11 files changed, 19 insertions, 10 deletions

diff --git a/__pycache__/hyperparameters.cpython-38.pyc b/__pycache__/hyperparameters.cpython-38.pyc
new file mode 100644
index 00000000..3162d698
--- /dev/null
+++ b/__pycache__/hyperparameters.cpython-38.pyc
diff --git a/__pycache__/losses.cpython-38.pyc b/__pycache__/losses.cpython-38.pyc
new file mode 100644
index 00000000..7c0a4b73
--- /dev/null
+++ b/__pycache__/losses.cpython-38.pyc
diff --git a/checkpoint-images/img-epoch0.jpg b/checkpoint-images/img-epoch0.jpg
new file mode 100644
index 00000000..4836b9e8
--- /dev/null
+++ b/checkpoint-images/img-epoch0.jpg
diff --git a/checkpoint-images/img-epoch100.jpg b/checkpoint-images/img-epoch100.jpg
new file mode 100644
index 00000000..dfcf28e2
--- /dev/null
+++ b/checkpoint-images/img-epoch100.jpg
diff --git a/checkpoint-images/img-epoch200.jpg b/checkpoint-images/img-epoch200.jpg
new file mode 100644
index 00000000..13fde882
--- /dev/null
+++ b/checkpoint-images/img-epoch200.jpg
diff --git a/checkpoint-images/img-epoch300.jpg b/checkpoint-images/img-epoch300.jpg
new file mode 100644
index 00000000..13fde882
--- /dev/null
+++ b/checkpoint-images/img-epoch300.jpg
diff --git a/checkpoint-images/img-epoch400.jpg b/checkpoint-images/img-epoch400.jpg
new file mode 100644
index 00000000..13fde882
--- /dev/null
+++ b/checkpoint-images/img-epoch400.jpg
diff --git a/hyperparameters.py b/hyperparameters.py
index 75528742..ac2beda8 100644
--- a/hyperparameters.py
+++ b/hyperparameters.py
@@ -9,7 +9,7 @@ Number of epochs. If you experiment with more complex networks you
 might need to increase this. Likewise if you add regularization that
 slows training.
 """
-num_epochs = 500
+num_epochs = 200
 
 """
 A critical parameter that can dramatically affect whether training
@@ -20,6 +20,8 @@ learning_rate = 1e2
 
 momentum = 0.01
 
-alpha = 1
+alpha = .05
+beta = 5
+# alpha = 1e-5
 
-beta = 100
+# beta = 1e-2
diff --git a/losses.py b/losses.py
index 99b1eded..6405df6a 100644
--- a/losses.py
+++ b/losses.py
@@ -11,16 +11,19 @@ class YourModel(tf.keras.Model):
     def __init__(self, content_image, style_image): #normalize these images to float values
         super(YourModel, self).__init__()
        
-        self.content_image = transform.resize(content_image, tf.shape(style_image), anti_aliasing=True)
+        self.content_image = transform.resize(content_image, tf.shape(style_image), anti_aliasing=True, preserve_range=True, clip=False)
         self.content_image = tf.expand_dims(self.content_image, axis=0)
+       #  self.content_image = self.content_image.astype('uint8')
         print(self.content_image)
         
         #perhaps consider cropping to avoid distortion
-        self.style_image = transform.resize(style_image, tf.shape(style_image), anti_aliasing=True)
+        self.style_image = transform.resize(style_image, tf.shape(style_image), anti_aliasing=True, preserve_range=True, clip=False)
         self.style_image = tf.expand_dims(self.style_image, axis=0)
+       #  self.style_image = self.style_image.astype('uint8')
         #self.x = tf.Variable(initial_value = self.content_image.numpy().astype(np.float32), trainable=True) 
         self.x = tf.Variable(initial_value = np.random.rand(self.content_image.shape[0],
-        self.content_image.shape[1], self.content_image.shape[2], self.content_image.shape[3]).astype(np.float32), trainable=True) 
+        self.content_image.shape[1], self.content_image.shape[2], self.content_image.shape[3]).astype(np.float32), trainable=True)
+       #  self.x = self.x.astype('uint8')
 
         self.alpha = hp.alpha
         self.beta = hp.beta
@@ -121,7 +124,7 @@ class YourModel(tf.keras.Model):
         for i in range(len(photo_layers)):
               pl = photo_layers[i]
               il = input_layers[i]
-              L_content = tf.math.add(L_content, tf.reduce_mean(tf.square(pl - il)))
+              L_content = tf.math.add(L_content, tf.reduce_sum(tf.square(pl - il)))
        
         #print('content loss', L_content)
         return L_content
@@ -149,8 +152,10 @@ class YourModel(tf.keras.Model):
 
        # get the loss per each lateral layer
        # N depends on # of filters in the layer, M depends on hight and width of feature map
-       M_l = art_layer.shape[0]
-       N_l = art_layer.shape[1]
+       # M_l = art_layer.shape[0]
+       # N_l = art_layer.shape[1]
+       M_l = self.content_image.shape[0]*self.content_image.shape[1]
+       N_l = art_layer.shape[0]
        
        # layer.filters might not work
        E_l = 1/4 * (M_l**(-2)) *(N_l**(-2)) * tf.reduce_sum(tf.square(G_l - A_l))
diff --git a/main.py b/main.py
index 4497cce0..85ee4862 100644
--- a/main.py
+++ b/main.py
@@ -48,6 +48,7 @@ def train(model):
         model.train_step()
 
 def save_image(filename, image):
+    image = transform.resize(image, tf.shape(image), anti_aliasing=True).astype('uint8')
     imsave(filename, image)
 
 
@@ -69,8 +70,9 @@ def main():
     train(my_model)
     
     final_image = tf.squeeze(my_model.x)
+    final_image = transform.resize(final_image, tf.shape(final_image), anti_aliasing=True).astype('uint8')
 
-    plt.imshow(final_image).astype('uint8')
+    plt.imshow(final_image)
 
     imsave(ARGS.savefile, final_image)
 
diff --git a/save.jpg b/save.jpg
index e35f2cbf..cca810d7 100644
--- a/save.jpg
+++ b/save.jpg