Technology Profile |
Advanced platform
for traffic analytics
vehicle categories and
pedestrians are determined
by analysing visual appearance.
Such systems can work
repetitively and reliably 24/7
while still reducing the process
complexity. Solutions like
GoodVision Video Insights
collect all physical aspects
of a vehicle or pedestrian’s
behaviour on the traffic scene.
This includes capturing its full
trajectory which is a baseline
for the deeper analysis of traffic
flows and identification of
important traffic events. The
company’s resources can then
be assigned to more valueadded
tasks in keeping with
staff skills.
When choosing a system
for traffic data analysis,
following key data areas are
essential to consider.
The first is having exact
vehicle trajectories. This is
helpful not only for visual
analytics but also for calculation
of further behavioural metrics.
Next comes traffic counting.
Defining traffic movements
interactively by entry and exit
areas regardless of the type of
by traffic modellers reviewing
spreadsheets and traffic video.
Traditional data collection
techniques are either resourceintensive
or time-consuming.
Human counters have a high
degree of inaccuracy and
provide static one-purpose
data. Invasive sensors must
be physically installed and
maintained, and have limited
capabilities. Older solutions
with computer vision also
have limitations such as high
dependence on scene setup
and climatic conditions, and an
inability to deal with shadows,
reflections, low light situations
and traffic density with
overlapping vehicles or vehicles
moving in groups.
Deep traffic data
All the limitations of methods
described lead to infrequent
analyses based on small samples
or outdated data. If the data is
missing, there is an option to
estimate traffic volume across
the entire road network using
a mathematical model. But
is it accurate?
Traditional solutions provide
single-purpose traffic data in
spreadsheets based on strict
input conditions which do not
allow the analysis required to
understand behaviour of the
traffic attendants on roads and
pathways. Advanced
information must be identified
by traffic analysts manually
requiring time and effort.
If cities and organisations
are really making decisions
with 30-year implications based
on potentially poor data the
question is how to get it reliably
for traffic analysts and
modellers? The uncertainty of
how to obtain traffic data in the
electronic format ends with
emersion in process automation
and Artificial Intelligence (AI).
With modern AI-based systems,
Traffic data is the raw
material of traffic
engineering business.
Engineers, analysts and
modellers all rely on its accuracy
and high-quality as a crucial
starting point to deliver projects.
Traffic engineering
companies and smart cities
obtain information about traffic
behaviour in order to prevent
traffic congestion, to enact
efficient street design and
increase traffic safety and
fluidity on the roads. That
becomes a huge bottleneck of
data-gathering which is mainly
reliant on human counters.
Having good traffic data
analysis in place is a necessary
baseline of every traffic
engineering project.
Traffic data analysis is
a fragmented market, with
confusion as to how and where
information can be found.
Common scenarios for
obtaining data for traffic
projects include traffic video
recording done by suppliers,
traffic counting is outsourced
from other suppliers – mainly
human counters, and advanced
traffic micro-analyses are
performed ad-hoc, manually
Left: Xxxxxxx
the junction is how to prevent
the user error and to perform
TMC queries repeatedly with
altered filters.
Third, knowing precise travel
time of individual vehicles or
pedestrian is valuable for exact
speed estimation and analysis
of the vehicle acceleration inside
of the junction.
Fourth, collected traffic
data should include information
about occupancy time, essential
when analysing waiting times
on lights or on vehicle yielding
on turns.
Fifth, time-gaps between
vehicles shall be calculated from
individual vehicle passages.
This helps traffic managers
to understand traffic flow
behaviour, identify aggressive
drivers, and spot traffic
congestion
Next, when it comes to
jaywalkers, u-turns and illegal
manoeuvres, data collection
should not be limited to road
and crosswalk corridors. It
should be captured anywhere
on the scene to detect jaywalkers
or illegal movements such as
wrong way running.
| Need to know
Video Insights by
Goodvision offers
its clients an easy
way of collecting and
analysing traffic data
> Combines 95% accurate
traffic data collection with
deep data analytics and
reporting
> Compatible with existing
camera infrastructure
> Data without additional
deployment costs
Traffic Technology International January/February 2020
www.TrafficTechnologyToday.c 068 om
/www.TrafficTechnologyToday.c