Update README.md

.gitignore

@@ -1,4 +1,2 @@
 .DS_Store
-images/
-.vscode/
 /images/

README.md

@@ -1 +1,21 @@
 # ML facial Generation
+
+ML algorithn to generate faces based on race and gender.
+
+# Installation
+This can be done using virtual environments. 
+The steps with different virtual environments are shown below
+
+## Virtual Environments
+
+### ****Conda***
+
+#### Create an environment
+
+```conda create --name nameofenvironment --file requirements.txt```
+
+#### Activating environment
+```conda activate nameofenvironment```
+
+# Run
+```python cli.py age race gender```

cli.py

@@ -18,17 +18,42 @@ def main(age,region,sex):
     z_dim = 100
     learning_rate = 0.0002
     beta1 = 0.5
-    epochs = 2
+    epochs = 20
     if region=='black':
-        pass
-        #genblack.train(epochs, batch_size, z_dim, learning_rate, beta1, get_batches, shape)
+        if sex == 'female':
+            data_files = genwhite.glob(os.path.join(data_dir, 'black/female/*.jpg'))
+            data_files.extend(genwhite.glob('*.png'))
+            shape = len(data_files), IMAGE_WIDTH, IMAGE_HEIGHT, 3
+            with tf.Session() as sess:
+                sess.run(tf.global_variables_initializer())
+            with tf.Graph().as_default():
+                genwhite.train(epochs, batch_size, z_dim, learning_rate, beta1,shape)
+        elif sex == 'male':
+            data_files = genwhite.glob(os.path.join(data_dir, 'black/male/*.jpg'))
+            data_files.extend(genwhite.glob('*.png'))
+            shape = len(data_files), IMAGE_WIDTH, IMAGE_HEIGHT, 3
+            with tf.Session() as sess:
+                sess.run(tf.global_variables_initializer())
+            with tf.Graph().as_default():
+                genwhite.train(epochs, batch_size, z_dim, learning_rate, beta1,shape)
+
     elif region=='white':
-        data_files = genwhite.glob(os.path.join(data_dir, 'white/*.jpg'))
-        shape = len(data_files), IMAGE_WIDTH, IMAGE_HEIGHT, 3
-        with tf.Session() as sess:
-            sess.run(tf.global_variables_initializer())
-        with tf.Graph().as_default():
-            genwhite.train(epochs, batch_size, z_dim, learning_rate, beta1,shape)
+        if sex == 'female':
+            data_files = genwhite.glob(os.path.join(data_dir, 'white/female/*.jpg'))
+            data_files.extend(genwhite.glob('*.png'))
+            shape = len(data_files), IMAGE_WIDTH, IMAGE_HEIGHT, 3
+            with tf.Session() as sess:
+                sess.run(tf.global_variables_initializer())
+            with tf.Graph().as_default():
+                genwhite.train(epochs, batch_size, z_dim, learning_rate, beta1,shape)
+        elif sex == 'male':
+            data_files = genwhite.glob(os.path.join(data_dir, 'white/male/*.jpg'))
+            data_files.extend(genwhite.glob('*.png'))
+            shape = len(data_files), IMAGE_WIDTH, IMAGE_HEIGHT, 3
+            with tf.Session() as sess:
+                sess.run(tf.global_variables_initializer())
+            with tf.Graph().as_default():
+                genwhite.train(epochs, batch_size, z_dim, learning_rate, beta1,shape)
     else:
         click.echo("Enter either black or white as region")
 

genwhite.py

@@ -12,9 +12,11 @@
 # Image configuration
 IMAGE_HEIGHT = 28
 IMAGE_WIDTH = 28
-data_files = glob(os.path.join(data_dir, 'white/*.jpg'))
+data_files = glob('images/**/*.*', recursive=True)
 shape = len(data_files), IMAGE_WIDTH, IMAGE_HEIGHT, 3
 
+# print(len(data_files))
+
 def get_image(image_path, width, height, mode):
     """
     Read image from image_path
@@ -223,7 +225,7 @@ def train(epoch_count, batch_size, z_dim, learning_rate, beta1, data_shape):
                 _ = sess.run(d_opt, feed_dict={input_real: batch_images, input_z: batch_z})
                 _ = sess.run(g_opt, feed_dict={input_real: batch_images, input_z: batch_z})
 
-                if steps % 400 == 0:
+                if steps % 20 == 0:
                     # At the end of every 10 epochs, get the losses and print them out
                     train_loss_d = d_loss.eval({input_z: batch_z, input_real: batch_images})
                     train_loss_g = g_loss.eval({input_z: batch_z})

requirements.txt

@@ -0,0 +1,8 @@
+python=3.6
+click
+requests
+tqdm
+tensorflow
+numpy
+matplotlib
+pillow