Update app.py

app.py

@@ -130,13 +130,15 @@ class TransferVAE_Video(nn.Module):
         backward = lstm_out[:, 0, self.hidden_dim:2 * self.hidden_dim]
         frontal = lstm_out[:, self.frames - 1, 0:self.hidden_dim]
         lstm_out_f = torch.cat((frontal, backward), dim=1)
+        
         f_mean = self.f_mean(lstm_out_f)
-        f_logvar = self.f_logvar(lstm_out_f)
-        f_post = self.reparameterize(f_mean, f_logvar, random_sampling=False)
+        f_post = f_mean
+        
         features, _ = self.z_rnn(lstm_out)
+        
         z_mean = self.z_mean(features)
-        z_logvar = self.z_logvar(features)
-        z_post = self.reparameterize(z_mean, z_logvar, random_sampling=False)
+        z_post = z_mean
+
         return f_post, z_post
 
     
@@ -150,16 +152,6 @@ class TransferVAE_Video(nn.Module):
         x = x.view(-1, x_shape[-3], x_shape[-2], x_shape[-1])
         x_embed = self.encoder(x)[0]
         return x_embed.view(x_shape[0], x_shape[1], -1)
-    
-
-    def reparameterize(self, mean, logvar, random_sampling=True):
-        if random_sampling is True:
-            eps = torch.randn_like(logvar)
-            std = torch.exp(0.5 * logvar)
-            z = mean + eps * std
-            return z
-        else:
-            return mean
 
 
     def forward(self, x, beta):
@@ -171,9 +163,7 @@ class TransferVAE_Video(nn.Module):
         zf = torch.cat((z_post, f_expand), dim=2)
         recon_x = self.decoder_frame(zf)
         return f_post, z_post, recon_x
-    
-
-    
+        
 
 def name2seq(file_name):
     images = []
@@ -235,7 +225,7 @@ def MyPlot(frame_id, src_orig, tar_orig, src_recon, tar_recon, src_Zt, tar_Zt, s
     axs[1, 3].imshow(src_Zf_tar_Zt)
     axs[1, 3].axis('off')
     
-    plt.subplots_adjust(hspace=0.01, wspace=0.0)
+    plt.subplots_adjust(hspace=0.0125, wspace=0.0)
     
     save_name = 'MyPlot_{}.png'.format(frame_id)