add docs for OneHot

Author: lorenzoh

docs/literate/preprocessing.md

@@ -5,4 +5,5 @@ This library also implements some general transformations useful for getting dat
 - [`ToEltype`](#)`(T)` converts the element type of any [`AbstractArrayItem`](#) to `T`.
 - [`ImageToTensor`](#) converts an image to an `ArrayItem` with another dimension for the color channels 
 - [`Normalize`](#) normalizes image tensors
+- [`OneHot`](#) to one-hot encode multi-class masks ([`MaskMulti`](#)s)
 

src/preprocessing.jl

@@ -186,6 +186,17 @@ tensortoimage(a::AbstractArray{T, 2}) where T = colorview(Gray, a)
 
 # OneHot encoding
 
+"""
+    OneHot([T = Float32])
+
+One-hot encodes a `MaskMulti` with `n` classes and size `sz` into
+an array item of size `(sz..., n)` with element type `T`. Supports [`apply!`].
+
+```julia
+item = MaskMulti(rand(1:4, 100, 100), 1:4)
+apply(OneHot(), item)
+```
+"""
 struct OneHot{T} <: Transform end
 OneHot() = OneHot{Float32}()