Marine Mammal Science,
Год журнала:
2022,
Номер
39(2), С. 422 - 437
Опубликована: Окт. 11, 2022
Abstract
The
SPLASH
project
(2004–2006)
revealed
complex
population
structure
and
migratory
connections,
but
no
regional
effort
was
made
along
the
southern
Mexican
Pacific
coast
until
dedicated
research
initiated
in
2010.
It
is
unclear
whether
humpback
whales
documented
this
region
belong
to
or
Central
American
units.
This
study
aimed
establish
relationship
between
of
Mexico
those
surrounding
Humpback
whale
photo‐identification
images
were
compared
from
multiple
locations
north,
central,
America.
Whales'
movements
among
regions
estimated
using
Interchange
Index
(
InI
)
Movement
MI
).
results
showed
higher
sampling
areas,
suggesting
that
“Central
unit.”
unit
migrates
north
feeding
areas
US
West
Coast
a
corridor
mainland
mouth
Gulf
California
Baja
Peninsula.
degree
interaction
with
other
units
unknown.
More
studies
are
needed
understand
disentangle
photographic
capture
on
migration
where
their
main
wintering
area.
Royal Society Open Science,
Год журнала:
2024,
Номер
11(2)
Опубликована: Фев. 1, 2024
For
the
40
years
after
end
of
commercial
whaling
in
1976,
humpback
whale
populations
North
Pacific
Ocean
exhibited
a
prolonged
period
recovery.
Using
mark–recapture
methods
on
largest
individual
photo-identification
dataset
ever
assembled
for
cetacean,
we
estimated
annual
ocean-basin-wide
abundance
species
from
2002
through
2021.
Trends
estimates
describe
strong
post-whaling
era
population
recovery
16
875
(±
5955)
to
peak
estimate
33
488
4455)
2012.
An
apparent
20%
decline
2012
2021,
26
662
4192),
suggests
abruptly
reached
carrying
capacity
due
loss
prey
resources.
This
was
particularly
evident
whales
wintering
Hawai‘i,
where,
by
had
declined
34%
2013,
down
levels
previously
seen
2006,
and
contrasted
an
absence
Mainland
Mexico
breeding
humpbacks.
The
strongest
marine
heatwave
recorded
globally
date
during
2014–2016
appeared
have
altered
course
recovery,
with
enduring
effects.
Extending
this
time
series
will
allow
serve
as
indicator
ecosystem
face
changing
climate.
Frontiers in Marine Science,
Год журнала:
2021,
Номер
8
Опубликована: Сен. 27, 2021
Motivated
by
the
need
to
estimate
abundance
of
marine
mammal
populations
inform
conservation
assessments,
especially
relating
fishery
bycatch,
this
paper
provides
background
on
estimation
and
reviews
various
methods
available
for
pinnipeds,
cetaceans
sirenians.
We
first
give
an
“entry-level”
introduction
estimation,
including
fundamental
concepts
importance
recognizing
sources
bias
obtaining
a
measure
precision.
Each
primary
mammals
is
then
described,
data
collection
analysis,
common
challenges
in
implementation,
assumptions
made,
violation
which
can
lead
bias.
The
main
method
estimating
pinniped
extrapolation
counts
animals
(pups
or
all-ages)
land
ice
whole
population.
Cetacean
sirenian
primarily
estimated
from
transect
surveys
conducted
ships,
small
boats
aircraft.
If
individuals
species
be
recognized
natural
markings,
mark-recapture
analysis
photo-identification
used
number
using
study
area.
Throughout,
we
cite
example
studies
that
illustrate
described.
To
population,
key
issues
include:
defining
population
estimated,
considering
candidate
based
strengths
weaknesses
relation
range
logistical
practical
issues,
being
aware
resources
required
collect
analyze
data,
understanding
made.
conclude
with
discussion
some
given
arise
during
implementation.
Mammalian Biology,
Год журнала:
2022,
Номер
102(3), С. 1005 - 1023
Опубликована: Апрель 5, 2022
Abstract
Determining
which
species
are
at
greatest
risk,
where
they
most
vulnerable,
and
what
the
trajectories
of
their
communities
populations
is
critical
for
conservation
management.
Globally
distributed,
wide-ranging
whales
dolphins
present
a
particular
challenge
in
data
collection
because
no
single
research
team
can
record
over
biologically
meaningful
areas.
Flukebook.org
an
open-source
web
platform
that
addresses
these
gaps
by
providing
researchers
with
latest
computational
tools.
It
integrates
photo-identification
algorithms
management,
sharing,
privacy
infrastructure
whale
dolphin
research,
enabling
global
collaborative
study
species.
With
seven
automatic
identification
trained
15
different
species,
resulting
37
species-specific
pipelines,
Flukebook
extensible
foundation
continually
incorporates
emerging
AI
techniques
applies
them
to
cetacean
photo
through
continued
collaboration
between
computer
vision
researchers,
software
engineers,
biologists.
2.0
million
photos
52,000
identified
individual
animals
submitted
250
enables
comprehensive
understanding
populations,
fostering
international
cross-institutional
while
respecting
ownership
privacy.
We
outline
technology
stack
architecture
Flukebook,
its
performance
on
real-world
imagery,
development
as
example
scalable,
extensible,
reusable
software.
step
change
our
ability
conduct
large-scale
cetaceans
across
geographic
ranges,
rapidly
iterate
population
assessments
abundance
trajectories,
engage
public
actions
protect
them.
Journal of Marine Science and Engineering,
Год журнала:
2023,
Номер
11(3), С. 595 - 595
Опубликована: Март 11, 2023
Monitoring
marine
mammals
is
of
broad
interest
to
governments
and
individuals
around
the
globe.
Very
high-resolution
(VHR)
satellites
hold
promise
reaching
remote
challenging
locations
fill
gaps
in
our
knowledge
mammal
distribution.
The
time
has
come
create
an
operational
platform
that
leverages
increased
resolution
satellite
imagery,
proof-of-concept
research,
advances
cloud
computing,
machine
learning
monitor
world’s
oceans.
Geospatial
Artificial
Intelligence
for
Animals
(GAIA)
initiative
was
formed
address
this
challenge
with
collaborative
innovation
from
government
agencies,
academia,
private
sector.
In
paper,
we
share
lessons
learned,
challenges
faced,
vision
how
VHR
imagery
can
enhance
understanding
cetacean
distribution
future.
Methods in Ecology and Evolution,
Год журнала:
2023,
Номер
14(10), С. 2611 - 2625
Опубликована: Июль 13, 2023
Abstract
Researchers
can
investigate
many
aspects
of
animal
ecology
through
noninvasive
photo–identification.
Photo–identification
is
becoming
more
efficient
as
matching
individuals
between
photos
increasingly
automated.
However,
the
convolutional
neural
network
models
that
have
facilitated
this
change
need
training
images
to
generalize
well.
As
a
result,
they
often
been
developed
for
individual
species
meet
threshold.
These
single‐species
methods
might
underperform,
ignore
potential
similarities
in
identifying
characteristics
and
photo–identification
process
among
species.
In
paper,
we
introduce
multi‐species
model
based
on
state‐of‐the‐art
method
human
facial
recognition,
ArcFace
classification
head.
Our
uses
two
such
heads
jointly
classify
identities,
allowing
share
information
parameters
within
network.
demonstration,
trained
with
50,796
from
39
catalogues
24
cetacean
species,
evaluating
its
predictive
performance
21,192
test
same
catalogues.
We
further
evaluated
external
entirely
composed
identities
did
not
see
during
training.
The
achieved
mean
average
precision
(MAP)
0.869
set.
Of
these,
10
representing
seven
MAP
score
over
0.95.
For
some
there
was
notable
variation
catalogues,
largely
explained
by
photo
quality.
Finally,
appeared
well,
scoring
similarly
their
species'
counterparts
larger
From
our
application,
provide
list
recommendations
users
model,
focusing
those
example,
high
quality
animals
identified
dorsal
nicks
notches
should
expect
near
optimal
performance.
Users
decreasing
higher
proportions
indistinct
or
poor
photos.
note
currently
freely
available
code
GitHub
repository
graphical
user
interface,
additional
functionality
collaborative
data
management,
via
Happywhale.com.