fixed bugs in dataloaders and augmentations, adjusted models for new …

…default in_channels

models/FCN.py

@@ -45,7 +45,7 @@ def forward(self, x):
         return x
 
 class FullyConvNetwork(nn.Module):
-    def __init__(self, input_dims = 10, num_classes = 2):
+    def __init__(self, input_dims = 9, num_classes = 2):
 
         """
         DESC

models/resnet_tsc.py

@@ -63,7 +63,7 @@ class ResNetTSC(nn.Module):
     ResNetTSC Model Instance
     """
 
-    def __init__(self, num_classes=2, in_channels=10):
+    def __init__(self, num_classes=2, in_channels=9):
         super(ResNetTSC, self).__init__()
 
         # Configurations of Stages 

utils/DataLoaders.py

@@ -66,7 +66,8 @@ def __init__(self, root):
         self.transforms = torchvision.transforms.Compose([
             RotateColsRandomly(X_POS_COL, Y_POS_COL),
             RotateColsRandomly(X_VEL_COL, Y_VEL_COL),
-            RotateColsRandomly(factor = NOICE_FACTOR),
+            GaussianNoise(factor = NOICE_FACTOR),
+
         ])
 
     def __len__(self):
@@ -77,13 +78,13 @@ def __getitem__(self, idx):
         X = self.X[idx]
         Y = self.Y[idx]
 
-        X = torch.tensor(X)
+        X = torch.from_numpy(X)
         Y = torch.tensor(Y).long()
 
         X = self.transforms(X)
 
         # cast to tensor and return
-        return X, torch.tensor(Y).long()
+        return X, Y
 
     def collate_fn(batch):
 

utils/augmentations.py

@@ -56,7 +56,7 @@ def __init__(self, col_1, col_2):
     def __call__(self, x):
 
         # create an identity matrix of shape dim
-        dims = self.shape[2]
+        dims = x.shape[1]
         A = torch.eye(dims)
 
         # calculate random radians to rotate