MareArts ANPR

ANPR Detection	Road Object Detection	Mobile App

Automatic Number Plate Recognition — detection + OCR for 80+ countries, GPU acceleration, ARM/Windows support, and cloud Vehicle Intelligence (make, model, color, type, face, plate nation, and server-side OCR).

One license covers everything: Python SDK · REST API Server · Mobile App · Road Object Detection

Get your license at marearts.com/products/anpr

---

What's Included in One License

Component	What It Does
Python SDK	Detect & read plates in your own code — CPU, CUDA, DirectML
REST API Server	Production server with web dashboard, history, alerts, watchlist
Mobile App	On-device ANPR for iOS & Android (~100-160 ms per scan)
Road Object Detection	Detect persons, vehicles, and 2-wheelers (MareArts Road Objects)
Vehicle Intelligence	Cloud AI: make, model, color, type, face, plate nation, server-side OCR

---

Mobile App

On-device ANPR for iOS and Android — parking, security, fleet management.

• On-device AI — ~100-160 ms, 100% offline capable
• Team Work — leader/member collaboration with shared data
• Webhooks — Discord, Slack, or any custom endpoint
• Cloud Sync — sync across devices via marearts.com
• Vehicle Intelligence — make, model, color, type, face, plate nation, server-side OCR
• Map View, CSV Export, Whitelist/Blacklist, Stats

No additional license required — the app works as your ANPR license.

Mobile App Guide →

---

Python SDK

Detect and read license plates in 5 lines of Python.

pip install marearts-anpr

from marearts_anpr import (
    ma_anpr_detector_v16, ma_anpr_ocr_v16, marearts_anpr_from_image_file
)

detector = ma_anpr_detector_v16("640p_fp32", user_name, serial_key, signature)
ocr = ma_anpr_ocr_v16("fp32", "univ", user_name, serial_key, signature)

result = marearts_anpr_from_image_file(detector, ocr, "car.jpg")
print(result)
# {'results': [{'ocr': 'ABC1234', 'ocr_conf': 99, 'ltrb': [...], ...}], ...}

Full SDK Documentation & Examples →

---

REST API Server

Production-ready server with web dashboard — zero code deployment.

ma-anpr config                     # one-time credential setup
ma-anpr server start               # http://localhost:8000

curl -X POST http://localhost:8000/api/anpr -F "image=@car.jpg"

Dashboard	Detection	History

20+ API endpoints: detect, batch, history, watchlist, alerts, stats, config, model selection, and more.

Server Documentation →

---

Docker

Deploy the ANPR server as a Docker container with GPU support — no setup, no dependencies.

docker run -d --gpus all -p 8000:8000 \
  -e MAREARTS_ANPR_USERNAME="your@email.com" \
  -e MAREARTS_ANPR_SERIAL_KEY="your_serial_key" \
  -e MAREARTS_ANPR_SIGNATURE="your_signature" \
  marearts-anpr-server:latest

curl -X POST http://localhost:8000/api/anpr -F "image=@car.jpg"

Includes CUDA acceleration, automatic CPU fallback, web dashboard, and all 20+ API endpoints.

Docker Guide →

---

Vehicle Intelligence (Cloud AI)

Go beyond plate text — identify the vehicle itself with a single API call.

Feature	Example Output
Make	Toyota, BMW, Hyundai, …
Model	Camry, 3 Series, Tucson, …
Color	White, Black, Silver, …
Type	Sedan, SUV, Truck, Van, …
Face	front, rear
Plate Nation	KR, DE, US, …
Server-side OCR	Cross-check plate text via cloud

from marearts_anpr import ma_anpr_mmc

mmc = ma_anpr_mmc(user_name, serial_key, signature)
info = mmc.enrich(image, anpr_results)
print(info["make"], info["model"], info["color"])
# Toyota Camry White

Run local ANPR first, then pass the results to mmc.enrich() for cloud enrichment. See the full MMC example →

Also available via the REST API server — just call the /api/anpr/mmc endpoint.

---

Road Object Detection

Detect persons, vehicles, and 2-wheelers in real-time — included in your ANPR license at no extra cost.

from marearts_road_objects import ma_road_object_detector

rod = ma_road_object_detector("640p_fp32", user_name, serial_key, signature)
result = rod.detect("street.jpg")
# [{'class': 'car', 'conf': 0.97, 'ltrb': [...]}, {'class': 'person', ...}]

Use it alongside ANPR for complete traffic scene understanding — plate reading + vehicle/pedestrian awareness in one pipeline.

Road Objects Documentation →

---

Supported Regions

80+ countries across 12 regional groups with per-country character sets for maximum accuracy.

Region	Countries	Code
🇪🇺 Europe+	37 countries (EU + Balkans + Indonesia)	eu
🇷🇺 Ex-USSR	15 countries (Russia, Ukraine, Kazakhstan, …)	ru
🌏 Asia	17 countries (Japan, Thailand, Vietnam, …)	asia
🇺🇸 North America	USA, Canada, Mexico	na
🇧🇷 South America	Brazil, Argentina	sa
🇿🇦 Africa	South Africa, Nigeria	af
🇦🇺 Oceania	Australia, New Zealand	oc
🇬🇧 UK	England, Scotland, Wales, N. Ireland	uk
🇨🇳 China	All provinces	cn
🇰🇷 Korea	All plate types	kr
🇯🇵 Japan	All prefectures	jp
🌍 Universal	All of the above	univ

Pass a 2-letter country code (e.g. de, au, th) for best accuracy, or use a group code. Unknown codes fall back to univ.

Full Region & Country Reference →

---

Performance (V16)

V16 ships one unified model for detection and one for OCR — no model-size selection needed.

Component	Model	Precision	Notes
Detector	320p_fp32	fp32 / int8	Fast — ~2× speed of 640p
Detector	640p_fp32	fp32 / int8	High detection — best for distant/small plates
OCR	fp32	fp32 / int8	Single model, all regions, dynamic switching

int8 variants available for smaller footprint and faster inference on edge devices.

---

Testing

We include a full test suite to verify SDK and server functionality.

cd tests/
python test_sdk.py       # SDK unit tests
python test_server.py    # Server API integration tests

Test Documentation →

---

More from MareArts

• MareArts Xcolor — Color extraction and similarity
• MareArts MAST — Real-time panoramic stitching

---

Support

Homepage	marearts.com
License	marearts.com/products/anpr
Live Demo	live.marearts.com
Contact	hello@marearts.com
YouTube	Video Examples

---

© 2026 MareArts. All rights reserved.