Scan a receipt
1
Open receipt capture
Start from the add flow and choose the camera or photo library.
2
Run on-device OCR
Google ML Kit extracts text on the device. The image is not sent to a Salli
server.
3
Reconstruct the receipt
Salli uses spatial text layout and deterministic rules to infer merchant,
total, date, and likely category.
4
Review the draft
Correct every field before saving. OCR is an assistant, not proof of a
transaction.
Image retention
The receipt-scan image used for OCR is not retained automatically. If you explicitly add a receipt or file as an attachment, Salli copies it into app-owned local storage and links it to the transaction.Backup behavior
JSON backup contains attachment references, not the image binaries. Restoring to another phone therefore restores the ledger metadata but cannot recreate the original files.Delete behavior
Deleting an attachment removes only a path-confined file inside Salli-owned storage. Delete-all removes the complete Salli receipt directory, including orphan files left by an interrupted edit.Platform notes
- Android receipt OCR builds and runs on supported physical devices.
- iOS receipt OCR targets physical devices on iOS 15.5 or later.
- The current native ML Kit binary does not link on Apple-silicon iOS simulators.
- Web does not provide the production OCR path.
