.
Retain a source copy when needed
The OCR review image is temporary and is removed when review ends; saving the recognised text does not automatically attach the photo to the note. Keep a photo-library copy or explicitly add an attachment if you need a durable record of the layout, diagram or original quotation. Check which material is allowed to be retained or shared before doing so.
Key points
- Standard recognition is on device and advanced recognition is in the cloud; both need an online Credits authorization.
- Capture one image and review the result before saving.
- Saved note text is searchable; an unsaved image is not a note-search corpus.
- Retain your own source copy if it will be needed after OCR review.