The scanner that can’t phone home.
Your pages never leave your phone. No internet permission — check for yourself.
AndroidManifest.xml — merged, as shipped
<uses-permission android:name="android.permission.CAMERA" />
<uses-permission android:name="android.permission.READ_MEDIA_IMAGES" />
<uses-permission android:name="android.permission.INTERNET" /> // stripped, build-enforcedA Gradle task fails the build outright if INTERNET reappears in the merged manifest. It isn’t a policy promise — it’s a check that runs on every build.
Three steps, zero uploads
Illustrative mockup of boundary detection — not an actual in-app screenshot. Real screenshots are coming with the Play Store listing.
Point
Aim your camera at a page or a book. Live boundary detection finds the edges and auto-captures the moment they hold steady.
Shoot
Curved book pages get straightened by a dewarp model running on your phone. No upload, no processing queue.
Export
On-device OCR makes every page searchable, then it’s a PDF sitting in your files — not in someone’s cloud.
What it does
Book dewarp
Classical computer vision plus a UVDoc neural model flattens curved book pages, running entirely on your phone.
On-device OCR
ML Kit text recognition runs locally to build a searchable text layer. Your document’s contents never transit a network.
Searchable PDF
Export lands as a real PDF with a positioned text layer, ready to search, share, or file away.
No ads
No ad SDKs, no ad network calls, nothing to serve them over anyway — there’s no internet permission to carry the traffic.
How it stacks up
| Talpatra Scan | Other scanners | |
|---|---|---|
| Network access | None — no INTERNET permission, build-enforced | Required for cloud sync, accounts, or ads |
| Ads | None | Common in free tiers |
| Book dewarp | Yes — fully offline | Varies; when present, usually cloud-processed |
“We don’t know what you scan, and we designed it so we can’t.”
Built by one developer. No team, no analytics, no idea what’s in your documents — on purpose.