Handwriting and Text Recognition (Mobile)
The Memories mobile editor can turn handwriting and photos into text. All recognition runs on the device; no image, stroke or recognized text is sent to any server.
Handwriting
Tap the pen icon in the editor toolbar to open the canvas. Tools: pen, eraser (removes touched strokes), undo and clear. Choose Chinese or English, then:
- Recognize text uses Google ML Kit Digital Ink Recognition
(
google_mlkit_digital_ink_recognition). The language model is downloaded once, only after you confirm; recognition itself is on-device. The result is shown in an editable preview before it is inserted at the cursor. - Save as image (memory editor only) keeps the drawing as a PNG. When the
memory is saved, each image is stored as a separate private, encrypted file
asset titled "
(handwriting)". Buffers are wiped after saving.
Scan text (OCR)
Tap the scan icon and pick the camera, the photo library or an image file. Text
is read with google_mlkit_text_recognition using the bundled Chinese and
Latin script models (on-device ML Kit on iOS as well); the result with more
characters wins. An editable preview is shown before inserting. Temporary
image copies made by the camera/photo pickers are deleted right after
recognition; files chosen from a desktop file system are never deleted.
Protected cards
The hidden secret field of the protected-card form has the same handwriting and scan buttons. Only text is produced (no image is kept), the preview stays inside the app, and the value goes straight into the hidden field.
Permissions
- Android:
CAMERA(requested when you choose the camera). The system photo picker needs no storage permission. - iOS:
NSCameraUsageDescriptionandNSPhotoLibraryUsageDescription.
If access is refused, a message explains how to allow it in Settings.