trainer and loader works :D

dataloader/dataloader.py

@@ -2,6 +2,8 @@
 import glob
 import skimage
 import torch
+import numpy as np
+
 
 class ImageLoader():
     def __init__(self, Images, Annotations,
@@ -36,8 +38,6 @@ def __getitem__(self, index):
         else:
             image = skimage.io.imread(self.images[index])
             label = skimage.io.imread(self.target_images[index])
-        image= torch.from_numpy(image)
-        label= torch.from_numpy(label)
         if self.transform:
             image = self.transform(image)
         if torch.cuda.is_available():

train.py

@@ -7,7 +7,10 @@
 from dataloader.dataloader import ImageLoader, TrainSet, TestSet
 from collections import defaultdict
 import torch
-params = hyperparameters(train_percentage=0.6, batch_size=10, epoch=4)
+from torchvision import transforms
+
+transforms_compose= transforms.Compose([transforms.ToTensor()])
+params = hyperparameters(train_percentage=0.6, batch_size=1, epoch=4)
 if torch.cuda.is_available():
     net= UNeT(n_class=1).cuda()
 else:
@@ -20,7 +23,7 @@
     Annotations=ANNOTATIONS_DIR,
     train_percentage=0.7)
 loss_val = Loss()
-Train = TrainSet(Images.train_set, extension="tif", transform=None)
+Train = TrainSet(Images.train_set, extension="tif", transform=transforms_compose)
 Test = TestSet(Images.test_set, extension="tif", transform=None)
 TrainLoder = DataLoader(
     Train,
@@ -37,7 +40,7 @@
     running_loss = 0.0
     for i, data in enumerate(TrainLoder, 0):
         inputs, labels= data["Image"], data["Label"]
-        inputs, labels= inputs.transpose(0, 3).transpose(1, 2), labels.transpose(0, 3).transpose(1, 2)
+        #inputs, labels= inputs.permute(0, 3, 1, 2), labels.permute(0, 3, 1, 2)
         optimizer.zero_grad()
         outputs = net(inputs)
         loss = loss_val.calc_loss(outputs, labels, metrics, bce_weight=0.5)