Recognition accuracy
PAG: in-house neural network, 99.6% on real Baku sites, trained on Azerbaijani, Georgian and Turkish plates. Recognition quality is what drives queues and revenue leakage.
Parqour is a Kazakhstan-born parking automation startup based on plate recognition, operating in several countries including Azerbaijan. PAG is a platform built in Baku with its own recognition engine and the largest smart parking network in the country. Below is an honest comparison on what matters to a site owner.
PAG: in-house neural network, 99.6% on real Baku sites, trained on Azerbaijani, Georgian and Turkish plates. Recognition quality is what drives queues and revenue leakage.
PAG: more than 40 sites — airport, Park Bulvar, Hyatt Regency, Fairmont Flame Towers, Port Baku, the Bravo chain, hospitals, residences. The largest smart parking network in Azerbaijan.
PAG: computer vision, software and hardware engineers are in Baku. Same-day site visits and fixes, 24/7 support in Azerbaijani and Russian.
PAG: hardware, installation and maintenance at our expense, paid back through a revenue share. Live in 15 days.
PAG: recognition and the gate-opening decision happen locally on site; the system does not depend on internet or an external cloud.
PAG: one vendor for parking, smart residential barriers, Velum Vision AI video surveillance, speed control and a driver mobile app.
Number of bays, occupancy, entrance layout — to [email protected].
Within 1 day we return a financial model and rollout plan.
Installation, integration with barriers and payments, staff training.
…and more sites across the country. The list grows every month.
Yes. We connect to existing cameras and barriers, migrate subscriptions and whitelists, run in parallel and switch over without stopping the car park.
A local development and support team, an engine trained on local plates, 40+ live sites nearby you can visit, and a zero-investment model.
Yes: QR, kiosk, PAG mobile app, bank cards, Apple Pay and Google Pay, plus subscriptions and validation for tenants.
Yes. The engine reads plates of 100+ countries, with pilots in Georgia and the Czech Republic and sites in preparation in the UAE.
Write to us — we will model the economics of your site and come back with a financial model within 1 day.