Technology Profile |
Vehicle classification and ALPR
in one customizable system
CARRIDA-Technologies’
mission is to read license
plates and classify vehicles,
as is reflected in its name: ‘Car’
as in vehicles, and ‘Rida’,
pronounced ‘reader’. Its
software module for automatic
license plate recognition (ALPR)
is designed for all kinds of
applications.
The CARRIDA software
engine is easy to integrate into
existing security and surveillance
ITS applications. It is fully
hardware independent and runs
on Windows and Linux PCs, as
well as advanced RISC machine
(ARM) architectures on Linux and
Android. A very fast processing
speed of 15-20ms per frame
enables the reading of license
plates from vehicles traveling up
to 150mph (240km/h).
The software can read license
plates even when they are
partially damaged, dirty or
distorted. It also works on noisy,
low-contrast or low-resolution
images. Readings of up to 99%
accuracy can be achieved under
real-world conditions. The
software can be used with any C
and C++ project via an
application program interface
(API), so the user can program a
broad range of applications and
can easily integrate CARRIDA’s
module into existing systems.
Working worldwide
The CARRIDA software engine
is able to handle the plates of
most countries worldwide,
including state/country
recognition. By this important
property CARRIDA’s partners
can rely on one proven package
for any application anywhere.
This reduces the development
effort for companies offering
solutions globally.
In addition to the license plate
reading, a new Make&Model
recognition function offers even
more security and functionality
| Need to know by allowing the user to check
Key features of
CARRIDA software
Traffic Technology International November/December 2019
www.TrafficTechnologyToday.com
048
computer itself has been trained
how to recognize a particular
vehicle type. Therefore, it is
possible to run much more
complex and thus more reliable
recognitions. This was
introduced with software
version 4.3. Like the ALPR
CARRIDA software module,
the Make&Model module is
hardware and manufacturer
independent.
‘Read’ cars with Android
The newest technological
achievement from CARRIDATechnologies
is an Androidoperated
software development
kit (SDK). A reading time of
approximately 300ms can be
reached for full HD images on
medium-class Android mobile
devices. A sample application
can be found on the company’s
website. CARRIDA for Android
is fully optimized for this
embedded ARM architecture.
Furthermore, the SDK includes
color recognition for GCC, state
recognition and stacked
characters for different countries.
A single-unit solution
A small, intelligent embedded
camera can be combined
with the CARRIDA software
to create a self-contained,
standalone solution. Using
an ARM camera can result in
an impressively low power
consumption of about 3 watts.
It is able to process data
collected without any additional
hardware, thereby saving space
and money. The entire ALPR
process takes place directly on
the camera. Whether access
control, red light monitoring
or toll systems – thanks to the
integrated CARRIDA software
engine, the camera handles all
processes. Its memory can also
be used to store self-defined
blacklists and whitelists for
vehicles, which are then
> Can read cars at speeds of
up to 150mph (240km/h)
> API Compatible with C and
C++ projects
> Recognition of plates
worldwide
> Make&Model vehicle
classification function
with deep learning
> Android SDK enables
300ms reading time on
Android mobile devices
> On-camera processing
whether the vehicle type matches
the number plate. It is also
possible to categorize vehicles
whose number plates are not
stored in the ALPR system itself.
This allows intelligent traffic
management systems and
parking operators to increase
their efficiency. Readings with
a positive match between the car
and its license plate can be
archived in the same system.
The software module is
based on the latest artificial
intelligence (AI) technology,
ensuring high recognition rates
of vehicle makes and models
from all over the world.
CARRIDA uses deep
learning for Make&Model
recognition. This means that the
/www.TrafficTechnologyToday.com