International Journal of Applied Research in Bioinformatics,
Год журнала:
2024,
Номер
13(1), С. 1 - 22
Опубликована: Ноя. 30, 2024
Black
sigatoka
is
a
leaf
spot
disease
affecting
banana
plants
that
has
caused
significant
yield
reductions
of
up
to
50%
(Arman
et
al.,
2023).
This
research
presents
data
visualizations
8,761
points
related
black
in
plants,
encompassing
attributes
such
as
time,
canopy
temperature,
and
relative
humidity.
The
paper
also
reviews
work,
including
the
application
mining
plant
studies,
use
deep
learning
neural
networks
for
data,
machine
predicting
crop
yields
detecting
disease.
Additionally,
it
big
20,952
values
obtained
from
web-accessible
Pathogen–Host
Interactions
database
(PHI-base),
covering
various
categories
providing
an
epidemiological
analysis
prevalent
causative
agents
specific
diseases.
Artificial Intelligence Review,
Год журнала:
2025,
Номер
58(3)
Опубликована: Янв. 17, 2025
Abstract
Plant
diseases
cause
significant
damage
to
agriculture,
leading
substantial
yield
losses
and
posing
a
major
threat
food
security.
Detection,
identification,
quantification,
diagnosis
of
plant
are
crucial
parts
precision
agriculture
crop
protection.
Modernizing
improving
production
efficiency
significantly
affected
by
using
computer
vision
technology
for
disease
diagnosis.
This
is
notable
its
non-destructive
nature,
speed,
real-time
responsiveness,
precision.
Deep
learning
(DL),
recent
breakthrough
in
vision,
has
become
focal
point
agricultural
protection
that
can
minimize
the
biases
manually
selecting
spot
features.
study
reviews
techniques
tools
used
automatic
state-of-the-art
DL
models,
trends
DL-based
image
analysis.
The
techniques,
performance,
benefits,
drawbacks,
underlying
frameworks,
reference
datasets
more
than
278
research
articles
were
analyzed
subsequently
highlighted
accordance
with
architecture
deep
models.
Key
findings
include
effectiveness
imaging
sensors
like
RGB,
multispectral,
hyperspectral
cameras
early
detection.
Researchers
also
evaluated
various
architectures,
such
as
convolutional
neural
networks,
transformers,
generative
adversarial
language
foundation
Moreover,
connects
academic
practical
applications,
providing
guidance
on
suitability
these
models
environments.
comprehensive
review
offers
valuable
insights
into
current
state
future
directions
detection,
making
it
resource
researchers,
academicians,
practitioners
agriculture.
Frontiers in Plant Science,
Год журнала:
2025,
Номер
15
Опубликована: Фев. 11, 2025
In
natural
environments,
tomato
leaf
disease
detection
faces
many
challenges,
such
as
variations
in
light
conditions,
overlapping
symptoms,
tiny
size
of
lesion
areas,
and
occlusion
between
leaves.
Therefore,
an
improved
method,
DM-YOLO,
based
on
the
YOLOv9
algorithm,
is
proposed
this
paper.
Specifically,
firstly,
lightweight
dynamic
up-sampling
DySample
incorporated
into
feature
fusion
backbone
network
to
enhance
ability
extract
features
small
lesions
suppress
interference
from
background
environment;
secondly,
MPDIoU
loss
function
used
learning
details
margins
order
improve
accuracy
localizing
margins.
The
experimental
results
show
that
precision
(P)
model
increased
by
2.2%,
1.7%,
2.3%,
2%,
2.1%compared
with
those
multiple
mainstream
models,
respectively.
When
evaluated
dataset,
was
92.5%,
average
(AP)
mean
(mAP)
were
95.1%
86.4%,
respectively,
which
3%,
1.4%
higher
than
P,
AP,
mAP
YOLOv9,
baseline
model,
method
had
good
performance
potential,
will
provide
strong
support
for
development
smart
agriculture
control.
Agriculture,
Год журнала:
2025,
Номер
15(7), С. 733 - 733
Опубликована: Март 28, 2025
A
disease
detection
network
based
on
a
sparse
parallel
attention
mechanism
is
proposed
and
experimentally
validated
in
the
passion
fruit
(Passiflora
edulis
[Sims])
task.
Passiflora
edulis,
as
tropical
subtropical
tree,
loved
worldwide
for
its
unique
flavor
rich
nutritional
value.
The
experimental
results
demonstrate
that
model
performs
excellently
across
various
metrics,
achieving
precision
of
0.93,
recall
0.88,
an
accuracy
0.91,
mAP@50
(average
at
IoU
threshold
0.50)
0.90,
mAP@50–95
thresholds
from
0.50
to
0.95)
0.60,
F1-score
significantly
outperforming
traditional
object
models
such
Faster
R-CNN,
SSD,
YOLO.
experiments
show
offers
significant
advantages
with
multi-scale
complex
backgrounds.
This
study
proposes
lightweight
deep
learning
incorporating
(SPAM)
detection.
Built
upon
Convolutional
Neural
Network
(CNN)
backbone,
integrates
dynamically
selective
enhance
performance
cases
backgrounds
objects.
Experimental
has
superior
precision,
recall,
mean
average
(mAP)
compared
state-of-the-art
while
maintaining
computational
efficiency.
Frontiers in Plant Science,
Год журнала:
2025,
Номер
15
Опубликована: Янв. 9, 2025
Potatoes
and
tomatoes
are
important
Solanaceae
crops
that
require
effective
disease
monitoring
for
optimal
agricultural
production.
Traditional
methods
rely
on
manual
visual
inspection,
which
is
inefficient
prone
to
subjective
bias.
The
application
of
deep
learning
in
image
recognition
has
led
object
detection
models
such
as
YOLO
(You
Only
Look
Once),
have
shown
high
efficiency
identification.
However,
complex
climatic
conditions
real
environments
challenge
model
robustness,
current
mainstream
struggle
with
accurate
the
same
diseases
across
different
plant
species.
This
paper
proposes
SIS-YOLOv8
model,
enhances
adaptability
climates
by
improving
YOLOv8
network
structure.
research
introduces
three
key
modules:
1)
a
Fusion-Inception
Conv
module
improve
feature
extraction
against
backgrounds
like
rain
haze;
2)
C2f-SIS
incorporating
Style
Randomization
enhance
generalization
ability
crop
extract
more
detailed
features;
3)
an
SPPF-IS
boost
robustness
through
fusion.
To
reduce
model's
parameter
size,
this
study
employs
Dep
Graph
pruning
method,
significantly
decreasing
volume
19.9%
computational
load
while
maintaining
accuracy.
Experimental
results
show
outperforms
original
YOLOv8n
tasks
potatoes
tomatoes,
improvements
8.2%
accuracy,
4%
recall
rate,
5.9%
mAP50,
6.3%
mAP50-95.
Through
these
structure
optimizations,
demonstrates
enhanced
environments,
offering
solution
automatic
detection.
By
our
approach
not
only
advances
but
also
contributes
broader
adoption
AI-driven
solutions
sustainable
management
diverse
climates.
Applied Sciences,
Год журнала:
2025,
Номер
15(5), С. 2835 - 2835
Опубликована: Март 6, 2025
In
Japan,
local
governments
implore
residents
to
remove
the
batteries
from
small-sized
electronics
before
recycling
them,
but
some
products
still
contain
lithium-ion
batteries.
These
residual
may
cause
fires,
resulting
in
serious
injuries
or
property
damage.
Explosive
materials
such
as
mobile
(such
power
banks)
have
been
identified
fire
investigations.
Therefore,
these
fire-causing
items
should
be
detected
and
separated
regardless
of
whether
other
processes
are
use.
This
study
focuses
on
automatic
detection
using
deep
learning
electronic
products.
Mobile
were
chosen
first
target
this
approach.
study,
MATLAB
R2024b
was
applied
construct
You
Only
Look
Once
version
4
algorithm.
The
model
trained
enable
results
show
that
model’s
average
precision
value
reached
0.996.
Then,
expanded
three
categories
items,
including
batteries,
heated
tobacco
(electronic
cigarettes),
smartphones.
Furthermore,
real-time
object
videos
detector
carried
out.
able
detect
all
accurately.
conclusion,
technologies
significant
promise
a
method
for
safe
high-quality
recycling.
Theriogenology,
Год журнала:
2025,
Номер
245, С. 117504 - 117504
Опубликована: Май 29, 2025
The
morphological
characteristics
of
bull
spermatozoa
are
usually
evaluated
visually
using
bright-field
microscopy
according
to
the
guidelines
proposed
by
Society
for
Theriogenology
(SFT)
Bull
Breeding
Soundness
Evaluation
(BBSE).
However,
analysis
is
labor
consuming
and
requires
experienced
personnel
obtain
reliable
results.
Nevertheless,
artificial
insemination
industry
increasingly
demands
implementation
genomic
selection
schemes
young
bulls.
Hence,
there
a
growing
need
more
standardized
technique
analyze
semen
quality,
particularly
evaluation
abnormalities
that
affect
freezing
suitability
fertilizing
capacity.
Therefore,
an
Artificial
Intelligence
(AI)
algorithm
automated
classification
microscope-acquired
images
was
developed
neural
networks,
specifically
YOLO
based
on
convolutional
networks
(CNNs)
were
able
learn
extract
relevant
features
from
complex
visual
data
through
image
segmentation.
aim
assess
ability
identify
sperm
cells
in
images,
establish
their
viability
classify
morphology
simplified
scheme
which
included
only
normal
or
major/minor
defect
categories.
dataset
comprised
8243
labeled
annotated
with
bounding
boxes
allow
segmentation
learn.
performance
obtained
showed
accuracy
82
%,
although
it
not
observed
all
classes
(excluding
probable
case
overfitting
where
reached
100
%),
precision
85
%
correct
morphology.
Results
thereby
confirmed
potential
applicability
without
excluding
its
future
achieving
optimal
performance.