License plate recognition (ANPR / LPR) →
In-house engine at 99.6%, runs on any IP camera.
Made in Azerbaijan 🇦🇿 · Computer Vision · AI
Smart parking powered by plate recognition — from Baku
PAG is a software platform that simplifies parking management: automatic plate recognition at 99.6%, contactless payment and real-time data. More than 40 sites in Azerbaijan: airport, malls, hotels, hospitals, business centres and residential complexes.
Solutions
Smart barrier gate →
Opens by plate, guests by QR, no remotes or cards.
Smart parking →
Unattended paid parking: QR, kiosk, app, dynamic tariffs.
PAG vs Parqour →
An honest comparison for choosing a parking platform in Azerbaijan.
IT & AI development →
Computer vision, neural networks, mobile apps, video analytics, automation.
Where it already works
40+ · Azerbaijan
- Port Baku Residence2023
- Baku Convention Center2023
- Bravo Hypermarkets2023
- Azərbaycan Dəmir Yolları (ADY)2024
- Park Bulvar Mall2024
- Amburan Mall2024
- Sea Breeze Resort2024
- Liv BonaDea Hospital2024
- Olimpik Star Fitness2024
- East Park Residence2024
- Baku City Residence / ASAN 52024
- Fairmont / Flame Towers2024
- Bravo Bayıl / Yasamal / Günəşli2025
- JMR Plaza2025
- White City Fountain Square2025
- Dünya Göz Hospital2025
- Bilgəh Cardiology Sanatorium2025
- Hyatt Regency Baku2025
- Blue Mall & Residence2025
- White City Office Building2025
- Chinar Park Business Center2025
- Port Baku 22025
- Bakıxanov Mall2025
- Makro Park2025
…and more sites across the country. The list grows every month.
Our technologies
Built in-house by PAG
All of PAG’s core technologies are developed in-house by our team in Baku — from license plate recognition and Face ID to AI under-vehicle inspection, the AIRA assistant, robotics, AI territory protection and our own AI.
PAG Recognizer — license plate recognition →
In-house ANPR / LPR engine with 99.6% accuracy on live sites. Runs on ordinary IP cameras — at night, in rain and at an angle — locally at the barrier.
PAG Gabarit — vehicle type, size and colour
Detects vehicle type (sedan, crossover/SUV, van, minibus, bus, truck), size class and colour from the ANPR camera frame. Works on PAG sites together with plate recognition — for class-based tariffs and access rules.
PAG Face ID — face login
Face login for operators on all PAG sites. Anti-spoofing: random head-movement challenge, colour-flash liveness check and AI verification against photo or phone-video replay.
PAG UVIS — AI under-vehicle inspection
Under-vehicle inspection for security checkpoints: an under-road camera module with lighting scans the underbody as the car drives over; AI looks for foreign objects and explosives and compares the scan with earlier passes of the same plate. The barrier opens only after a clean result.
PAG AI — our own AI
PAG’s own AI running on our servers: web panel, terminal client and OpenAI-compatible API. Company data never leaves our infrastructure.
AIRA — PAG’s AI assistant
Voice and text AI assistant: calls right on the website and in WhatsApp, a Telegram secretary for management, and an operator assistant in the parking workstation (payments, debts, manual openings). Speaks Azerbaijani and Russian.
PAG Robotics — robotics
Development of AI-driven robots: voice conversation in the speaker’s language, cameras and recognition, control by PAG’s own AI.
Velum Vision — AI territory protection →
Perimeter and territory protection with AI video analytics: Face ID, ANPR radar, people and vehicle tracking, real-time intrusion alerts.
FAQ
Frequently asked questions
PAG (Park AI Global) is an IT and AI company founded in Baku, Azerbaijan. It builds its own license plate recognition engine, a parking management platform and Velum Vision smart video surveillance.
On more than 40 sites in Azerbaijan — Park Bulvar, Hyatt Regency, Fairmont, Port Baku, Baku Convention Center, Bravo, Liv BonaDea, ADY and others; pilots in Georgia and the Czech Republic.
Write to [email protected] — we return a financial model within 1 day.
Ready to launch smart parking?
Write to us — we will model the economics of your site and come back with a financial model within 1 day.