separated train and test transformation

Tune.py

@@ -8,15 +8,15 @@ from IdentitySubset import IdentitySubset
 from architectures.Model import Model, Architecture
 
 # numbre of classes 
-CLASS_SIZE = 30
+CLASS_SIZE = 20
 # batch 
-BATCH_SIZE = 16
+BATCH_SIZE = 8
 
 # size of images per class trainset + testset
 # 30 works best, more than that and we dont have enough data
 SAMPLE_SIZE = 30
 
-# this is then full sample - test sample
+# this is then (full_sample - test_sample)
 TRAINING_SMPLE = 28
 
 # learning rate
@@ -26,14 +26,18 @@ EPOCHS = 20
 # depends on model architecture
 # ResNet, DenseNet = 224
 # Inception = 299
-RESOLUTION = 299
+RESOLUTION = 224
 
-# model architecture
-arch = Architecture.INCEPTION
+# model architecture options are 
+#  - RESNET18
+#  - RESNET50
+#  - DENSENET121
+#  - INCEPTION
+arch = Architecture.RESNET18
 
 # DATA PREPARATION
 # load data set and prepare
-dataset = get_set(res = RESOLUTION)
+dataset = get_set()
 # select identities for experiment
 selected_identities = select_ids(
     dataset = dataset,
@@ -56,10 +60,12 @@ id_map = {old_id: new_id for new_id, old_id in enumerate(selected_identities)}
 
 # we remap identities because crossEntropyLoss requires in indices 0 -> (n-1)
 # where n = class size.
+tr_transform = train_transform(res = RESOLUTION)
 train_data = IdentitySubset(
-    dataset, 
-    train_indices, 
-    id_map)
+    dataset=dataset, 
+    indices=train_indices, 
+    id_mapping=id_map,
+    transform=tr_transform)
 
 train_loader = DataLoader(
     train_data, 
@@ -94,11 +100,14 @@ torch.save(
 print('Model saved!')
 
 # EVALUATE
+
+te_transform = test_transform(RESOLUTION)
 # Testing
 test_data = IdentitySubset(
-    dataset, 
-    test_indices, 
-    id_map)
+    dataset = dataset, 
+    indices=test_indices, 
+    id_mapping=id_map,
+    transform=te_transform)
 
 test_loader = DataLoader(
     test_data,