FastVision
"""
Mask{N, T}(classes) <: Block
Block for an N-dimensional categorical mask. `obs` is valid for
`Mask{N, T}(classes)`
if it is an N-dimensional array with every element in `classes`.
"""
struct
Mask
{
N
,
T
}
<:
Block
classes
::
AbstractVector
{
T
}
end
Mask
{
N
}
(
classes
::
AbstractVector
{
T
}
)
where
{
N
,
T
}
=
Mask
{
N
,
T
}
(
classes
)
function
checkblock
(
block
::
Mask
{
N
,
T
}
,
a
::
AbstractArray
{
T
,
N
}
)
where
{
N
,
T
}
return
all
(
map
(
x
->
x
∈
block
.
classes
,
a
)
)
end
function
mockblock
(
mask
::
Mask
{
N
,
T
}
)
where
{
N
,
T
}
rand
(
mask
.
classes
,
ntuple
(
_
->
16
,
N
)
)
::
AbstractArray
{
T
,
N
}
endVisualization
function
showblock!
(
io
,
::
ShowText
,
block
::
Mask
{
2
}
,
obs
)
img
=
_maskimage
(
obs
,
block
.
classes
)
ImageInTerminal
.
imshow
(
io
,
img
)
end
function
_maskimage
(
mask
,
classes
)
classtoidx
=
Dict
(
class
=>
i
for
(
i
,
class
)
in
enumerate
(
classes
)
)
colors
=
distinguishable_colors
(
length
(
classes
)
,
transform
=
deuteranopic
)
return
map
(
x
->
colors
[
classtoidx
[
x
]
]
,
mask
)
end
function
_maskimage
(
mask
::
AbstractArray
{
<:
Gray
{
T
}
}
,
args
...
)
where
{
T
}
_maskimage
(
reinterpret
(
T
,
mask
)
,
args
...
)
end
function
_maskimage
(
mask
::
AbstractArray
{
<:
Normed
{
T
}
}
,
args
...
)
where
{
T
}
_maskimage
(
reinterpret
(
T
,
mask
)
,
args
...
)
end
@
testset
"
OneHot [encoding]
"
begin
enc
=
OneHot
(
)
testencoding
(
enc
,
Mask
{
2
}
(
1
:
10
)
,
rand
(
1
:
10
,
50
,
50
)
)
end