Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies,
Год журнала:
2025,
Номер
9(1), С. 1 - 38
Опубликована: Март 3, 2025
Human
activity
recognition
(HAR)
using
ambient
sensors
in
smart
homes
has
numerous
applications
for
human
healthcare
and
wellness.
However,
building
general-purpose
HAR
models
that
can
be
deployed
to
new
home
environments
requires
a
significant
amount
of
annotated
sensor
data
training
overhead.
Most
vary
significantly
their
layouts,
i.e.,
floor
plans
the
specifics
embedded,
resulting
low
generalizability
trained
specific
homes.
We
address
this
limitation
by
introducing
novel,
layout-agnostic
modeling
approach
systems
utilizes
transferrable
representational
capacity
natural
language
descriptions
raw
data.
To
end,
we
generate
Textual
Descriptions
Of
Sensor
Triggers
(TDOST)
encapsulate
surrounding
trigger
conditions
provide
cues
underlying
activities
models.
Leveraging
textual
embeddings,
rather
than
data,
create
predict
standard
across
without
(re-)training
or
adaptation
target
Through
an
extensive
evaluation,
demonstrate
effectiveness
TDOST-based
unseen
through
experiments
on
benchmark
Orange4Home
CASAS
datasets.
Furthermore,
conduct
detailed
analysis
how
individual
components
our
affect
downstream
performance.
Sensors,
Год журнала:
2023,
Номер
23(9), С. 4259 - 4259
Опубликована: Апрель 25, 2023
Work-related
musculoskeletal
disorders
(WMSDs)
are
a
major
contributor
to
disability
worldwide
and
substantial
societal
costs.
The
use
of
wearable
motion
capture
instruments
has
role
in
preventing
WMSDs
by
contributing
improvements
exposure
risk
assessment
potentially
improved
effectiveness
work
technique
training.
Given
the
versatile
potential
for
wearables,
this
article
aims
provide
an
overview
their
application
related
prevention
trunk
upper
limbs
discusses
challenges
technology
support
measures
future
opportunities,
including
research
needs.
relevant
literature
was
identified
from
screening
recent
systematic
reviews
overviews,
more
studies
were
search
using
Web
Science
platform.
Wearable
enables
continuous
measurements
multiple
body
segments
superior
accuracy
precision
compared
observational
tools.
also
real-time
visualization
exposures,
automatic
analyses,
feedback
user.
While
miniaturization
usability
wearability
can
expand
occupational
settings
increase
among
safety
health
practitioners,
several
fundamental
remain
be
resolved.
opportunities
increased
usage
devices
work-related
may
require
international
collaborations
creating
common
standards
measurements,
metrics,
which
epidemiologically
based
categories
disorders.
Sensors,
Год журнала:
2025,
Номер
25(3), С. 724 - 724
Опубликована: Янв. 25, 2025
Human
activity
recognition
(HAR)
using
radar
technology
is
becoming
increasingly
valuable
for
applications
in
areas
such
as
smart
security
systems,
healthcare
monitoring,
and
interactive
computing.
This
study
investigates
the
integration
of
convolutional
neural
networks
(CNNs)
with
conventional
signal
processing
methods
to
improve
accuracy
efficiency
HAR.
Three
distinct,
two-dimensional
techniques,
specifically
range-fast
Fourier
transform
(FFT)-based
time-range
maps,
time-Doppler-based
short-time
(STFT)
smoothed
pseudo-Wigner–Ville
distribution
(SPWVD)
are
evaluated
combination
four
state-of-the-art
CNN
architectures:
VGG-16,
VGG-19,
ResNet-50,
MobileNetV2.
positions
radar-generated
maps
a
form
visual
data,
bridging
image
representation
domains
while
ensuring
privacy
sensitive
applications.
In
total,
twelve
preprocessing
configurations
analyzed,
focusing
on
trade-offs
between
complexity
accuracy,
all
which
essential
real-time
Among
these
results,
MobileNetV2,
combined
STFT
preprocessing,
showed
an
ideal
balance,
achieving
high
computational
rate
96.30%,
spectrogram
generation
time
220
ms
inference
2.57
per
sample.
The
comprehensive
evaluation
underscores
importance
interpretable
features
resource-constrained
environments,
expanding
applicability
radar-based
HAR
systems
augmented
reality,
autonomous
edge
Sensors,
Год журнала:
2023,
Номер
23(11), С. 5281 - 5281
Опубликована: Июнь 2, 2023
Smart
living,
a
concept
that
has
gained
increasing
attention
in
recent
years,
revolves
around
integrating
advanced
technologies
homes
and
cities
to
enhance
the
quality
of
life
for
citizens.
Sensing
human
action
recognition
are
crucial
aspects
this
concept.
living
applications
span
various
domains,
such
as
energy
consumption,
healthcare,
transportation,
education,
which
greatly
benefit
from
effective
recognition.
This
field,
originating
computer
vision,
seeks
recognize
actions
activities
using
not
only
visual
data
but
also
many
other
sensor
modalities.
paper
comprehensively
reviews
literature
on
smart
environments,
synthesizing
main
contributions,
challenges,
future
research
directions.
review
selects
five
key
i.e.,
Technology,
Multimodality,
Real-time
Processing,
Interoperability,
Resource-Constrained
they
encompass
critical
required
successfully
deploying
living.
These
domains
highlight
essential
role
sensing
play
developing
implementing
solutions.
serves
valuable
resource
researchers
practitioners
seeking
further
explore
advance
field
IEEE Access,
Год журнала:
2023,
Номер
11, С. 105140 - 105169
Опубликована: Янв. 1, 2023
Human
activity
recognition
(HAR)
has
become
increasingly
popular
in
recent
years
due
to
its
potential
meet
the
growing
needs
of
various
industries.
Electromyography
(EMG)
is
essential
clinical
and
biological
settings.
It
a
metric
that
helps
doctors
diagnose
conditions
affect
muscle
activation
patterns
monitor
patients'
progress
rehabilitation.
Despite
widespread
Application,
existing
methods
for
recording
interpreting
EMG
data
need
more
signal
detection
robust
categorization.
Recent
material
science
Artificial
Intelligence
(AI)
developments
have
significantly
improved
detection.
With
an
elderly
patient
population,
HAR
used
Activities
Daily
Living
(ADLs)
healthcare
also
being
security
settings
identify
suspect
behavior,
Surface
(sEMG)
non-invasive
treatment
since
it
monitors
contractions
during
exercise.
sEMG
AI
revolutionized
systems
years.
Sophisticated
are
required
recognize,
break
down,
manufacture,
classify
signals
obtained
by
muscles.
This
review
summarizes
research
papers
based
on
with
EMG.
made
tremendous
contributions
biomedical
classification.
The
different
approaches
preprocessing,
feature
extraction,
Reduction
techniques,
Deep
Learning
(DL)
Machine
(ML)
classification
then
briefly
explained.
We
focused
latest
ML/DL
HAR,
Hardware
involved
Application.
discovered
open
issues
future
direction
may
point
new
lines
inquiry
ongoing
toward
EMG-based
Algorithms,
Год журнала:
2023,
Номер
16(2), С. 77 - 77
Опубликована: Фев. 1, 2023
Over
the
last
few
years,
human
activity
recognition
(HAR)
has
drawn
increasing
interest
from
scientific
community.
This
attention
is
mainly
attributable
to
proliferation
of
wearable
sensors
and
expanding
role
HAR
in
such
fields
as
healthcare,
sports,
monitoring.
Convolutional
neural
networks
(CNN)
are
becoming
a
popular
approach
for
addressing
problems.
However,
this
method
requires
extensive
training
datasets
perform
adequately
on
new
data.
paper
proposes
novel
deep
learning
model
pre-trained
scalograms
generated
using
continuous
wavelet
transform
(CWT).
Nine
CNN
architectures
different
CWT
configurations
were
considered
select
best
performing
combination,
resulting
evaluation
more
than
300
models.
On
source
KU-HAR
dataset,
selected
achieved
classification
accuracy
an
F1
score
97.48%
97.52%,
respectively,
which
outperformed
contemporary
state-of-the-art
works
where
dataset
was
employed.
target
UCI-HAPT
proposed
resulted
maximum
F1-score
increase
0.21%
0.33%,
whole
2.82%
2.89%,
subset.
It
concluded
that
usage
model,
particularly
with
frozen
layers,
results
improved
performance,
faster
training,
smoother
gradient
descent
small
datasets.
use
sufficiently
large
may
lead
negative
transfer
degradation.
IEEE Access,
Год журнала:
2023,
Номер
11, С. 19122 - 19134
Опубликована: Янв. 1, 2023
Security
has
always
been
a
significant
concern
since
the
dawn
of
human
civilization.
That
is
why
we
build
houses
to
keep
ourselves
and
our
belongings
safe.
And
do
not
hesitate
spend
lot
on
front-door
locks
install
CCTV
cameras
monitor
security
threats.
This
paper
presents
an
innovative
automatic
Front
Door
(FDS)
algorithm
that
uses
Human
Activity
Recognition
(HAR)
detect
four
different
threats
at
front
door
from
real-time
video
feed
with
73.18%
accuracy.
The
activities
are
recognized
using
combination
GoogleNet-BiLSTM
hybrid
network.
network
receives
camera
classifies
activities.
proposed
this
classification
alert
any
attempts
break
by
kicking,
punching,
or
hitting.
Furthermore,
FDS
effective
in
detecting
gun
violence
door,
which
further
strengthens
security.
(HAR)-based
novel
demonstrates
potential
ensuring
better
safety
71.49%
precision,
68.2%
recall,
F1-score
0.65.
IEEE Communications Surveys & Tutorials,
Год журнала:
2024,
Номер
26(2), С. 890 - 929
Опубликована: Янв. 1, 2024
Due
to
the
ever-growing
powers
in
sensing,
computing,
communicating
and
storing,
mobile
devices
(e.g.,
smartphone,
smartwatch,
smart
glasses)
become
ubiquitous
an
indispensable
part
of
people's
daily
life.
Until
now,
have
been
adopted
many
applications,
e.g.,
exercise
assessment,
life
monitoring,
human-computer
interactions,
user
authentication,
etc.
Among
various
Human
Activity
Recognition
(HAR)
is
core
technology
behind
them.
Specifically,
HAR
gets
sensor
data
corresponding
human
activities
based
on
built-in
sensors
devices,
then
adopts
suitable
recognition
approaches
infer
type
activity
data.
The
last
two
decades
witnessed
ever-increasing
research
HAR.
However,
new
challenges
opportunities
are
emerging,
especially
for
devices.
Therefore,
this
paper,
we
review
aiming
advance
following
area.
Firstly,
give
overview
including
general
rationales,
main
components
challenges.
Secondly,
analyze
progress
from
each
aspect,
activities,
data,
preprocessing,
approaches,
evaluation
standards
application
cases.
Finally,
present
some
promising
trends
future
research.
Sensors,
Год журнала:
2025,
Номер
25(2), С. 441 - 441
Опубликована: Янв. 13, 2025
This
paper
presents
an
approach
for
event
recognition
in
sequential
images
using
human
body
part
features
and
their
surrounding
context.
Key
points
were
approximated
to
track
monitor
presence
complex
scenarios.
Various
feature
descriptors,
including
MSER
(Maximally
Stable
Extremal
Regions),
SURF
(Speeded-Up
Robust
Features),
distance
transform,
DOF
(Degrees
of
Freedom),
applied
skeleton
points,
while
BRIEF
(Binary
Independent
Elementary
HOG
(Histogram
Oriented
Gradients),
FAST
(Features
from
Accelerated
Segment
Test),
Optical
Flow
used
on
silhouettes
or
full-body
capture
both
geometric
motion-based
features.
Feature
fusion
was
employed
enhance
the
discriminative
power
extracted
data
physical
parameters
calculated
by
different
extraction
techniques.
The
system
utilized
a
hybrid
CNN
(Convolutional
Neural
Network)
+
RNN
(Recurrent
classifier
recognition,
with
Grey
Wolf
Optimization
(GWO)
selection.
Experimental
results
showed
significant
accuracy,
achieving
98.5%
UCF-101
dataset
99.2%
YouTube
dataset.
Compared
state-of-the-art
methods,
our
achieved
better
performance
recognition.
Electronics,
Год журнала:
2023,
Номер
12(8), С. 1892 - 1892
Опубликована: Апрель 17, 2023
Cyber-physical
security
is
vital
for
protecting
key
computing
infrastructure
against
cyber
attacks.
Individuals,
corporations,
and
society
can
all
suffer
considerable
digital
asset
losses
due
to
attacks,
including
data
loss,
theft,
financial
reputation
harm,
company
interruption,
damage,
ransomware
espionage.
A
cyber-physical
attack
harms
both
physical
assets.
system
more
challenging
than
software-level
because
it
requires
inspection
monitoring.
This
paper
proposes
an
innovative
effective
algorithm
strengthen
(CPS)
with
minimal
human
intervention.
It
approach
based
on
activity
recognition
(HAR),
where
GoogleNet–BiLSTM
network
hybridization
has
been
used
recognize
suspicious
activities
in
the
perimeter.
The
proposed
HAR-CPS
classifies
from
real-time
video
surveillance
average
accuracy
of
73.15%.
incorporates
machine
vision
at
IoT
edge
(Mez)
technology
make
latency
tolerant.
Dual-layer
ensured
by
operating
hybrid
a
cloud
server,
which
ensures
system.
optimization
scheme
makes
possible
only
USD
4.29±0.29
per
month.