naxtlab

BUSINESS

A company specializing in automation technology that combines deep learning and image processing technologies

Beyond ALPR

개요

Summary

Service
Conceptual Chart

Download Product Introduction Visit our Youtube Channel

Introduction

Beyond ALPR (Automatic License Plate Recognition) is a vehicle recognition solution and service that utilizes state-of-the-art Artificial Intelligence technology. ALPR's algorithm involves detecting license plates, vehicle colors, and brands to accurately derive at the vehicle's identity.

구성

Components

Components

  • 01.
    License Plate Recognition
  • Expand and enhance plate number
  • Recognize license plate type
  • Recognize plate number / characters
  • 02.
    Vehicle Type Recognition
  • Recognize car brand
  • Recognize model
  • Recognize model year
  • Recognize color
  • 03.
    Vehicle Ownership Verification
  • Recognize license plate/vehicle type
  • Verify identity
  • Verify vehicle registration certificate
  • 04.
    Exterior Damage Recognition
  • Recognize vehicle appearance
  • Recognize car damage location and type
  • 05.
    Under vehicle imaging and damage recognition
  • Portable under vehicle imaging solution
  • Recognize under vehicle damage position

Beyond ALPR
View

License plate recognition
Vehicle type recognition
Exterior Damage Recognition
Verifying car registration

주요 기능

Main Feature

Technical Details Diagram

Characteristics
of license plate
recognition technology

Expand and Enhance using Super Resolution technology
  • In order to raise recognition rate of photos taken by cameras at the entrance of parking lots to cameras installed on patrol cars, Super Resolution technology is applied to expand and correct the image in real-time so that the images can be identified easily with naked eyes.
  • Example

50x25 Low Resolution Image
(Difficult to identify)

200x100 High Resolution Image
(Apply super resolution)

200x100 Original Image
(Ground Truth)

Provided in various forms, from embedded devices to cloud services
  • The technology can be distributed and serviced in various forms from imbeded devices to cloud - services
    according to the intended use without sacrificing accuracy.
Supports both old and new license plates
  • Supports both old and new license plates

The history of license plates in Korea

Division Shape
Private vehicle
  • black lettering on white background
  • no area identifier
Commercial vehicle
  • black lettering on yellow background
  • name of the registration area(city / province) on the left side
Embassy or Consulate vehicle
  • white lettering on royal blue background
  • two letters on the left side, 123-123 format
Electric vehicle
  • black lettering on skyblue background
  • electric vehicle symbols
1979 -1995
  • white lettering on green background
  • characters of area name and one digit on the top of the plate
  • 가 1234 format(one korean letter followed by four numbers)
1996 - 2003
  • white lettering on green background
  • characters of area name and two numbers on the top of the plate
  • the same format as previous one
2004 - 2006
  • white lettering on green background
  • On the top of the plate, two numbers and one korean-character, area name exlcuded
  • four numbers format
2006 - 08. 2019
  • black lettering on white background
  • one column layout from previous two column
  • European license plates format
Sep, 2019 ~
  • start with three numbers due to lack of available license number
  • font style for better recognition

Vehicle type
recognition technology

Capable of recognizing not only the maker and brand, but also the detailed model
  • Capable of recognizing various information at once through the following vehicle type recognition process

Step 1 : Recognize vehicle frame

  • Recognizes the vehicle frame and
    crops unnecessary background area
  • Can concurrently recognize multiple vehicles from a piece of photo

Step 2 : Recognize direction of
the photo

  • Recognize one of the eight directions below and select the optimized learning model for vehicle recognition

Step 3 : Recognize vehicle type /
model year

  • Recognition results can be verified through the heat map below
Capable of recognizing vehicles regardless of the direction of photo shooting
Identical models can also be recognized according to their model years
  • Example

Demo Video

Go To Youtube