Android Solutions — On-Device OCR and Field Debugging

Two jobs on an Android handheld get in the way of everything else: getting printed data into your business app, and getting a device debugged on a customer site.

HappyLab solves both, on the device and in the browser.

• Android OCR turns a PDA into an OCR terminal — press the scan key, and the printed text lands in the active field.
• AI-ADB runs a full ADB console in a browser tab — no drivers, no Python, no software on the customer’s machine.

Neither needs a backend, and neither sends your data anywhere.

Android OCR: read printed text on the device

A barcode scanner beeps all day and reads nothing but barcodes. Ask it for the printed serial number on a label, a batch code, an expiry date or a document reference, and it has nothing to say — those are printed characters, not encoded data. Until now, that meant typing the value by hand or running a separate OCR workflow.

Android OCR runs on the PDA as a background service. It reads printed text through the camera or the built-in scan engine and types the result straight into whatever field has focus. Press the scan key and the value appears — the same gesture your team already uses for barcodes.

Everything is computed on the device. No network, no server, no image ever leaving the PDA.

1
2
3
4
5
6
7

Business app (cursor in a field)
   │  press the scan key / send a broadcast / tap the floating ball
   ▼
Resident OCR service
   │  capture → crop → on-device OCR → rule extraction
   ▼
The result is typed into the business app's field — as natural as a barcode scan

Two capture paths

Three ways to trigger it

The broadcast channel is what makes integration cheap. One line in your app is enough:

1
2

Intent intent = new Intent("ocr.scan");
sendBroadcast(intent);   // the PDA opens the capture window in the current mode

Getting the result into your system

Recognition engine

Detect plus recognise, with in-house pre- and post-processing. A single region takes roughly 37 ms, only the framed area is processed, and you can switch tiers at runtime without reinstalling.

A 15-filter pre-processing chain and a regex rule library then turn the recognised text into the exact field you wanted:

1
2
3
4

Rule:     Serial number
Regex:    [A-Z0-9]{12,18}
Text:     "SN 4SC4-316-abc-123 batch B2026"
Result:   "4SC4-316-abc-123"

Supported devices

Tested Android PDA families include EDA5S · EDA10A · EDA51 · EDA52 · EDA56 · EDA61K · CT32 · CT40 · CT45 · CT47 · CK62 · CK65 · CK67 · RT10A.

Using a model that is not listed? We support additional devices through custom development — send us the model.

Built for the next hardware generation

Android OCR runs a standard ONNX model on a standard Android inference stack — a deliberate choice:

• ✅ Runs on today’s hardware — no special silicon required; mainstream PDAs are enough.
• 🔮 Gets faster on tomorrow’s — the next generation of PDAs ships dedicated NPUs, and because the engine uses standard execution providers, those devices accelerate it with no code changes.
• 📈 The model can grow — upgrade or swap the model without touching the hardware you already own.

Recognition capability should not be frozen at the moment of purchase.

AI-ADB: a full ADB console in the browser

Debugging an Android device should not require a toolchain. Installing ADB, matching drivers, fixing environment variables, hunting for a data cable, keeping a second phone ready to film the screen — the setup often takes longer than the fix. On a customer site, with a locked-down laptop and no admin rights, it can take all afternoon.

AI-ADB is a complete Android device console that runs entirely in a web browser. Plug in a USB cable, open a tab, and start working. No ADB install, no Python, no drivers, no software to deploy on the customer’s machine.

All communication happens over the USB cable itself. Nothing is sent to a server, so there is zero latency and zero data exposure.

Mirroring uses the scrcpy protocol — H.264 hardware encoding on the device, decoded in the browser through WebCodecs and rendered to a canvas. The result is low-latency mirroring with click, swipe and drag mapped precisely to the device, plus a full navigation bar, screen wake and sleep, rotation, live screenshots and screen recording to a downloadable video. No plugins, no separate mirroring app.

Where it fits

Pricing

All licenses are one-time — no subscriptions, nothing to renew. Custom development starts at $300 for a specific output format, a non-listed device or an extra language.

Requirements and limits

• Android OCR: Android 9+ (minSdk 29). Scan-engine mode needs a compatible scan engine; camera mode needs a camera. The license is bound to the device serial number. Key injection is system-restricted on Android, so accessibility-based injection is used today — native keyboard injection is in progress.
• AI-ADB: a Chromium-based browser only (Chrome or Edge); Firefox and Safari do not expose WebUSB. USB debugging must be enabled, and some vendor ROMs restrict ADB by default. USB only — this is not a wireless ADB tool.
• Interface language: Chinese for both today; the English build is in progress.
• Fully offline: no backend, no database, no data leaving the device.

Try it first

Open AI-ADB in your browser with a device on the end of a USB cable and see it for yourself. For OCR on the desktop, try the online OCR tool.

Related reading: Android-OCR: On-Device OCR for Android PDAs and AI-ADB: Control Any Android Device From Your Browser.

💬 Questions or a custom requirement?

Tell us your device model and what you need it to do, and we will tell you whether it works out of the box or needs a custom build.

📧 sales@happylab.me · 🌐 global.happylab.me