Frontiers in Marine Science,
Год журнала:
2024,
Номер
11
Опубликована: Май 28, 2024
Aquatic
biodiversity
monitoring
relies
on
species
recognition
from
images.
While
deep
learning
(DL)
streamlines
the
process,
performance
of
these
method
is
closely
linked
to
large-scale
labeled
datasets,
necessitating
manual
processing
with
expert
knowledge
and
consume
substantial
time,
labor,
financial
resources.
Semi-supervised
(SSL)
offers
a
promising
avenue
improve
DL
models
by
utilizing
extensive
unlabeled
samples.
However,
complex
collection
environments
long-tailed
class
imbalance
aquatic
make
SSL
difficult
implement
effectively.
To
address
challenges
in
within
scheme,
we
propose
Wavelet
Fusion
Network
Consistency
Equilibrium
Loss
function.
The
former
mitigates
influence
data
environment
fusing
image
information
at
different
frequencies
decomposed
through
wavelet
transform.
latter
improves
scheme
refining
consistency
loss
function
adaptively
adjusting
margin
for
each
class.
Extensive
experiments
are
conducted
FishNet
dataset.
As
expected,
our
existing
up
9.34%
overall
classification
accuracy.
With
accumulation
data,
improved
limited
shows
potential
advance
conservation.
BMC Research Notes,
Год журнала:
2024,
Номер
17(1)
Опубликована: Ноя. 27, 2024
These
data
enable
the
development
of
machine
learned
models
to
detect
unintended
passage
salmonids
over
in-stream
barriers.
Such
are
key
fully
characterizing
effectiveness
selective
systems,
as
they
and
quantify
fish
which
occurs
outside
intended
transit
selection
mechanism.
were
used
construct
custom
surveillance
tools
for
FishPass
(
https://www.glfc.org/fishpass.php
),
a
20-year
restoration
project
provide
up-
down-stream
desirable
fishes
while
simultaneously
blocking
or
removing
undesirable
fishes.
The
datasets
contain
2300
annotated
images
emerged
collected
in
natural
riverine
environment.
stem
from
video
during
2022
2023
fall
runs
several
pacific
salmonid
species
introduced
Laurentian
Great
Lakes
on
Boardman
(Ottaway)
River
Traverse
City,
MI,
USA.
In
addition
salmonids,
provided
containing
partially
submerged
fish,
other
wildlife
present
environmental
conditions
represented
by
most
clear
partly
cloudy.
could
be
develop
object
detection
environments.
International Journal of Advanced Technology and Engineering Exploration,
Год журнала:
2024,
Номер
11(111)
Опубликована: Фев. 29, 2024
The
ocean,
serving
as
a
vast
reservoir
of
resources
crucial
for
the
economy
and
human
sustenance,
plays
pivotal
role
in
influencing
economies
specific
countries.This
impact
is
particularly
evident
through
expansion
fisheries
sector
related
marine
industries
[1].To
strategically
develop
ensure
sustainable
growth
these
industries,
application
data
mining,
classification,
analyses
becomes
indispensable.Data
set
techniques
focused
on
extracting
pertinent
information
from
extensive
databases
across
diverse
business
domains,
stands
key
tool
informed
decision-making
[2].However,
existing
literature
this
field
faces
challenges
that
warrant
careful
consideration.
Journal of Marine Science and Engineering,
Год журнала:
2024,
Номер
12(3), С. 488 - 488
Опубликована: Март 14, 2024
Resource
management
for
fisheries
plays
a
pivotal
role
in
fostering
sustainable
industry.
In
Japan,
resource
surveys
rely
on
manual
measurements
by
staff,
incurring
high
costs
and
limitations
the
number
of
feasible
measurements.
This
study
endeavors
to
revolutionize
implementing
image-recognition
technology.
Our
methodology
involves
developing
system
that
detects
individual
fish
regions
images
automatically
identifies
crucial
keypoints
accurate
length
We
use
grounded-segment-anything
(Grounded-SAM),
foundation
model
instance
segmentation.
Additionally,
we
employ
Mask
Keypoint
R-CNN
trained
image
bank
(FIB),
which
is
an
original
dataset
images,
accurately
detect
significant
keypoints.
Diverse
were
gathered
evaluation
experiments,
demonstrating
robust
capabilities
proposed
method
detecting
both
Frontiers in Marine Science,
Год журнала:
2024,
Номер
11
Опубликована: Май 28, 2024
Aquatic
biodiversity
monitoring
relies
on
species
recognition
from
images.
While
deep
learning
(DL)
streamlines
the
process,
performance
of
these
method
is
closely
linked
to
large-scale
labeled
datasets,
necessitating
manual
processing
with
expert
knowledge
and
consume
substantial
time,
labor,
financial
resources.
Semi-supervised
(SSL)
offers
a
promising
avenue
improve
DL
models
by
utilizing
extensive
unlabeled
samples.
However,
complex
collection
environments
long-tailed
class
imbalance
aquatic
make
SSL
difficult
implement
effectively.
To
address
challenges
in
within
scheme,
we
propose
Wavelet
Fusion
Network
Consistency
Equilibrium
Loss
function.
The
former
mitigates
influence
data
environment
fusing
image
information
at
different
frequencies
decomposed
through
wavelet
transform.
latter
improves
scheme
refining
consistency
loss
function
adaptively
adjusting
margin
for
each
class.
Extensive
experiments
are
conducted
FishNet
dataset.
As
expected,
our
existing
up
9.34%
overall
classification
accuracy.
With
accumulation
data,
improved
limited
shows
potential
advance
conservation.
