bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2023,
Volume and Issue:
unknown
Published: Nov. 21, 2023
ABSTRACT
Digital
cameras
have
the
ability
to
capture
daily
images
of
plant
roots,
allowing
for
estimation
root
biomass.
However,
complexities
structures
and
noisy
image
backgrounds
pose
challenges
advanced
phenotyping.
Manual
segmentation
methods
are
laborious
prone
errors,
which
hinders
experiments
involving
several
plants.
This
paper
introduces
Rhizonet,
a
supervised
deep
learning
approach
semantic
images.
Rhizonet
harnesses
Residual
U-Net
backbone
enhance
prediction
accuracy,
incorporating
convex
hull
operation
precisely
outline
largest
connected
component.
The
primary
objective
is
accurately
segment
biomass
roots
analyze
their
growth
over
time.
input
data
comprises
color
various
samples
within
hydroponic
environment
known
as
EcoFAB,
subject
specific
nutrition
treatments.
Validation
tests
demonstrate
robust
generalization
model
across
experiments.
research
pioneers
advances
in
phenotype
analysis
by
standardizing
processes
facilitating
thousands
while
reducing
subjectivity.
proposed
algorithms
contribute
significantly
precise
assessment
dynamics
under
diverse
conditions.
Applied Sciences,
Journal Year:
2024,
Volume and Issue:
14(5), P. 1884 - 1884
Published: Feb. 25, 2024
Accurate
identification
of
fruits
in
greenhouse
environments
is
an
essential
need
for
the
precise
functioning
agricultural
robots.
This
study
presents
a
solution
to
problem
distinguishing
cucumber
from
their
stems
and
leaves,
which
often
have
similar
colors
natural
environment.
The
proposed
algorithm
fruit
relies
on
color
segmentation
form
matching.
First,
we
get
boundary
details
acquired
image
sample.
edge
information
described
reconstructed
by
utilizing
shape
descriptor
known
as
Fourier
order
acquire
matching
template
image.
Subsequently,
generate
multi-scale
amalgamating
computational
real-world
data.
target
subjected
conditioning
enhance
segmenacktation
region
inside
HSV
space.
Then,
segmented
compared
based
its
shape.
method
decreases
presence
unwanted
image,
hence
improving
effectiveness
An
analysis
was
performed
set
200
photos
that
were
obtained
field.
findings
indicate
presented
this
surpasses
conventional
recognition
algorithms
terms
accuracy
efficiency,
with
rate
up
86%.
Moreover,
system
has
exceptional
proficiency
identifying
targets
within
greenhouses.
attribute
renders
it
great
resource
offering
technical
assistance
robots
operate
accuracy.
Journal of Vibroengineering,
Journal Year:
2024,
Volume and Issue:
26(5), P. 1150 - 1165
Published: May 27, 2024
The
selection
of
weight
matrices
Q
and
R
in
the
LQR
control
strategy
for
active
suspension
is
susceptible
to
subjective
interference.
To
address
this
issue,
a
modified
differential
evolutionary
algorithm
proposed
optimize
controller,
ensuring
that
weighting
coefficients
are
set
their
optimal
values.
exhibits
drawbacks
terms
its
slow
convergence
rate
significant
impact
parameter
settings
on
obtained
results.
An
adaptive
two
candidate
mutation
strategies
adaptively
adjusts
scaling
factor
crossover
so
as
better
improve
ability
jumping
out
local
optimum
global
search.
algorithm's
functionality
verified
by
constructing
1/4
model
Simulink
software
platform
implementing
evolution
program
written
C++
language
using
MATLAB.
iterates
through
inputs
obtain
fitness
value
three
comfort
indices.
By
comparing
results
with
those
from
passive
traditional
suspension,
optimizing
based
can
effectively
reduce
vehicle
vibration
amplitude
while
considering
overall
performance
enhancement,
thereby
significantly
improving
ride
handling
stability.
Biomimetics,
Journal Year:
2024,
Volume and Issue:
9(10), P. 604 - 604
Published: Oct. 8, 2024
The
field
of
evolutionary
multitasking
optimization
(EMTO)
has
been
a
highly
anticipated
research
topic
in
recent
years.
EMTO
aims
to
utilize
algorithms
concurrently
solve
complex
problems
involving
multiple
tasks.
Despite
considerable
advancements
this
field,
numerous
continue
use
single
search
operator
(ESO)
throughout
the
evolution
process.
This
strategy
struggles
completely
adapt
different
tasks,
consequently
hindering
algorithm's
performance.
To
overcome
challenge,
paper
proposes
via
an
adaptive
bi-operator
(BOMTEA).
BOMTEA
adopts
and
adaptively
controls
selection
probability
each
ESO
according
its
performance,
which
can
determine
most
suitable
for
various
In
experiment,
showed
outstanding
results
on
two
well-known
benchmark
tests,
CEC17
CEC22,
significantly
outperformed
other
comparative
algorithms.
Agriculture,
Journal Year:
2024,
Volume and Issue:
14(7), P. 1075 - 1075
Published: July 4, 2024
Cassava
crop
age
estimation
is
crucial
for
optimizing
irrigation,
fertilization,
and
pest
management,
which
are
key
components
of
precision
agriculture.
Accurate
knowledge
allows
effective
resource
application,
minimizing
environmental
impact
enhancing
yield
predictions.
The
Bare
Land
Referenced
Algorithm
from
Hyper-Temporal
Data
(BRAH)
used
bare
land
classification
cassava
estimation,
but
it
traditionally
requires
manual
NDVI
thresholding,
challenging
with
large
datasets.
To
address
this
limitation,
we
propose
automating
the
thresholding
process
using
Otsu’s
method
image
contrast
histogram
equalization.
This
study
applies
these
enhancements
to
BRAH
algorithm
in
Ratchaburi,
Thailand,
utilizing
a
dataset
604
Landsat
satellite
images
1987
2024.
Our
research
demonstrates
accuracy
practicality
algorithm,
providing
94%
detecting
validation
locations
an
average
deviation
8.78
days
between
acquisition
date
validated
date.
approach
facilitates
precise
agricultural
planning
promoting
sustainable
farming
practices
supporting
several
Sustainable
Development
Goals
(SDGs).
