Effect of Hyperparameter Tuning on the Performance of YOLOv8 for Multi Crop Classification on UAV Images
Applied Sciences,
Год журнала:
2024,
Номер
14(13), С. 5708 - 5708
Опубликована: Июнь 29, 2024
This
study
investigates
the
performance
of
YOLOv8,
a
Convolutional
Neural
Network
(CNN)
architecture,
for
multi-crop
classification
in
mixed
farm
with
Unmanned
Aerial
Vehicle
(UAV)
imageries.
Emphasizing
hyperparameter
optimization,
specifically
batch
size,
study’s
primary
objective
is
to
refine
model’s
size
improved
accuracy
and
efficiency
crop
detection
classification.
Using
Google
Colaboratory
platform,
YOLOv8
model
was
trained
over
various
sizes
(10,
20,
30,
40,
50,
60,
70,
80,
90)
automatically
identify
five
different
classes
(sugarcane,
banana
trees,
spinach,
pepper,
weeds)
present
on
UAV
images.
The
assessed
using
accuracy,
precision,
recall
aim
identifying
optimal
size.
results
indicate
substantial
improvement
classifier
from
10
up
while
significant
dips
peaks
were
recorded
at
70
90.
Based
analysis
obtained
results,
Batch
60
emerged
best
overall
automatic
Although
F1
score
moderate,
combination
high
makes
it
most
balanced
option.
However,
Size
80
also
shows
very
precision
(98%)
(84%),
which
suitable
if
focus
achieving
precision.
findings
demonstrate
robustness
identification
highlighting
impact
tuning
appropriate
performance.
Язык: Английский
Optimized Membrane Fouling Prediction and Mitigation for Improved Water Treatment: a Review
International Journal of Chemical Engineering and Materials,
Год журнала:
2024,
Номер
3, С. 162 - 180
Опубликована: Дек. 31, 2024
This
review
article
presents
recent
advancements
in
membrane
filtration
technologies,
particularly
focusing
on
fouling
mechanisms
affecting
reverse
osmosis
(RO)
membranes.
It
a
comprehensive
analysis
of
various
studies
conducted
over
the
past
two
decades,
highlighting
complexities
caused
by
natural
organic
matter
(NOM),
particulate
matter,
and
biofouling.
The
also
examines
innovative
modelling
approaches
to
predict
behaviour,
including
development
Membrane
Fouling
Index-Ultrafiltration
(MFI-UF)
method
application
advanced
characterization
techniques
such
as
optical
coherence
tomography
(OCT)
Near-Edge
X-ray
Absorption
Fine
Structure
(NEXAFS)
spectroscopy.
Additionally,
it
discusses
effectiveness
pre-treatment
strategies,
coagulation
flocculation
mitigating
enhancing
performance.
Finally,
integration
artificial
intelligence
(AI)
predicting
behaviour
is
highlighted,
with
emphasis
its
potential
optimize
operational
parameters
systems.
Язык: Английский