Cli separation and data parsing

features.py

@@ -8,9 +8,9 @@
 
 import os
 
-from keras import applications
-from keras.applications.resnet50 import preprocess_input
-from keras.preprocessing import image
+from tensorflow.keras import applications, models, Model
+from tensorflow.keras.applications.resnet50 import preprocess_input
+from tensorflow.keras.preprocessing import image
 import numpy as np
 from pandas import DataFrame as DF
 
@@ -47,7 +47,7 @@ def _extract(fp, model):
         # Load the image, setting the size to 224 x 224
         img = image.load_img(fp, target_size=(224, 224))
 
-        # Convert the image to a numpy array, resize it (1, 2, 244, 244), and preprocess it
+        # Convert the image to a numpy array, resize it (1, 224, 224, 3), and preprocess it
         img_data = image.img_to_array(img)
         img_data = np.expand_dims(img_data, axis=0)
         img_data = preprocess_input(img_data)
@@ -67,9 +67,27 @@ def extract_features(filepath, model='ResNet50', write_to=None):
     print('Extracting features')
 
     # Get the model
-    print('Acquiring model "{}"'.format(model), end='')
-    m = named_model(model)
-    print('\rAcquired model\t\t\t\t\t')
+    print('Acquiring model "{}"'.format(model))
+
+    if type(model) == str:
+
+        # From filepath
+        if os.path.exists(model):
+            print('Assuming model argument is a filepath')
+            m = models.load_model(model)
+
+        # From standard named models
+        else:
+            print('Assuming model argument is a named model')
+            m = named_model(model)
+
+    # Model already in memory
+    else:
+        print('Assuming model argument is a loaded model')
+        m = model
+
+    assert isinstance(m, Model), 'Model \'{}\' is not a tf.keras.Model'.format(model)
+    print('\rSuccessfully acquired model\t\t\t\t\t')
 
     # Get the image filepaths
     filepath = filepath.replace('\\', '/')

tsne_reducer.py

@@ -23,7 +23,7 @@ def tsne(features, dims=2, write_to=None, tsne_kwargs={}):
     # Don't consider the first unique ID column
     features_salient = features.copy().drop(columns=[id_col_name], axis=1)
 
-    reduced = pd.DataFrame(TSNE(**tsne_kwargs).fit_transform(features_salient))
+    reduced = pd.DataFrame(TSNE(**tsne_kwargs).fit_transform(features_salient), dtype=object)
     reduced.insert(0, id_col_name, features[[id_col_name]])
 
     print('Success')

umap_reducer.py

@@ -25,7 +25,7 @@ def umap(features, dims=2, write_to=None):
     # Don't consider the first unique ID column
     features_salient = features.copy().drop(columns=[id_col_name], axis=1)
 
-    reduced = pd.DataFrame(UMAP(spread=0.5).fit_transform(features_salient))
+    reduced = pd.DataFrame(UMAP(spread=0.5).fit_transform(features_salient), dtype=object)
     reduced.insert(0, id_col_name, features[[id_col_name]])
 
     print('Success')