Journal of Computing in Civil Engineering,
Год журнала:
2023,
Номер
38(1)
Опубликована: Сен. 25, 2023
Construction
materials
undergo
appearance
and
textural
changes
during
the
construction
process.
Accurate
recognition
of
these
is
critical
for
effectively
understanding
status;
however,
recognizing
various
levels
detailed
material
conditions
not
sufficiently
explored.
The
primary
challenge
in
availability
labeled
training
data.
To
address
this
challenge,
study
proposes
a
novel
state-of-the-art
deep
learning
model
that
leverages
transfer
learning,
utilizing
pretrained
Inception
V3
to
knowledge
limited
data
set
context.
This
enables
learn
meaningful
representations
from
data,
enhancing
its
ability
accurately
classify
conditions.
In
addition,
gray-level
co-occurrence
matrix
(GLCM)–based
texture
features
are
extracted
images
capture
materials,
which
then
concatenated
with
transferred
convolutional
neural
network
(CNN)
create
more
comprehensive
representation
proposed
achieved
an
overall
classification
accuracy
95%
71%
(208
images)
very
small
(70
sets,
respectively.
It
outperformed
different
experimental
architectures,
including
CNN
models
developed
using
without
augmentation,
augmentation
separate
local
binary
pattern
(LBP)
GLCM
super
learners
trained
augmented
findings
suggest
model,
combines
GLCM-based
features,
effective
even
can
contribute
improved
management
monitoring.
Applied Sciences,
Год журнала:
2023,
Номер
13(4), С. 2304 - 2304
Опубликована: Фев. 10, 2023
Regular
inspection
and
monitoring
of
buildings
infrastructure,
that
is
collectively
called
the
built
environment
in
this
paper,
critical.
The
includes
commercial
residential
buildings,
roads,
bridges,
tunnels,
pipelines.
Automation
robotics
can
aid
reducing
errors
increasing
efficiency
tasks.
As
a
result,
robotic
has
become
significant
research
topic
recent
years.
This
review
paper
presents
an
in-depth
qualitative
content
analysis
269
papers
on
use
robots
for
infrastructure.
found
nine
different
types
systems,
with
unmanned
aerial
vehicles
(UAVs)
being
most
common,
followed
by
ground
(UGVs).
study
also
five
applications
monitoring,
namely,
maintenance
inspection,
construction
quality
progress
as-built
modeling,
safety
inspection.
Common
areas
investigated
researchers
include
autonomous
navigation,
knowledge
extraction,
motion
control
sensing,
multi-robot
collaboration,
implications,
data
transmission.
findings
provide
insight
into
developments
field
will
benefit
researchers,
facility
managers,
developing
implementing
new
solutions.
Automation in Construction,
Год журнала:
2022,
Номер
139, С. 104312 - 104312
Опубликована: Май 6, 2022
Construction
sites
are
highly
hazardous
due
to
the
dynamic
interaction
between
workers
and
moving
equipment,
with
high
fatality
rates
caused
by
collision
falling
from
height,
etc.
Hence,
identifying
unsafe
behaviors
among
is
crucial
for
enhancing
site
safety,
such
as
tracking
their
on-site
movement
personal
protective
equipment
(PPE).
Vision-based
video
processing
has
been
actively
used
automatically
recognize
on
construction
sites.
However,
existing
studies
mainly
monitor
within
a
single
camera
capturing
only
small
sub-region.
As
typically
move
around
fairly
large
sites,
continuously
across
multiple
cameras
would
enable
more
comprehensive
behavioral
analyses.
this
paper
proposes
framework
monitoring
safety
compliance
workers,
combining
worker
re-identification
(ReID)
PPE
classification.
Deep
learning-based
approaches
developed
address
challenges
these
two
tasks
respectively.
For
ReID,
new
loss
function
named
similarity
designed
encourage
deep
learning
models
learn
discriminative
human
features,
realizing
robust
of
individual
workers.
classifying
statuses,
weighted-class
strategy
proposed
mitigate
model
bias
when
given
imbalanced
samples
classes,
improved
performance
despite
limited
training
samples.
By
ReID
classification
results,
workflow
log
any
incident
not
wearing
necessary
PPEs.
With
an
actual
dataset,
methods
improve
4%
13%
accuracies
respectively,
which
will
facilitate
analytics
inspection
Developments in the Built Environment,
Год журнала:
2023,
Номер
16, С. 100247 - 100247
Опубликована: Окт. 11, 2023
Effective
progress
monitoring
is
ineviTable
for
completing
the
construction
of
building
and
infrastructure
projects
successfully.
In
this
digital
transformation
era,
with
data-centric
management
control
approach,
effectiveness
methods
expected
to
improve
dramatically.
"Digital
Twin,"
which
creates
a
bidirectional
communication
flow
between
physical
entity
its
counterpart,
found
be
crucial
enabling
technology
information-aware
decision-making
systems
in
manufacturing
other
automotive
industries.
Recognizing
benefits
production
construction,
researchers
have
proposed
Digital
Twin
Construction
(DTC).
DTC
leverages
information
modeling
processes,
lean
practices,
on-site
data
collection
mechanisms,
Artificial
Intelligence
(AI)
based
analytics
improving
planning
processes.
Progress
monitoring,
key
component
control,
can
significantly
benefit
from
DTC.
However,
some
knowledge
gaps
still
need
filled
practical
implementation
built
environment
domain.
This
research
reviews
existing
vision-based
methods,
studies
evolution
automated
research,
highlights
methodological
technological
that
must
addressed
DTC-based
predictive
monitoring.
Subsequently,
it
proposes
framework
closed-loop
through
Finally,
way
forward
fully
automated,
real-time
upon
concept
proposed.
Alexandria Engineering Journal,
Год журнала:
2024,
Номер
88, С. 80 - 90
Опубликована: Янв. 12, 2024
Building
information
modeling
(BIM)
technology
can
organically
combine
data
and
virtual
reality,
compare
them
with
actual
construction
objects
to
realize
the
smart
collaboration
of
entity
model
in
processes,
greatly
reducing
early
stage
mistakes
improve
efficiency.
This
paper
proposed
an
automation
system
framework
based
on
BIM
platform,
discussed
element
configuration
finally
elaborated
working
mechanism
automated
platform.
The
study
results
show
that
platform
supports
rapid
large-scale
three-dimensional
scenes
precise
integration
multi-source
data,
bring
together
asset
from
different
sources
form
open,
safe,
accessible
digital
environment.
application
cost
management
has
brought
huge
economic
benefits,
efficiency
been
increased
by
65%,
period
shortened
30%,
labor
intensity
reduced
27%,
productivity
39%,
which
considerable
indirect
benefits
projects.
