►
From YouTube: UniMatrix GB 17 Meeting 20210929 Meeting Recording
Description
Homepage: https://gitlab.com/unimatrix
A
C
A
B
C
B
D
B
Okay,
so
I
sent
out
for
the
agenda
reflecting
that
for
some
time
there
has
been
some
questions
around
the
the
scope
etc
of
your
metrics,
and
I
ask
you
all
to
to
send
some
feedback
on
what
we
intentions
are
and
the
only
reply
I
got
was
from
from
ipro
very
detailed
reply
for
my
pro
on
a
technical
level,
a
lot
of
it.
But
I
think
some
of
the
reply,
somehow
I
shall
say,
coincides
with
also
access
view,
and
that
is,
I
think
we
are
a
little
bit.
B
B
A
A
A
So
it's
use
cases
that
are
beyond
safety
and
security.
It's
it's
not
really
within
the
the
budget
of
security
manager,
maybe
also
atari.
I
think
it
addresses
some
of
your
concerns
last
month
during
the
call
there
what
we
also
see
from
the
cost
perspective
with
an
operational
efficiency.
There
is
a
higher
acceptance
to
a
higher
cost
for
the
camera
and,
as
we
discussed
potential
components
for
any
metrics
project
project
and
some
fear
that
those
could
drive
the
demands
of
the
hardware
up.
A
So
I
think
what
we're
trying
to
say
is
that
it
probably
would
be
a
good
exercise
to
to
discuss
really
what
market
we're
trying
to
address
and
who
will
be
the
the
the
largest
group
that
benefit
from
from
uni
metrics
initiative
apart
from
camera
manufacturers.
So
what
are
the
developers
who
who
are
the
developers
that
will
be
using
that
and
what
they
will
be
trying
to
do?
A
D
Yeah,
I
I
I
totally
agree
with
you
with
your
view.
So
that's
why,
from
our
part
to
for
this
meeting
and
we,
we
prepare
a
little
bit
some
some
simple
slide
to
to
write
a
little
bit
on
our
conserve
or
our
information.
We
want
to
share
because
we
we
have
the
same
same
same
thinking
as
well,
that
if
we
want
to
do
this
first
thing,
we
need
to
have
to
work
out
a
very
thoughtful
targeting
targeting
strategy.
D
Firstly,
so
without
sata
it's
extremely
difficult
to
do
something
because
we
we
all
of
all
of
us
we
we
are.
We
are
capacity
at
the
resource,
limited
and
market
limited
and
even
experience
limited.
So
it's
it's
very
dangerous
or
risky
to
to
target
very
how
to
say
expensive
group
what
I
mean,
but
at
the
first
step
we
need
to
be
to
be
clarified
on
the
target,
so
yeah
from
our
part
of
we.
We
have
something
to
share
yeah,
maybe
later.
B
E
To
video
surveillance,
in
fact,
if
you
remember
when
we
met
in
lund
almost
two
years
ago,
I
asked
very
from
the
very
beginning
to
remove
the
the
word
video,
because
I
see
application
that
is
not
only
bound
in
in
video
audio
can
be
something
that
they
can
have.
A
lot
of,
for
example,
analytics
just
to
make
an
example,
shot
detectors
or
something
like
this,
but.
E
C
B
E
B
F
B
But
while
you're
you're
doing
that,
I
open
up
the
mission
here
on
the
governance,
ford
charter
and
it
actually
says
to
create
an
open
source
reference
implementation
using
industry,
standard
apis
and
frameworks
to
allow
management
of
cloud
native
application
in
the
physical
security
industry.
Ecosystems.
B
B
D
D
We
think
we
need
to
discuss
more
closely
on
this
kind
of
point.
Of
course,
hikvision
is
one
of
the
initial
participants,
in
that
we
we
kept
a
very
close
communication
and
the
contact
with
the
with
everyone
in
gp4
that
we
in
in
technical
order,
so
the
the
strategy
is
that
we,
we
really
would
like
to
play
a
show
inside
of
this
organization
at
the
very
beginning,
of
course.
So
that's
why
that's
why
we
from
our
rng
team?
We
are
committed
to
follow
up
a
load
of
framework.
D
Even
at
the
beginning,
we
we
didn't.
We
didn't
agree
on
some
kind
of
technical
technical
attraction,
but
that
is
fun
so
anyway,
we
can
put
our
own
reference
framework
as
a
sub
library
and
a
larger
framework.
So
this
is
a
this
kind
of
commun
commitment.
We
can
show
that
we
want
to
do
with
within
this
this
unimatrix
operation
and
also
dora
would
work
out
the
api
definition
so
based
on
tp
request
so
and
from
our
side
also.
D
D
The
most,
I
think,
is
that
we
can
be
the
biggest
how
to
say
promoter
of
the
whole
concept,
so
once
it
goes
up
to
market,
as
we
can,
hyphena
can
use
our
global
communication
network
and
distribution
channels,
so
we
can
send
out
this
message
to
our
massively
covered
customer
base
and
recruit
partners
or
educate
partners
through
different
approaches
through
webinar
training
or
everything
we
have
so,
but
anyway,
every
every
company,
including
axis
panasonic
or
any
other
company.
When
we
do
investment
in
investment,
we
need
to
think
about
the
return
right.
D
So
the
roi
is
the
most
important.
So
that's
why
we
are
very
urged
to
push
all
of
our
colleagues
inside
of
this
organization
to
have
a
clear-cut
business
model
or
the
very
clarified
direction.
So
this
is
the
most
important
without
that
that
will
become
a
very
hard
to
say,
challenging
for
us
to
to
asking
company
continued
investment
or
increase
their
investment
in
future,
because,
yes,
everyone
needs
to
see
a
good
perspective,
of
course.
So
without
that
it's
a
very
dangerous
for
everyone
right
yeah.
So.
D
D
Do
we
still
stick
on
this
initial
concept
or
not
because
from
this
concept
I
think
it's
a
very
eye-catching
or
very
good
position
concept,
but
until
now
from
our
part
of
visiting
it,
it
lacks
some
some
keyword,
key
environment,
for
example,
management
of
cloud
native
applications.
So
what
is
that?
How
to
do
that?
So
I
don't
think
it's
very
clear
right
and.
D
D
D
I
mentioned
it's:
it's
a
developer
base
that
we
talk
it
or
we
want
to
target
more,
not
just
with
with
some
different
vertical
market,
with
a
different
technology
like
business
intelligence.
So
we've
seen
lots
of
business
intentions
coming
out
from
these
two
years
and
also
in
industrial
manufacturing
area.
There
are
a
lot
of
players
doing
some
very
specific
automation
algorithm.
So
it's
also
another
trend
so,
but
all
of
the
both
of
them
are
very
quite
new
to
security
manufacturing
to
us.
So
how
do
we
react
to
that
kind
of
interaction?
D
D
I
don't
think
any
developers
want
to
want
to
invest
six
months
to
develop
an
app
without
any
benefits
right
also,
also
from
our
side.
We
see
a
lot
of
the
ai
and
the
deep
learning
framework
with
hardware
some
organization
or
make
a
company
that
they
have
their
own,
offering
available
on
in
market,
so
how
to
differentiate
ourselves
to
these
mega
players
offerings.
So
this
is
a
very
interesting
question.
G
D
D
First,
if,
if
we
want
to
say
the
first
thing
is
mostly
like
commercial
cctv
players
with
some
specific
detection
or
some
some
mpr,
some
phase
algorithm
providers-
maybe
everyone
has
some
some
partner
in
this
area,
but
yeah
he's
more
clear,
but
if
we
we
want
to
do
that
most
of
the
peers.
They
are
very
happy
or
very
comfortable
to
do
one
or
two
age
computing
with
with
hardware
manufacturers
with
the
axis
a
cap
or
hours
or
panasonic
or
other
other
players,
age
computing
framework.
D
So
this
is
still
good
to
recruit
them
to
a
logic
family,
but
what's
the
most,
how
to
say
most
carrying
it's,
not
the
age
computing,
the
age
capacity
itself,
but
the
whole
the
whole
solution
level,
including
them
vms,
layout,
vms
support,
because
they
need
to
have
some
kind
of
application
to
display
their
specific
algorithm
right.
If
this
is
just
a
generic
generic
algorithm,
of
course,
they
cannot
compete
with
our
with
cctv
hardware
manufacturers.
D
But
but
right
now
I
don't.
I
don't
see
many
many
many
famous
partners
coming
into
our
unimatch.
So
this
is
one
one
one
thing
another
the
second
day
and
the
third
days,
it's
quite
a
similar,
but
a
different
vertical.
The
second
one
is
some.
There
are
lots,
a
lot
of
startup
or
niche
competency,
focusing
on
detail
business
intelligence,
omni
channel
analytics
such
a
berries.
F
D
D
They
have
such
a
hands-on
capability
to
do
their
own
gpu
based
cameras
using
intel
or
a
media
gpu
so
and
the
and
the
they
they
also
they
will
have
their
own
application
levels
as
well,
but
the
most
challenging
for
them
is
that
they
need
to
keep
the
whole
hardware
production
in
the
supply
chain.
So
it's
a
little
bit
difficult
for
them,
but
the
quantity-
I
don't
know.
I
just
see
a
very
few
of
them
doing
such
things.
D
So
that's
why
that's
why
this
is
just
some
condensed
targeting
partners.
We
can
have
right
now,
so
maybe
there
are
more
from
different
areas.
We
don't
know
yet,
but
I
think
for
if
we
want
to,
we
want
to
make
uni
metrics
going
successfully
to
the
next
step.
We
need
to
do
the
targeting
as
a
first
as
the
first
tax,
because
we
need
to
scope
down.
D
D
We
don't
have
to
say,
jump
out
of
scope
to
think
about
how
to
compete
with
our
solution
to
other
to
other
competitors
or
providers,
for
example,
gpu
based
hardware
providers
or
some
other
providers,
and
if,
if
each
computing
framework
is
a
is
our
direction
but
and
as
a
target
the
partner
we
can
figure
out,
then
we
need
to
think
about
the
total
investment
cost.
It's
not
only
hardware
cause,
but
here
it's
a
it's
a
it's
a
it's.
D
A
general
total
cost
that
a
which
includes
the
development
time,
migration
codes
from
other
platform
to
our
platform
and
also
the
total
solution
codes,
including
equity
application
level.
So
they
need
to
vertical
vertical
algorithm
places.
They
need
to
think
about
how
to
how
to
show
their
application
in
cloud
or
in
our
without
cloud
etc,
and
also
we
need
to
think
about
how
to
go
to
market
how
to
market
the
such
such
a
development
work
for
our
app
developers.
D
If
we
want
to
be
very
successful
or
very
powerful
to
this
data,
I
just
I
just
have
some
some
brainstorming
ideas
that
if,
if
our
target
is
something
related
to
cloud,
almost
all
the
cloud
players
aws
or
microsoft,
azure
they
they
have
their
own
market,
their
own
revenue,
sharing
structure
and
their
own
their
own
strategy.
So
why
not?
We
can
use
their
their
structure
to
to
facilitate
some
up
developers,
some
to
to
help
the
app
to
earn
money
or
to
do
some
testing
to
do
some
religion,
marketing
things.
D
D
A
I
think
it's
very
interesting.
First
of
all,
thank
you
adler
for
for
for
sharing
that.
I'm
really
glad
to
see
that
a
lot
of
of
points
that
you've
highlighted
is
exactly
what
we
said
in
the
beginning
from
our
end,
so
I
think
it's
really
showing
that
there
is
a
need
to
revisiting
who
we're
really
a
target
and
also,
if
we
agree
on
the
targeting,
do
we
have
right
participants
in
the
project.
A
So
I
think
it's
my
personal
view
is
that
I
think
it's
great
that
we
kind
of
have
very
similar
view.
One
thing
when
you
talked
about
when
you
showed
that
the
wording
from
from
the
definition
of
unimetrics
you
know
the
colorful
and
catchy
phrases
I
mean
again.
This
is
a
collaboration
project,
so
if
we
believe
that
it
should
be
some
somewhat
different
or
need
to
be
defined,
it's
it's
up
to
this
group
to
define
it.
B
A
E
B
C
B
B
And
we
I
I
think,
also
we
see
a
lot
of
manufacturers,
maybe
depending
on
the
on
the
concern
about
ai
in
the
general
digitization
market,
that
manufacturers
tend
maybe
to
have
to
to
choose
to
have
that
in-house
and
because
they
see
more
value
in
in
providing
this
as
an
add-on
to
their
own
product
than
depending
on
the
next
on
a
third-party
network.
That's
my
conclusion.
A
I
agree
so
exactly
as
I
understand
I
mean
this
is
what
happened
like
almost
all
vendors,
providing
their
own
detection
analytics
that
are
widely
used
within
security,
so
solutions
from
application
developers
in
the
larger
extent
build
on
the
metadata
provided
by
those
applications
already,
so
it
would
be.
I
mean.
A
The
market
have
changed,
so
cameras
becomes
more
and
more
competent
and
a
lot
of
functionality
included
from
the
beginning.
So
I
think
we
see
also
in
in
our
development
community.
We
see
that
the
develop
developers
within
security
becomes
fewer
and
fewer
and
developers
within
field
outside
pure
security
becomes
more
and
more
so
they're
moving
towards
that,
because
because
actually
camera
manufacturers
are
taking
this
market
from
them,
yeah.
D
Everything
comes
to
during
the
last
recent
years,
everything
inside
the
city,
tv
market,
coming
to
from
hardware
level
coming
to
a
very
big,
generalized
generalization
in
from
a
level
of
specification,
because
every
manufacturer
have
a
high
megapixel,
4k
camera
or
low-light
cameras.
But
we
don't
have
a
lot
of
room
to
play
our
own
specificity
right.
So
only
thing
we
can
do
is
just
develop
our
own
algorithm
or
cooperate
with
other
algorithm
provider
to
do
some
kind
of
solution
solution
offering.
So
this
is
the
only
choice
for
every
camera
manufacturer.
D
How
to
see
a
server,
a
server
marketplace
for
their
general
marketplace
and
all
of
their
developer
can
use
some
kind
of
aws
or
a
microsoft,
deep
learning
capacity
to
do
some
app
or
then
testing
their
app
or
development
there
or
deployment
to
their
app
or
publish
their
app.
And
then
our
our
framework
can
have
some
connection
to
their
cloud
platform
and
download
some
some
app
directly
to
our
hardware.
D
And
for
this
point
I
think
it's
very
no
innovative
for
app
developers
because
they
are
very
comfortable
by
using
by
using
makeup
players
framework,
including
aws
and
microsoft,
and
also
their
marketplace
can
help
them
to
earn
more
money
or
more
marketing
exposure.
And
for
us
that
we
will
how
to
say,
we
will
make
a
coordination
together
with
them
and
get
more
cloud
based
up
into
age-based
app.
So
this
condition
synergy
that
maybe
this
is
some
point
we
can
think
about
that
and
yeah.
D
D
A
But
but,
as
you
said
at
the
moment,
it's
rather
difficult
for
the
developers
who
are
very
used
to
azure
environment
or
aws
environment
and
tool
chains
to
take
that
and
start
coding
in
embedded
and
rather
limited
environment
from
resource
point
of
view.
So
I
think
that's
that's.
This
is
where
we
could
collaborate
and
then
see
how
how
to
make
this
transition
for
developers
easier,
eventually
agree
on
on
the
tools
that
we
could
standardize
within
this
project,
in
collaboration
with
cloud
providers
if
needed.
D
Yeah,
it's
it's.
I
think
it's
it's!
It's
also
the
main
direction
of
larry,
the
writer,
because
from
a
lot
of
the
framework
in
the
u,
you
prefer
to
use
some
kind
of
publicly
known
the
reference
framework
and
with
the
opencv
or
something
like
that.
So
this
is
a
very
how
to
say
compatible
with
with
with
almost
all
the
mega
players.
So
yeah
technical-
I
I
don't
know
any
more
details,
but
the
technical
speaking.
A
How
do
me
do
we
move
from
here,
because,
first
of
all,
we
kind
of
agreed
that
that
the
scope
discussion
to
be
revisited
yeah
and
it's
likely
to
be
broader
than
just
security
focus.
Also.
I
think
we
agreed
that
participations
from
cloud
vendors
as
well
as
application
developers,
is
something
that
project
would
benefit
from.
D
C
G
A
A
C
So
in
conclusion,
so
many
kinds
of
vendors
will
be
aware:
join
unimatrix
project
and
according
to
other
song,
so
the
aws
or
microsoft-
I
don't
know,
but
so
the
marketplace
is
needed
for
the
application
distribution
is
my
understanding
correct
if
we
we
introduce
the
concept
after
now.
So
what
is
the
american
germanic?
So
we
should
make
clear
that
points,
and
so
so
we
should
this
decide
which
direction
we
should.
E
C
C
C
B
B
B
Everyone
in
favor
of
this
yes
very
good,
yeah,
very
good
so
and
then
we
can
move
on
to
the
other
parts.
Ipro
mykotosan
has
proposed
that
we
will
use
a
a
schematic
in
excel
a
task
list
that
we
in
the
gb
will
tell
the
the
the
communication
board
and
the
technical
board
to
map
down
their
tasks
and
and
map
them
on
a
timeline.
B
So
we
can
easily
know
where
we
are
on
the
in
on
the
undertaking.
So
it's
a
simple
question:
yes
or
no:
yes,
okay!
Yes,
for
that,
one,
and
also
the
next
question
is
we
are
about
to
register
the
project
with
the
linux
foundation
on
the
crowdfunding
site
is.
Is
that
yes,
that's
a
very
convenient
way
of
raising
some
money
or
receiving
donations?
If
we
want
to
spend
money
for
things?
B
Yes,
yes,
yes,
and
then
also
we
have
discussed
a
little
bit
of
keeping
mailman
as
the
reflector,
because
linux
foundation,
they
do
provide
a
reflector,
but
if
we
keep
it,
we
keep
also
the
the
capability
of
administrating
that
very
same
very
easy.
Instead
of
being,
depending
on
on
linux
foundation,
we
have
learned
that
it
takes
some
time
before
they
react
when
you
send
them
emails.