def forward(self, inputs):
"""
Forward process
Parameters
----------
inputs: Input tensor
Returns
-------
outputs: Predictions of network, shape is (batch, num_classes).
The outputs are called logits for classification net because the
softmax activations are not applied in this forward process
"""
use_bn = self.config.get('use_bn')
block_filters = self.config.get('block_filters')
block_conv_nums = self.config.get('block_conv_nums')
outputs = inputs
for i, filters in enumerate(block_filters):
outputs = self.build_blocks(outputs, filters,
block_conv_nums[i])
if i != len(block_filters) - 1:
outputs = maxpool(outputs, pool_size=2)
else:
outputs = global_average_pool(outputs)
if use_bn:
outputs = batch_normalization(outputs, training=self.training)
outputs = dense(outputs, self.config.get('num_classes'))
return outputs
def forward(self, inputs):
"""
Forward process
Parameters
----------
inputs: Input tensor
Returns
-------
outputs: Predictions of network, shape is (batch, num_classes).
The outputs are called logits for classification net because the
softmax activations are not applied in this forward process
"""
block_filters = self.config.get('block_filters')
block_dense_nums = self.config.get('block_dense_nums')
outputs = inputs
for i, filters in enumerate(block_filters):
first_blocks = i == 0
growth_rate = self.config.get('growth_rates')[i]
outputs = self.build_dense_blocks(outputs, filters,
block_dense_nums[i], growth_rate,
first_blocks)
if i != len(block_filters) - 1:
outputs = maxpool(outputs, pool_size=2)
else:
outputs = global_average_pool(outputs)
outputs = dense(outputs, self.config.get('num_classes'))
return outputs
def forward(self, inputs):
"""
Forward process
Parameters
----------
inputs: Input tensor
"""
# encoder process
outputs = inputs
encoders = []
for i, filters in enumerate(self.block_filters):
first_blocks = i == 0
outputs = self.build_res_blocks(outputs, filters,
self.block_res_nums[i],
first_blocks)
if i != len(self.block_filters) - 1:
encoders.append(outputs)
outputs = maxpool(outputs, pool_size=2)
# decoder process
encoder_num = len(encoders)
for k, encoder in enumerate(encoders[::-1]):
i = encoder_num - k - 1
first_blocks = i == 0
filters = int(encoder.get_shape()[-1])
outputs = conv_bn_relu(outputs,
filters,
3,
1,
1,
use_bn=self.use_bn,
training=self.training)
outputs = upsample(outputs, 2, self.interpolation_type)
outputs = concat([encoder, outputs])
outputs = self.build_res_blocks(outputs, filters,
self.block_res_nums[i],
first_blocks)
outputs = conv(outputs, self.config.get('output_channel'))
outputs = softmax(outputs)
return outputs