Fix UNet for 3D convolutions (specify ndim to convxlayer and ResBlock) (

#263)

* Fix UNet for 3D convolutions (specify ndim to convxlayer and ResBlock)

FastVision/src/models/unet.jl

@@ -39,9 +39,9 @@ function UNetDynamic(backbone,
                       inputsize;
                       m_middle = UNetMiddleBlock,
                       skip_upscale = fdownscale,
-                      kwargs...)
+        kwargs...)
     outsz = Flux.outputsize(unet, inputsize)
-    return Chain(unet, final(outsz[end - 1], k_out))
+    return Chain(unet, final(outsz[end - 1], k_out, length(outsz) - 2))
 end
 
 function catchannels(x1, x2)
@@ -50,7 +50,7 @@ function catchannels(x1, x2)
 end
 
 function unetlayers(layers,
-                    sz;
+    sz;
                     k_out = nothing,
                     skip_upscale = 0,
                     m_middle = _ -> (identity,))
@@ -81,7 +81,8 @@ function unetlayers(layers,
         return UNetBlock(Chain(layer, childunet),
                          k_in,  # Input channels to upsampling layer
                          k_mid,
-                         k_out)
+                         k_out,
+                         length(outsz) - 2)
     end
 end
 
@@ -95,28 +96,28 @@ Given convolutional module `m` that halves the spatial dimensions
 and outputs `k_in` filters, create a module that upsamples the
 spatial dimensions and then aggregates features via  a skip connection.
 """
-function UNetBlock(m_child, k_in, k_mid, k_out = 2k_in)
+function UNetBlock(m_child, k_in, k_mid, k_out = 2k_in, ndim = 2)
     return Chain(upsample = SkipConnection(Chain(child = m_child,                              # Downsampling and processing
-                                                 upsample = PixelShuffleICNR(k_mid, k_mid)),
+                                                 upsample = PixelShuffleICNR(k_mid, k_mid, ndim)),
                                            Parallel(catchannels, identity, BatchNorm(k_in))),
                  act = xs -> relu.(xs),
-                 combine = UNetCombineLayer(k_in + k_mid, k_out))
+                 combine = UNetCombineLayer(k_in + k_mid, k_out, ndim))
 end
 
-function PixelShuffleICNR(k_in, k_out; r = 2)
-    return Chain(convxlayer(k_in, k_out * (r^2), ks = 1), Flux.PixelShuffle(r))
+function PixelShuffleICNR(k_in, k_out, ndim; r = 2)
+    return Chain(convxlayer(k_in, k_out * (r^ndim), ks = 1, ndim = ndim), Flux.PixelShuffle(r))
 end
 
-function UNetCombineLayer(k_in, k_out)
-    return Chain(convxlayer(k_in, k_out), convxlayer(k_out, k_out))
+function UNetCombineLayer(k_in, k_out, ndim)
+    return Chain(convxlayer(k_in, k_out, ndim = ndim), convxlayer(k_out, k_out, ndim = ndim))
 end
 
-function UNetMiddleBlock(k)
-    return Chain(convxlayer(k, 2k), convxlayer(2k, k))
+function UNetMiddleBlock(k, ndim)
+    return Chain(convxlayer(k, 2k, ndim = ndim), convxlayer(2k, k, ndim = ndim))
 end
 
-function UNetFinalBlock(k_in, k_out)
-    return Chain(ResBlock(1, k_in, k_in), convxlayer(k_in, k_out, ks = 1))
+function UNetFinalBlock(k_in, k_out, ndim)
+    return Chain(ResBlock(1, k_in, k_in, ndim = ndim), convxlayer(k_in, k_out, ks = 1, ndim = ndim))
 end
 
 """
@@ -139,4 +140,7 @@ end
 
     model = UNetDynamic(Models.xresnet18(), (128, 128, 3, 1), 4, fdownscale = 1)
     @test Flux.outputsize(model, (128, 128, 3, 1)) == (64, 64, 4, 1)
+
+    model = UNetDynamic(Models.xresnet18(ndim = 3), (128, 128, 128, 3, 1), 4)
+    @test Flux.outputsize(model, (128, 128, 128, 3, 1)) == (128, 128, 128, 4, 1)
 end end