steffen/hom_traj_gen
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

restructured

Browse Source
This commit is contained in:
Steffen Illium
2020-03-11 19:34:05 +01:00
parent 1f4edae95c
commit 7b795c2f7b
3 changed files with 49 additions and 69 deletions
Download Patch File Download Diff File Expand all files Collapse all files

112
lib/models/generators/cnn.py
View File

@ -1,5 +1,3 @@
from statistics import mean
from random import choice
import torch
@ -36,6 +34,7 @@ class CNNRouteGeneratorModel(LightningBaseModule):
kld_loss = -0.5 * torch.sum(1 + logvar - mu.pow(2) - logvar.exp())
# Dimensional Resizing TODO: Does This make sense? Sanity Check it!
# kld_loss /= reduce(mul, self.in_shape)
kld_loss *= self.hparams.data_param.dataset_length / self.hparams.train_param.batch_size
loss = (kld_loss + element_wise_loss) / 2
return dict(loss=loss, log=dict(element_wise_loss=element_wise_loss, loss=loss, kld_loss=kld_loss))
@ -83,7 +82,6 @@ class CNNRouteGeneratorModel(LightningBaseModule):
self.in_shape = self.dataset.map_shapes_max
# Todo: Better naming and size in Parameters
self.feature_dim = self.hparams.model_param.lat_dim * 10
self.feature_mixed_dim = self.feature_dim + self.feature_dim + 1
# NN Nodes
###################################################
@ -99,81 +97,64 @@ class CNNRouteGeneratorModel(LightningBaseModule):
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.map_res_1 = ResidualModule(self.map_conv_0.shape, ConvModule, 2, conv_kernel=3, conv_stride=1,
conv_padding=1, conv_filters=self.hparams.model_param.filters[0],
self.map_res_1 = ResidualModule(self.map_conv_0.shape, ConvModule, 2, conv_kernel=5, conv_stride=1,
conv_padding=2, conv_filters=self.hparams.model_param.filters[0],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.map_conv_1 = ConvModule(self.map_res_1.shape, conv_kernel=3, conv_stride=1, conv_padding=0,
self.map_conv_1 = ConvModule(self.map_res_1.shape, conv_kernel=5, conv_stride=1, conv_padding=0,
conv_filters=self.hparams.model_param.filters[1],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.map_res_2 = ResidualModule(self.map_conv_1.shape, ConvModule, 2, conv_kernel=3, conv_stride=1,
conv_padding=1, conv_filters=self.hparams.model_param.filters[1],
self.map_res_2 = ResidualModule(self.map_conv_1.shape, ConvModule, 2, conv_kernel=5, conv_stride=1,
conv_padding=2, conv_filters=self.hparams.model_param.filters[1],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.map_conv_2 = ConvModule(self.map_res_2.shape, conv_kernel=3, conv_stride=1, conv_padding=0,
self.map_conv_2 = ConvModule(self.map_res_2.shape, conv_kernel=7, conv_stride=1, conv_padding=0,
conv_filters=self.hparams.model_param.filters[2],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.map_res_3 = ResidualModule(self.map_conv_2.shape, ConvModule, 2, conv_kernel=3, conv_stride=1,
conv_padding=1, conv_filters=self.hparams.model_param.filters[2],
self.map_res_3 = ResidualModule(self.map_conv_2.shape, ConvModule, 2, conv_kernel=7, conv_stride=1,
conv_padding=3, conv_filters=self.hparams.model_param.filters[2],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.map_conv_3 = ConvModule(self.map_res_3.shape, conv_kernel=5, conv_stride=1, conv_padding=0,
conv_filters=self.hparams.model_param.filters[2]*2,
self.map_conv_3 = ConvModule(self.map_res_3.shape, conv_kernel=11, conv_stride=1, conv_padding=0,
conv_filters=self.hparams.model_param.filters[2],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.map_flat = Flatten(self.map_conv_3.shape)
self.map_lin = nn.Linear(reduce(mul, self.map_conv_3.shape), self.feature_dim)
#
# Trajectory Encoder
self.traj_conv_1 = ConvModule(self.in_shape, conv_kernel=3, conv_stride=1, conv_padding=0,
conv_filters=self.hparams.model_param.filters[0],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.traj_conv_2 = ConvModule(self.traj_conv_1.shape, conv_kernel=3, conv_stride=1, conv_padding=0,
conv_filters=self.hparams.model_param.filters[0],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.traj_conv_3 = ConvModule(self.traj_conv_2.shape, conv_kernel=3, conv_stride=1, conv_padding=0,
conv_filters=self.hparams.model_param.filters[0],
use_norm=self.hparams.model_param.use_norm,
use_bias=self.hparams.model_param.use_bias)
self.traj_flat = Flatten(self.traj_conv_3.shape)
self.traj_lin = nn.Linear(reduce(mul, self.traj_conv_3.shape), self.feature_dim)
#
# Mixed Encoder
self.mixed_lin = nn.Linear(self.feature_mixed_dim, self.feature_mixed_dim)
self.mixed_norm = nn.BatchNorm1d(self.feature_mixed_dim) if self.hparams.model_param.use_norm else lambda x: x
self.mixed_lin = nn.Linear(self.feature_dim, self.feature_dim)
self.mixed_norm = nn.BatchNorm1d(self.feature_dim) if self.hparams.model_param.use_norm else lambda x: x
#
# Variational Bottleneck
self.mu = nn.Linear(self.feature_mixed_dim, self.hparams.model_param.lat_dim)
self.logvar = nn.Linear(self.feature_mixed_dim, self.hparams.model_param.lat_dim)
self.mu = nn.Linear(self.feature_dim, self.hparams.model_param.lat_dim)
self.logvar = nn.Linear(self.feature_dim, self.hparams.model_param.lat_dim)
#
# Alternative Generator
self.alt_lin_1 = nn.Linear(self.hparams.model_param.lat_dim, self.feature_dim)
self.alt_lin_2 = nn.Linear(self.feature_dim, reduce(mul, self.traj_conv_3.shape))
# Todo Fix This Hack!!!!
reshape_shape = (1, self.map_conv_3.shape[1], self.map_conv_3.shape[2])
self.reshape_to_map = Flatten(reduce(mul, self.traj_conv_3.shape), self.traj_conv_3.shape)
self.alt_lin_2 = nn.Linear(self.feature_dim, reduce(mul, reshape_shape))
self.alt_deconv_1 = DeConvModule(self.traj_conv_3.shape, self.hparams.model_param.filters[2],
conv_padding=0, conv_kernel=5, conv_stride=1,
self.reshape_to_map = Flatten(reduce(mul, reshape_shape), reshape_shape)
self.alt_deconv_1 = DeConvModule(reshape_shape, self.hparams.model_param.filters[2],
conv_padding=0, conv_kernel=9, conv_stride=1,
use_norm=self.hparams.model_param.use_norm)
self.alt_deconv_2 = DeConvModule(self.alt_deconv_1.shape, self.hparams.model_param.filters[1],
conv_padding=0, conv_kernel=3, conv_stride=1,
conv_padding=0, conv_kernel=5, conv_stride=1,
use_norm=self.hparams.model_param.use_norm)
self.alt_deconv_3 = DeConvModule(self.alt_deconv_2.shape, self.hparams.model_param.filters[0],
conv_padding=1, conv_kernel=3, conv_stride=1,
conv_padding=1, conv_kernel=5, conv_stride=1,
use_norm=self.hparams.model_param.use_norm)
self.alt_deconv_out = DeConvModule(self.alt_deconv_3.shape, 1, activation=None,
conv_padding=1, conv_kernel=3, conv_stride=1,
@ -214,34 +195,33 @@ class CNNRouteGeneratorModel(LightningBaseModule):
return alt_tensor
def encode(self, map_array, trajectory, label):
map_tensor = self.map_conv_0(map_array)
map_tensor = self.map_res_1(map_tensor)
map_tensor = self.map_conv_1(map_tensor)
map_tensor = self.map_res_2(map_tensor)
map_tensor = self.map_conv_2(map_tensor)
map_tensor = self.map_res_3(map_tensor)
map_tensor = self.map_conv_3(map_tensor)
map_tensor = self.map_flat(map_tensor)
map_tensor = self.map_lin(map_tensor)
label_array = torch.cat([torch.full((1, 1, self.in_shape[1], self.in_shape[2]), x.item())
for x in label], dim=0)
label_array = self._move_to_model_device(label_array)
combined_tensor = torch.cat((map_array, trajectory, label_array), dim=1)
combined_tensor = self.map_conv_0(combined_tensor)
combined_tensor = self.map_res_1(combined_tensor)
combined_tensor = self.map_conv_1(combined_tensor)
combined_tensor = self.map_res_2(combined_tensor)
combined_tensor = self.map_conv_2(combined_tensor)
combined_tensor = self.map_res_3(combined_tensor)
combined_tensor = self.map_conv_3(combined_tensor)
traj_tensor = self.traj_conv_1(trajectory)
traj_tensor = self.traj_conv_2(traj_tensor)
traj_tensor = self.traj_conv_3(traj_tensor)
traj_tensor = self.traj_flat(traj_tensor)
traj_tensor = self.traj_lin(traj_tensor)
combined_tensor = self.map_flat(combined_tensor)
combined_tensor = self.map_lin(combined_tensor)
mixed_tensor = torch.cat((map_tensor, traj_tensor, label.float().unsqueeze(-1)), dim=1)
mixed_tensor = self.mixed_norm(mixed_tensor)
mixed_tensor = self.activation(mixed_tensor)
mixed_tensor = self.mixed_lin(mixed_tensor)
mixed_tensor = self.mixed_norm(mixed_tensor)
mixed_tensor = self.activation(mixed_tensor)
combined_tensor = self.mixed_lin(combined_tensor)
combined_tensor = self.mixed_norm(combined_tensor)
combined_tensor = self.activation(combined_tensor)
combined_tensor = self.mixed_lin(combined_tensor)
combined_tensor = self.mixed_norm(combined_tensor)
combined_tensor = self.activation(combined_tensor)
#
# Parameter and Sampling
mu = self.mu(mixed_tensor)
logvar = self.logvar(mixed_tensor)
# logvar = torch.clamp(logvar, min=0, max=10)
mu = self.mu(combined_tensor)
logvar = self.logvar(combined_tensor)
z = self.reparameterize(mu, logvar)
return z, mu, logvar