International Journal of Comparative Psychology,
Journal Year:
2023,
Volume and Issue:
36(1)
Published: Sept. 25, 2023
Active,
real-time
observation
of
behavior
is
a
time-consuming
task,
which
heavily
resource-limited.
At
the
same
time,
simultaneous
several
individuals
often
paramount
to
increase
statistical
rigor
and
eliminate
potential
temporal
or
environmental
bias,
especially
in
natural
settings.
This
paper
describes
low-cost
video
recording
system
created
by
using
“off-the-shelf”
components.
The
easy
use
can
automatically
record
wide
variety
related
ecological
interactions
evolutionary
processes.
sensitive
enough
broad
range
animals
from
planarians,
small
insects
humans.
It
also
be
used
measure
plants.
will
work
during
daylight
hours
at
night
run
continuously
autonomously
for
48
hours,
longer
if
capture
motion-triggered
bigger
capacity
batteries
data
storage
facilities
are
used.
Animals,
Journal Year:
2023,
Volume and Issue:
13(20), P. 3276 - 3276
Published: Oct. 20, 2023
Animal
activity
recognition
(AAR)
using
wearable
sensor
data
has
gained
significant
attention
due
to
its
applications
in
monitoring
and
understanding
animal
behavior.
However,
two
major
challenges
hinder
the
development
of
robust
AAR
models:
domain
variability
difficulty
obtaining
labeled
datasets.
To
address
this
issue,
study
intensively
investigates
impact
unsupervised
adaptation
(UDA)
for
AAR.
We
compared
three
distinct
types
UDA
techniques:
minimizing
divergence-based,
adversarial-based,
reconstruction-based
approaches.
By
leveraging
UDA,
classifiers
enable
model
learn
domain-invariant
features,
allowing
trained
on
source
perform
well
target
without
labels.
evaluated
effectiveness
techniques
dog
movement
additional
from
horses.
The
application
across
positions
(neck
back),
sizes
(middle-sized
large-sized),
gender
(female
male)
within
data,
as
species
(dog
horses),
exhibits
improvements
classification
performance
reduced
discrepancy.
results
highlight
potential
mitigate
shift
enhance
various
settings
different
species,
providing
valuable
insights
practical
real-world
scenarios
where
is
scarce.
bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2024,
Volume and Issue:
unknown
Published: Oct. 14, 2024
Abstract
Cage-free
housing
systems
for
laying
hens,
and
their
accompanying
guidelines,
legislation,
audits,
are
becoming
more
common
around
the
world.
regulations
often
specify
requirements
floor
space
cage
height,
but
availability
of
three-dimensional
can
vary
depending
on
system
configurations.
Little
research
has
looked
at
how
much
vertical
a
hen
occupies
while
flapping
her
wings,
which
is
arguably
most
space-intensive
behavior.
Therefore,
objective
this
study
was
to
use
depth
sensing
camera
measure
maximum
height
hens
reach
when
wing
without
physical
obstructions.
Twenty-eight
individually
caged
Hy-line
W36
45
weeks
age
were
evaluated.
A
ceiling-mounted
centered
above
test
pen
calibrated
prior
collecting
data.
During
testing,
one
time
placed
in
recorded
wings.
From
footage,
minimum
distance
between
pixels
obtained
each
frame,
we
computed
reached
by
hen.
Results
used
during
event
showed
that
51.0
±
4.7
cm.
No
measures
correlated
with
from
(P>0.05).
Hens
single
strain,
old
enough
have
keel
damage,
cage-reared
housed,
preventing
us
generalizing
results
too
far.
However,
cameras
provide
useful
approach
varying
strains,
ages,
rearing/housing
methods
need
perform
dynamic
behaviors.
Sensors,
Journal Year:
2024,
Volume and Issue:
24(24), P. 7978 - 7978
Published: Dec. 13, 2024
Extracting
behavioral
information
from
animal
sounds
has
long
been
a
focus
of
research
in
bioacoustics,
as
sound-derived
data
are
crucial
for
understanding
behavior
and
environmental
interactions.
Traditional
methods,
which
involve
manual
review
extensive
recordings,
pose
significant
challenges.
This
study
proposes
an
automated
system
detecting
classifying
vocalizations,
enhancing
efficiency
analysis.
The
uses
preprocessing
step
to
segment
relevant
sound
regions
audio
followed
by
feature
extraction
using
Short-Time
Fourier
Transform
(STFT),
Mel-frequency
cepstral
coefficients
(MFCCs),
linear-frequency
(LFCCs).
These
features
input
into
convolutional
neural
network
(CNN)
classifiers
evaluate
performance.
Experimental
results
demonstrate
the
effectiveness
different
CNN
models
with
AlexNet,
DenseNet,
EfficientNet,
ResNet50,
ResNet152
being
evaluated.
achieves
high
accuracy
vocal
behaviors,
such
barking
howling
dogs,
providing
robust
tool
highlights
importance
systems
bioacoustics
suggests
future
improvements
deep
learning-based
methods
enhanced
classification
