def FCNN(): num_classes = 2 n_layers_enc = 32 n_layers_ctx = 128 n_input = 5 prob_drop = 0.25 layers = [] # Encoder pool = nn2.SpatialMaxPooling(2,2,2,2) layers.append(nn2.SpatialConvolution(n_input, n_layers_enc, 3, 3, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(nn2.SpatialConvolution(n_layers_enc, n_layers_enc, 3, 3, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(pool) # Context Module layers.append(nn2.SpatialDilatedConvolution(n_layers_enc, n_layers_ctx, 3, 3, 1, 1, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 2, 2, 2, 2)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 4, 4, 4, 4)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 8, 8, 8, 8)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 16, 16, 16, 16)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 32, 32, 32, 32)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 64, 64, 64, 64)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_enc, 1, 1)) layers.append(nn.ELU()) # Nao havia no paper # Decoder layers.append(nn2.SpatialMaxUnpooling(pool)) layers.append(nn2.SpatialConvolution(n_layers_enc, n_layers_enc, 3, 3, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(nn2.SpatialConvolution(n_layers_enc, num_classes, 3, 3, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(nn.SoftMax()) # Nao havia no paper return nn.Sequential(*layers)
def build_spatial_unpooling_net():
pool = nn.SpatialMaxPooling(2, 2, 2, 2)
unpool = nn.SpatialMaxUnpooling(pool)
return nn.Sequential().add(pool).add(unpool)
input_size=(3, 4, 5),
reference_fn=lambda i, _: i.select(1, 2)),
OldModuleTest(nn.SelectTable, (1, ),
input_size=[(1, ), (2, ), (3, ), (4, )],
reference_fn=lambda i, _: i[1]),
OldModuleTest(nn.SpatialAveragePooling, (2, 2), input_size=(2, 3, 6, 6)),
OldModuleTest(nn.SpatialAveragePooling, (2, 2, 2, 2),
input_size=(2, 3, 6, 6),
desc='stride'),
OldModuleTest(nn.SpatialAveragePooling, (2, 2, 2, 2, 1, 1),
input_size=(2, 3, 6, 6),
desc='stride_pad'),
OldModuleTest(
nn.SpatialAdaptiveMaxPooling, (4, 4),
input_size=(2, 3, 8, 8),
reference_fn=lambda i, _: nn.SpatialMaxPooling(2, 2).forward(i)),
OldModuleTest(nn.SpatialAdaptiveMaxPooling, (4, 4),
input_size=(2, 3, 7, 11),
desc='irregular'),
OldModuleTest(nn.SpatialConvolution, (3, 4, 3, 3),
input_size=(2, 3, 6, 6)),
OldModuleTest(nn.SpatialConvolution, (3, 4, 3, 3, 2, 2),
input_size=(2, 3, 6, 6),
desc='strided'),
OldModuleTest(nn.SpatialConvolution, (3, 4, 3, 3, 2, 2, 1, 1),
input_size=(2, 3, 6, 6),
desc='padding'),
OldModuleTest(nn.SpatialConvolutionLocal, (3, 2, 4, 4, 2, 2),
input_size=(1, 3, 4, 4)),
OldModuleTest(nn.SpatialConvolutionLocal, (3, 2, 6, 6, 2, 2, 2, 2),
input_size=(2, 3, 6, 6),
def test_max_pooling(self):
self._test_single_layer(nn.SpatialMaxPooling(3, 3, 1, 1, 1, 1))
