Fix backbone

Author: theabhirath

src/convnets/convmixer.jl

@@ -21,7 +21,7 @@ function convmixer(planes, depth; inchannels = 3, kernel_size = (9, 9),
                                                preact = true, groups = planes, pad = SamePad()), +),
                   conv_bn((1, 1), planes, planes, activation; preact = true)) for _ in 1:depth]
   head = Chain(AdaptiveMeanPool((1, 1)), MLUtils.flatten, Dense(planes, nclasses))
-  return Chain(stem, Chain(blocks), head)
+  return Chain(Chain(stem, Chain(blocks)), head)
 end
 
 convmixer_config = Dict(:base => Dict(:planes => 1536, :depth => 20, :kernel_size => (9, 9),

src/convnets/densenet.jl

@@ -28,8 +28,7 @@ Create a DenseNet transition sequence
 - `outplanes`: number of output feature maps
 """
 transition(inplanes, outplanes) =
-  Chain(conv_bn((1, 1), inplanes, outplanes; bias = false, rev = true),
-   MeanPool((2, 2)))
+  Chain(conv_bn((1, 1), inplanes, outplanes; bias = false, rev = true), MeanPool((2, 2)))
 
 """
     dense_block(inplanes, growth_rates)
@@ -68,8 +67,8 @@ function densenet(inplanes, growth_rates; reduction = 0.5, nclasses = 1000)
   for (i, rates) in enumerate(growth_rates)
     outplanes = inplanes + sum(rates)
     append!(layers, dense_block(inplanes, rates))
-    (i != length(growth_rates)) && 
-      append!(layers, transition(outplanes, floor(Int, outplanes * reduction)))
+    (i != length(growth_rates)) &&
+      push!(layers, transition(outplanes, floor(Int, outplanes * reduction)))
     inplanes = floor(Int, outplanes * reduction)
   end
   push!(layers, BatchNorm(outplanes, relu))

src/convnets/mobilenet.jl

@@ -31,12 +31,10 @@ function mobilenetv1(width_mult, config;
   for (dw, outch, stride, repeats) in config
     outch = Int(outch * width_mult)
     for _ in 1:repeats
-      layer = if dw
-        depthwise_sep_conv_bn((3, 3), inchannels, outch, activation; stride = stride, pad = 1)
-      else
-        conv_bn((3, 3), inchannels, outch, activation; stride = stride, pad = 1)
-      end
-      append!(layers, layer)
+      layer = dw ? depthwise_sep_conv_bn((3, 3), inchannels, outch, activation;
+                                         stride = stride, pad = 1) :
+                   conv_bn((3, 3), inchannels, outch, activation; stride = stride, pad = 1)
+      push!(layers, layer)
       inchannels = outch
     end
   end
@@ -120,7 +118,7 @@ function mobilenetv2(width_mult, configs; max_width = 1280, nclasses = 1000)
   # building first layer
   inplanes = _round_channels(32 * width_mult, width_mult == 0.1 ? 4 : 8)
   layers = []
-  append!(layers, conv_bn((3, 3), 3, inplanes, stride = 2))
+  push!(layers, conv_bn((3, 3), 3, inplanes, stride = 2))
 
   # building inverted residual blocks
   for (t, c, n, s, a) in configs
@@ -136,7 +134,7 @@ function mobilenetv2(width_mult, configs; max_width = 1280, nclasses = 1000)
   outplanes = (width_mult > 1) ? _round_channels(max_width * width_mult, width_mult == 0.1 ? 4 : 8) :
                                  max_width
 
-  return Chain(Chain(layers), conv_bn((1, 1), inplanes, outplanes, relu6, bias = false),
+  return Chain(Chain(Chain(layers), conv_bn((1, 1), inplanes, outplanes, relu6, bias = false)),
                Chain(AdaptiveMeanPool((1, 1)), MLUtils.flatten, Dense(outplanes, nclasses)))
 end
 
@@ -185,7 +183,7 @@ end
 (m::MobileNetv2)(x) = m.layers(x)
 
 backbone(m::MobileNetv2) = m.layers[1]
-classifier(m::MobileNetv2) = m.layers[2:end]
+classifier(m::MobileNetv2) = m.layers[2]
 
 # MobileNetv3
 
@@ -213,7 +211,7 @@ function mobilenetv3(width_mult, configs; max_width = 1024, nclasses = 1000)
   # building first layer
   inplanes = _round_channels(16 * width_mult, 8)
   layers = []
-  append!(layers, conv_bn((3, 3), 3, inplanes, hardswish; stride = 2))
+  push!(layers, conv_bn((3, 3), 3, inplanes, hardswish; stride = 2))
   explanes = 0
   # building inverted residual blocks
   for (k, t, c, r, a, s) in configs
@@ -232,7 +230,7 @@ function mobilenetv3(width_mult, configs; max_width = 1024, nclasses = 1000)
                      Dropout(0.2),
                      Dense(output_channel, nclasses))
 
-  return Chain(Chain(layers), conv_bn((1, 1), inplanes, explanes, hardswish, bias = false),
+  return Chain(Chain(Chain(layers), conv_bn((1, 1), inplanes, explanes, hardswish, bias = false)),
                Chain(AdaptiveMeanPool((1, 1)), MLUtils.flatten, classifier))
 end
 
@@ -308,4 +306,4 @@ end
 (m::MobileNetv3)(x) = m.layers(x)
 
 backbone(m::MobileNetv3) = m.layers[1]
-classifier(m::MobileNetv3) = m.layers[2:end]
+classifier(m::MobileNetv3) = m.layers[2]

src/convnets/resnet.jl

@@ -82,7 +82,7 @@ function resnet(block, residuals::AbstractVector{<:NTuple{2, Any}}, connection =
   inplanes = 64
   baseplanes = 64
   layers = []
-  append!(layers, conv_bn((7, 7), 3, inplanes; stride = 2, pad = 3, bias = false))
+  push!(layers, conv_bn((7, 7), 3, inplanes; stride = 2, pad = 3, bias = false))
   push!(layers, MaxPool((3, 3), stride = (2, 2), pad = (1, 1)))
   for (i, nrepeats) in enumerate(block_config)
     # output planes within a block

src/convnets/resnext.jl

@@ -40,7 +40,7 @@ function resnext(cardinality, width, widen_factor = 2, connection = (x, y) -> @.
   inplanes = 64
   baseplanes = 128
   layers = []
-  append!(layers, conv_bn((7, 7), 3, inplanes; stride = 2, pad = (3, 3)))
+  push!(layers, conv_bn((7, 7), 3, inplanes; stride = 2, pad = (3, 3)))
   push!(layers, MaxPool((3, 3), stride = (2, 2), pad = (1, 1)))
   for (i, nrepeats) in enumerate(block_config)
     # output planes within a block

src/convnets/vgg.jl

@@ -16,7 +16,7 @@ function vgg_block(ifilters, ofilters, depth, batchnorm)
   layers = []
   for _ in 1:depth
     if batchnorm
-      append!(layers, conv_bn(k, ifilters, ofilters; pad = p, bias = false))
+      push!(layers, conv_bn(k, ifilters, ofilters; pad = p, bias = false))
     else
       push!(layers, Conv(k, ifilters => ofilters, relu, pad = p))
     end
@@ -62,15 +62,12 @@ Create VGG classifier (fully connected) layers
 - `dropout`: the dropout level between each fully connected layer
 """
 function vgg_classifier_layers(imsize, nclasses, fcsize, dropout)
-  layers = []
-  push!(layers, MLUtils.flatten)
-  push!(layers, Dense(Int(prod(imsize)), fcsize, relu))
-  push!(layers, Dropout(dropout))
-  push!(layers, Dense(fcsize, fcsize, relu))
-  push!(layers, Dropout(dropout))
-  push!(layers, Dense(fcsize, nclasses))
-
-  return layers
+  return Chain(MLUtils.flatten,
+               Dense(Int(prod(imsize)), fcsize, relu),
+               Dropout(dropout),
+               Dense(fcsize, fcsize, relu),
+               Dropout(dropout),
+               Dense(fcsize, nclasses))
 end
 
 """
@@ -94,7 +91,7 @@ function vgg(imsize; config, inchannels, batchnorm = false, nclasses, fcsize, dr
   conv = vgg_convolutional_layers(config, batchnorm, inchannels)
   imsize = outputsize(conv, (imsize..., inchannels); padbatch = true)[1:3]
   class = vgg_classifier_layers(imsize, nclasses, fcsize, dropout)
-  return Chain(Chain(conv), Chain(class))
+  return Chain(Chain(conv), class)
 end
 
 const vgg_conv_config = Dict(:A => [(64,1), (128,1), (256,2), (512,2), (512,2)],