Industrial OCR Solutions — Read Text, Not Just Barcodes

A barcode scanner beeps all day and reads exactly one kind of thing: a barcode.

Ask it for the etched serial on a machined part, the lot number on a label, the expiry date, or a document reference, and it has nothing to say. Those are printed characters, not encoded data — there is no symbology to decode. On the line, that means typing the value by hand or running a second, separate OCR workflow.

HappyLab adds OCR to the devices you already have. Three form factors — handheld scanner, fixed-mount reader and Android PDA — all read printed text, extract the fields you want with rules, and put the result straight into the system that needs it.

All three run fully offline, on the hardware you already own. No cloud, no image ever leaving your plant.

The problem with the usual answers

Two approaches dominate, and both hit a wall on a real line.

OCR built into the device hardware is easy to deploy but frozen at the moment of manufacture. The model is fixed: it cannot follow a new character set, a new layout or an extra inspection point, and it cannot be tuned for a new industry.

Vision or multimodal large models are capable, but rarely practical on site. They need heavy compute, depend on the network — a data-compliance risk when images leave the building — add latency that misses takt time, and are complex and expensive to deploy.

Device-side OCR is too rigid. Large-model OCR is too heavy. Neither lands cleanly on a production line.

HappyLab takes the third path: a light, on-device model running on the PC or PDA you already have.

Three form factors, one engine

All three share the same four upper layers — pre-processing, engine, rules and output. Only the bottom capture layer changes with the hardware, so an upgrade to the model or the rules reaches every form factor at once.

How it works

1
2
3
4
5

Capture (USB · FTP · camera / scan engine)
   → 15-filter pre-processing (white balance, contrast, enhancement)
   → OCR engine (tiny 6 MB or small 30.5 MB)
   → Regex rules (extract only the fields you want)
   → Output (focus injection · broadcast · records)

Step 1 — capture the frame. On a scanner, squeezing the trigger returns the image immediately. On a fixed-mount reader, a photoelectric sensor triggers the IO and the image is pushed over FTP with no operator involved. On Android, a scan key, a broadcast or an on-screen button does it.

Step 2 — clean the image. A 15-filter chain handles dark, low-contrast, colour-cast and shadowed captures before recognition, so the engine is not fighting the lighting.

Step 3 — recognise. Detect plus recognise, with in-house pre- and post-processing. The default tiny model is 6 MB and processes a single region in roughly 37 ms; the small model (30.5 MB) is there for complex backgrounds. Switch between them at runtime — no reinstall.

Step 4 — extract only what you want. Configure named rules so the result is a clean field, not a wall of text:

1
2
3
4

Rule:     serial
Regex:    [A-Z0-9]{4}-[A-Z0-9]{4}-[A-Z0-9]{4}
Text:     "Ticket 5465-2270-8599 — winner!"
Result:   "5465-2270-8599"

Step 5 — put it where it belongs. Focus injection types the value into whatever field has focus — Excel, ERP, WMS, your in-house app — exactly like a barcode scan.

Capabilities

Where it fits

Requirements and limits

• Platform: Windows 10/11 64-bit for the scanner and fixed-mount products; Android 9+ for the PDA product. No GPU required.
• Capture: a supported handheld scanner or fixed-mount reader over USB, or an Android PDA camera or scan engine. Other models are supported through custom development.
• Interface language: Chinese today; the English build is in progress.
• Fully offline: no backend, no cloud, nothing leaving the machine or the device.
• Accuracy depends on image quality — flat, evenly lit and in focus gives the best result.

Start with the free online tool

You can try the same engine in your browser before talking to anyone: run OCR on an image and see the result. No install and no sign-up.

Then, when you are ready:

Every license is one-time — no subscriptions. A free 30-day trial is available on request.

Read the details

• Scanner-OCR V2: Add OCR to Your Barcode Scanner
• Fixed-Mount OCR V2: Read Characters and Grade Barcode Quality on the Line
• Android-OCR: On-Device OCR for Android PDAs
• Web OCR: Test On-Device OCR in Your Browser, Free

💬 Questions or a custom requirement?

Tell us what you need to read and where it is printed, and we will set up a fast, tailored version for you — including a free trial on your own parts.

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