- 01Printed labels
- 02Stable camera position
- 03Local journal
- 04Manual evidence
Set up a project you can demonstrate
Use a desk stand, steady lighting and your own project boxes. Start with two different marker labels. Choose a QR size and distance that your actual camera reads reliably; there is no universal distance promised by the code.
If you developed on a laptop and now want a self-contained Pi station, the integrated kit below is an option. Existing compatible hardware remains a valid route. Read the current bundle contents and camera instructions before buying.
Collect useful evidence
Record these observations in a small README:
- Your board/computer, OS, camera capture method and OpenCV version.
- A marker accepted once while held in view, and accepted again after the cooldown.
- A different marker accepted during that first label's cooldown.
- A web-address QR ignored without opening it.
- Capture failure reported when the camera is unavailable.
This is your device-validation work. The downloadable desktop tests do not replace it. Keep a tiny sample of the JSONL output beside your notes, using labels you created for the experiment.
Try it
Move the camera farther away until decoding becomes unreliable. Log that limit and improve lighting or marker size before changing the event policy. Recognition failures and cooldown decisions belong to different layers.