Measuring
and
predicting
Carbon
Emission
(CE)
is
important
to
enabling
the
main
culprit
of
various
urgent
environmental
issues
including
global
warming.
However,
prior
studies
did
not
fully
incorporate
impact
micro-level
urban
streetscapes,
which
might
lead
biased
prediction
CE.
To
fill
gap,
we
developed
an
effective
framework
predict
residential
CE
in
areas
from
widely
existing
publicly
available
street-view
images
(SVI)
using
machine
learning.
First,
used
a
semantic
segmentation
algorithm
classify
more
than
30
streetscape
elements
SVI
describe
built
environment
whose
features
affect
transportation
Second,
based
on
streetscapes
quantified,
trained
10-fold
cross-validation
method
with
learning
models
at
1KM
grid
level
data
PlanetData.
We
found
first,
such
as
sidewalks,
roads,
fences,
buildings,
walls
are
significantly
correlated
presence
buildings
subtle
(e.g.,
walls,
fences)
indicates
higher-density
related
Third,
vegetation
trees
grass)
reversely
Our
findings
shed
light
feasibility
single
open
source
(i.e.,
SVI)
effectively
model
neighborhood-level
for
regions
across
diverse
forms.
useful
planners
inform
new
town
development
regeneration
towards
low
goals.
ISPRS annals of the photogrammetry, remote sensing and spatial information sciences,
Journal Year:
2024,
Volume and Issue:
X-4/W5-2024, P. 211 - 218
Published: June 27, 2024
Abstract.
Urban
digital
twins,
and
3D
city
models
underpinning
them,
provide
novel
solutions
to
urban
management
but
tend
overlook
the
human
element.
The
trending
research
on
perception
reveals
people’s
perspective
towards
interpreting
experiencing
built
environment.
Advancing
representation
of
building
physics
descriptive
information
in
we
establish
addition
integration
notion
how
humans
perceive
buildings.
Unlocking
a
new
dimension
our
domain,
this
concept
can
facilitate
broader
adoption
semantic
data
socio-economic
fields
across
various
domains,
advance
existing
use
cases
GIS.
This
work
is
first
instance
integrating
such
attributes
models,
which
have
traditionally
been
confined
physical
objective
measures.
visual
each
evaluated
based
images
extracted
from
street
view
images.
We
add
as
an
CityJSON
dataset
representing
thousands
buildings
Amsterdam,
Netherlands.
To
robust
sustainable
integration,
develop
Extension
accommodate
validate
its
schema
successfully,
visualise
dataset.
Further,
present
two
demonstrate
usability
for
downstream
analysis.
One
concurrent
clustering
morphology
perception,
while
other
conducting
attribute-based
query
that
enables
stakeholders
identify
particular
interest
combining
both
traditional
attributes.
Heritage Science,
Journal Year:
2024,
Volume and Issue:
12(1)
Published: Dec. 3, 2024
Abstract
Under
the
background
of
transformation
resource-based
cities,
heritage
as
symbolic
cultural
representation
plays
a
synergistic
role
in
revitalizing
urban
vibrancy.
A
majority
contemporary
research
focuses
on
specific
restoration
and
renovation.
However,
scant
literature
has
been
concerned
with
an
integrated
corridor
upgrading
framework
from
spatial
quality
perspective,
which
limited
effects
promoting
socio-cultural
development.
This
aims
to
evaluate
through
GIS-based
environmental
model
(ESM)
multi-source
data
verification
AI-based
image
semantic
segmentation
analysis,
cultivating
suggestions
for
management
revitalize
holistic
urban–rural
areas.
The
takes
city,
Fengfeng
Mining
District
(FMD)
Handan,
China,
case.
found
heterogeneity
evaluation
results
their
geographical
distribution,
image-based
evidenced
suitability
reliability
ESM
assessment.
proposes
quantitative
assessing
improving
corridors.
optimization
corridors
should
combine
comprehensive,
precise,
people-oriented
assessment,
analysis
method
could
be
effective
decision-making
support
system.
Measuring
and
predicting
Carbon
Emission
(CE)
is
important
to
enabling
the
main
culprit
of
various
urgent
environmental
issues
including
global
warming.
However,
prior
studies
did
not
fully
incorporate
impact
micro-level
urban
streetscapes,
which
might
lead
biased
prediction
CE.
To
fill
gap,
we
developed
an
effective
framework
predict
residential
CE
in
areas
from
widely
existing
publicly
available
street-view
images
(SVI)
using
machine
learning.
First,
used
a
semantic
segmentation
algorithm
classify
more
than
30
streetscape
elements
SVI
describe
built
environment
whose
features
affect
transportation
Second,
based
on
streetscapes
quantified,
trained
10-fold
cross-validation
method
with
learning
models
at
1KM
grid
level
data
PlanetData.
We
found
first,
such
as
sidewalks,
roads,
fences,
buildings,
walls
are
significantly
correlated
presence
buildings
subtle
(e.g.,
walls,
fences)
indicates
higher-density
related
Third,
vegetation
trees
grass)
reversely
Our
findings
shed
light
feasibility
single
open
source
(i.e.,
SVI)
effectively
model
neighborhood-level
for
regions
across
diverse
forms.
useful
planners
inform
new
town
development
regeneration
towards
low
goals.
