def create_heatmap_representations(self, source_image, kp_driving_value, kp_source_value):
"""
Eq 6. in the paper H_k(z)
"""
spatial_size = source_image.shape[2:]
gaussian_driving = kp2gaussian(kp_driving_value, spatial_size=spatial_size, kp_variance=self.kp_variance)
gaussian_source = kp2gaussian(kp_source_value, spatial_size=spatial_size, kp_variance=self.kp_variance)
heatmap = gaussian_driving - gaussian_source
#adding background feature
zeros = torch.zeros(heatmap.shape[0], 1, spatial_size[0], spatial_size[1]).type(heatmap.type())
heatmap = torch.cat([zeros, heatmap], dim=1)
heatmap = heatmap.unsqueeze(2)
return heatmap
def create_heatmap_representations(self, source_image, kp_driving,
kp_source):
"""
Eq 6. in the paper H_k(z)
"""
spatial_size = source_image.shape[2:]
# TODO：空余时间理解下这部分的 kp 的高斯变换（猜测是把 key point 的权重用一个高斯分布表示）
gaussian_driving = kp2gaussian(kp_driving,
spatial_size=spatial_size,
kp_variance=self.kp_variance)
gaussian_source = kp2gaussian(kp_source,
spatial_size=spatial_size,
kp_variance=self.kp_variance)
heatmap = gaussian_driving - gaussian_source
#adding background feature
zeros = torch.zeros(heatmap.shape[0], 1, spatial_size[0],
spatial_size[1]).type(heatmap.type())
heatmap = torch.cat([zeros, heatmap], dim=1)
heatmap = heatmap.unsqueeze(2)
return heatmap
def create_heatmap_representations(self, source_image, kp_driving,
kp_source):
"""
Eq 6. in the paper H_k(z)
"""
spatial_size = source_image.shape[2:]
gaussian_driving = kp2gaussian(kp_driving,
spatial_size=spatial_size,
kp_variance=self.kp_variance)
gaussian_source = kp2gaussian(kp_source,
spatial_size=spatial_size,
kp_variance=self.kp_variance)
heatmap = gaussian_driving - gaussian_source
# adding background feature
zeros = paddle.to_tensor(
np.zeros((heatmap.shape[0], 1, spatial_size[0], spatial_size[1]),
dtype=np.float32))
heatmap = paddle.concat([zeros, heatmap], axis=1)
heatmap = heatmap.unsqueeze(2)
return heatmap
def forward(self, x, kp=None):
feature_maps = []
out = x
if self.use_kp:
heatmap = kp2gaussian(kp, x.shape[2:], self.kp_variance)
out = torch.cat([out, heatmap], dim=1)
for down_block in self.down_blocks:
feature_maps.append(down_block(out))
out = feature_maps[-1]
prediction_map = self.conv(out)
return feature_maps, prediction_map
def forward(self, x, kp=None):
feature_maps = []
out = x
if self.use_kp:
heatmap = kp2gaussian(kp, x.shape[2:], self.kp_variance)
out = fluid.layers.concat([out, heatmap], axis=1)
for down_block in self.down_blocks:
feature_maps.append(down_block(out))
out = feature_maps[-1]
if self.sn is not None:
self.conv.weight.set_value(self.sn(self.conv.parameters()[0]))
prediction_map = self.conv(out)
return feature_maps, prediction_map
def create_heatmap_representations(self,
source_image,
kp_driving,
kp_source,
oval=False):
"""
Eq 6. in the paper H_k(z)
"""
spatial_size = source_image.shape[2:]
if not oval:
gaussian_driving = kp2gaussian(kp_driving,
spatial_size=spatial_size,
kp_variance=self.kp_variance)
gaussian_source = kp2gaussian(kp_source,
spatial_size=spatial_size,
kp_variance=self.kp_variance)
else: #modified
gaussian_driving = kp2ESgaussian(kp_driving,
spatial_size=spatial_size,
kp_variance_1=self.kp_variance,
kp_variance_2=self.kp_variance *
5) # added
gaussian_source = kp2ESgaussian(kp_source,
spatial_size=spatial_size,
kp_variance_1=self.kp_variance,
kp_variance_2=self.kp_variance *
5) # added
heatmap = gaussian_driving - gaussian_source
#adding background feature
zeros = torch.zeros(heatmap.shape[0], 1, spatial_size[0],
spatial_size[1]).type(heatmap.type())
heatmap = torch.cat([zeros, heatmap], dim=1)
heatmap = heatmap.unsqueeze(2)
return heatmap