#!/usr/bin/env python
# coding: utf-8

# # One Hundred Layers Tiramisu

# In[1]:


import torch
import torch.nn as nn
import torch.nn.functional as F
from collections import OrderedDict


# ### Initial Conv Block

# In[29]:


class _FirstConv(nn.Sequential):
    def __init__(self, num_input_features):
        super(_FirstConv, self).__init__()
        self.add_module('conv0', nn.Conv2d(3, num_input_features, kernel_size=7, stride=2, padding=3, bias=False))
        self.add_module('norm0', nn.BatchNorm2d(num_input_features))
        self.add_module('relu0', nn.ReLU(inplace=True))
        self.add_module('pool0', nn.MaxPool2d(kernel_size=3, stride=2, padding=1))


# In[30]:


# Test
conv1 = _FirstConv(5)


# ### Dense Layer

# In[31]:


class _DenseLayer(nn.Sequential):
    def __init__(self, num_input_features, growth_rate, bn_size, drop_rate):
        super(_DenseLayer, self).__init__()
        self.add_module('norm.1', nn.BatchNorm2d(num_input_features)),
        self.add_module('relu.1', nn.ReLU(inplace=True)),
        self.add_module('conv.1', nn.Conv2d(num_input_features, bn_size *
                        growth_rate, kernel_size=1, stride=1, bias=False)),
        self.add_module('norm.2', nn.BatchNorm2d(bn_size * growth_rate)),
        self.add_module('relu.2', nn.ReLU(inplace=True)),
        self.add_module('conv.2', nn.Conv2d(bn_size * growth_rate, growth_rate,
                                            kernel_size=3, stride=1, padding=1, bias=False)),
        self.drop_rate = drop_rate

    def forward(self, x):
        new_features = super(_DenseLayer, self).forward(x)
        if self.drop_rate > 0:
            new_features = F.dropout(new_features, p=self.drop_rate, training=self.training)
        return torch.cat([x, new_features], 1)


# In[32]:


# Test
dense1 = _DenseLayer(5,3,1,.5)


# ### Dense Block

# In[33]:


class _DenseBlock(nn.Sequential):
    def __init__(self, num_layers, num_input_features, bn_size, growth_rate, drop_rate):
        super(_DenseBlock, self).__init__()
        for i in range(num_layers):
            layer = _DenseLayer(num_input_features + i * growth_rate, growth_rate, bn_size, drop_rate)
            self.add_module('denselayer%d' % (i + 1), layer)


# In[34]:


# Test
denseBlock1 = _DenseBlock(4,4,1,4,.5)


# ### Transition Up

# In[35]:


class _TransitionUp(nn.Sequential):
    def __init__(self, num_input_features, num_output_features):
        super(_TransitionUp, self).__init__()
        self.add_module('norm', nn.BatchNorm2d(num_input_features))
        self.add_module('relu', nn.ReLU(inplace=True))
        self.add_module('conv', nn.Conv2d(num_input_features, num_output_features,
                                          kernel_size=1, stride=1, bias=False))
        self.add_module('pool', nn.AvgPool2d(kernel_size=2, stride=2))


# In[36]:


# Test
transUp = _TransitionUp(5,10)


# ### Transition Down

# In[37]:


class _TransitionDown(nn.Sequential):
    def __init__(self, num_input_features, num_output_features):
        super(_TransitionDown, self).__init__()
        self.add_module('norm', nn.BatchNorm2d(num_input_features))
        self.add_module('relu', nn.ReLU(inplace=True))
        self.add_module('conv', nn.Conv2d(num_input_features, num_output_features,
                                          kernel_size=1, stride=1, bias=False))
        self.add_module('pool', nn.AvgPool2d(kernel_size=2, stride=2))


# In[38]:


# Test
transDown = _TransitionDown(5,10)


# ### Final Model

# In[45]:


class FCDenseNet(nn.Module):
    r"""FC-DenseNet model class, based on
    `"The One Hundred Layers Tiramisu: Fully Convolutional DenseNets for Semantic Segmentation" <https://arxiv.org/pdf/1611.09326>`

    Args:
        growth_rate (int) - how many filters to add each layer (`k` in paper)
        block_config (list of 4 ints) - how many layers in each pooling block
        num_init_features (int) - the number of filters to learn in the first convolution layer
        bn_size (int) - multiplicative factor for number of bottle neck layers
          (i.e. bn_size * k features in the bottleneck layer)
        drop_rate (float) - dropout rate after each dense layer
        num_classes (int) - number of classification classes
    """
    def __init__(self, growth_rate=32, block_config=(6, 12, 24, 16),
                 num_init_features=64, bn_size=4, drop_rate=0, num_classes=1000):

        super(FCDenseNet, self).__init__()
        
        self.features = nn.Sequential()
        self.features.add_module('firstConv', _FirstConv(num_init_features))

        # Each denseblock
        num_features = num_init_features
        for i, num_layers in enumerate(block_config):
            block = _DenseBlock(num_layers=num_layers, num_input_features=num_features,
                                bn_size=bn_size, growth_rate=growth_rate, drop_rate=drop_rate)
            self.features.add_module('denseblock%d' % (i + 1), block)
            num_features = num_features + num_layers * growth_rate
            if i != len(block_config) - 1:
                trans = _Transition(num_input_features=num_features, num_output_features=num_features // 2)
                self.features.add_module('transition%d' % (i + 1), trans)
                num_features = num_features // 2

        # Final batch norm
        self.features.add_module('norm5', nn.BatchNorm2d(num_features))

        # Linear layer
        self.classifier = nn.Linear(num_features, num_classes)

    def forward(self, x):
        features = self.features(x)
        out = F.relu(features, inplace=True)
        out = F.avg_pool2d(out, kernel_size=7).view(features.size(0), -1)
        out = self.classifier(out)
        return out


# In[46]:


# Test
model = FCDenseNet()


# In[ ]: