AI / IoT & devices / Web
Licence-Plate Recognition
Camera webhooks, an OCR worker and a live dashboard — plus a custom Lao plate model.
- My role
- Developer: API, worker, dashboard and model training.
- Platforms
- Web, API
- Year
- 2026
- Engagement
- Full-time role
Built for a client. Names, logos and identifiers are blurred; all data shown is demo data.
Architecture
Screens for this project are under a strict NDA. The diagram shows how the system fits together.
Overview
A production ANPR service: traffic cameras push events, an API stores them and a Python worker reads the plate, then a live dashboard shows the feed with filters, KPIs, OCR cost and retention rules. Alongside it, a two-stage YOLOv8 model for Lao plates — a detector at 0.998 mAP50 and a character and province reader still improving — trained in a labelling and retraining loop on a local GPU box.
What I built
- Camera webhook ingest and worker queue
- Live feed, KPIs and cost dashboard
- Vehicles, companies, gates and transactions (v2)
- Two-stage Lao plate model with a labelling tool
→Contact
Let's build something that stays up.
I'm looking for a team where I can own meaningful systems. Hiring, or have a project in mind? I usually reply within a day.
WhatsApp
+856 20 7817 2771Phone
+856 20 7817 2771Location
Vientiane, Laos · UTC+7